DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
This office action is in responsive to RCE filed on 03/17/2026. Claims remain pending in the application. Claims 1, 12, and 16 are independent.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-11 are rejected under 35 U.S.C. 103 as being unpatentable over ALLEGORICO et al. (US 2017/0320004 A1, pub. date: 11/09/2017), hereinafter ALLEGORICO in view of Knox et al. (US 2007/0176783 A1, pub. date: 08/02/2007), hereinafter Knox and Lennartz et al. (US 2004/0217857 A1, pub. date: 11/04/2004), hereinafter Lennartz .
Independent Claim 1
ALLEGORICO discloses a computing device for filter life prediction for an aspirating air filter system (ALLEGORICO, ¶¶ [0003], [0005], and [0007]: during long periods of operation the filter media may become saturated with particulate matter, which subsequently obstructs or impedes the flow of air and creates a significant pressure drop between the upstream and downstream side of the filter media; an estimation of the residual useful life of a filter arrangement would be of great assistance in suitably programming a maintenance intervention of this kind; provide an accurate and efficient estimation of the residual useful life of a filter arrangement; ¶¶ [0033]-[0035] and [0067] with FIGS. 1-2: the inlet air filter system 13 contains one or more filter arrangements (e.g., filter arrangement 25 in an upstream volume 21, and filter arrangements 29 , 31 , 33 in a downstream volume 23) to prevent or limit ingress of particulate matter, such as dust or salt, or other impurities, which is fluidly coupled through a clean air duct 14 to the suction side of the compressor; residual useful life estimation system configured for estimating the residual useful life of the filter arrangements 25, 29, 31, 33), comprising:
a memory (ALLEGORICO, ¶ [0067] with 45 in FIG. 2: storage unit); and a processor (ALLEGORICO, ¶ [0067] with 43 in FIG. 2: a central control unit) configured to execute executable instructions stored in the memory to (ALLEGORICO, ¶ [0067]: a central control unit 43 can be suitably programmed for performing the above described prognostic method):
receive operational data from the aspirating air filter system, wherein the operational data is received from the aspirating air filter system via a air filter system for a first time period to generate an initial data set (ALLEGORICO, ¶ [0036]: the accumulated particulate matter obstructs the passage of air through the filter arrangement, thus increasing the pressure loss, i.e., the pressure differential across the filter arrangement; the pressure loss, i.e., the pressure drop across the filter arrangement can thus be used as a degradation parameter, which provides information on the degradation of the filter arrangement; ¶¶ [0039]-[0042] with FIG. 2: each filter arrangement 25 , 29 , 31 , 33 is provided with its own residual useful life prognostic system; each prognostic system can comprise a differential pressure measuring arrangement, configured for measuring a pressure differential across the respective filter arrangement; in FIG . 2, a plurality of filter degradation measuring systems 41A , 41B , 41C , 41D are schematically shown, one for each filter arrangement 25 , 29 , 31 , 33; each filter degradation measuring systems can be comprised of a differential pressure measuring arrangement; differential pressure measuring arrangements can be configured to measure the total differential pressure across two or more sequentially arranged filter arrangements; reference profiles of filter degradation are defined as set of data representing a filter degradation parameter as a function of time, which are predetermined "a priori" with respect to the actual measurement of the degradation parameter; each reference profile of filter degradation or degradation curve can therefore be represented as a curve of pressure loss across the filter arrangement versus time; the pressure loss can be expressed in mmH2O or other pressure unit of measurement; ¶ [0043] with FIG. 3: represent the experimental data measured on seven different arrangements, e.g. operating in different environmental conditions and/or in different operating conditions; ¶ [0011]: the reference degradation curves are predetermined and can be obtained by data on existing filter arrangements; reference degradation curves can be generated starting from a limited number of experimentally obtained curves; in particular, the predetermined reference curves can be based on historical data of degradation of the filter arrangement; ¶¶ [0055]-[0056] with FIGS. 2 and 5: a threshold value Δp_alarm of the pressure drop across the filter arrangement is reported on the vertical axis ; upon reaching the threshold value Δp_ alarm the filter arrangement requires to be changed; the pressure drop Δp (i.e., the degradation parameter) across the filter arrangement has been measured by the respective pressure measuring arrangement, e.g., 41A , 41B , 41C or 41D (FIG. 2), between the start of the filter operation (T = 0) and the actual time Ti; at time Ti the future trend of the degradation curve CF shall be predicted, in order to obtain an estimation of the residual useful life of the filter arrangement, which is the coordinate on the abscissa of the point of intersection between the curve CF and the "ALARM line"; ¶¶ [0064]-[0068]: use matrices of data where the pressure drop values for the reference degradation curves Cj are stored; each reference degradation curve Cj can thus be defined by a set of coordinates (Ti, Xj,i); at each time instant Ti the degradation value detected by the differential pressure measuring arrangement is stored together with the previously detected values and the buffer of data (xi-n, xi) are used, together with the stored data defining the reference degradation curves Cj, in equations (2), (4) and (5) to calculate the estimated residual useful life of the filter arrangement; measured values of the degradation parameter can be stored in a storage unit 45, together with data defining the reference degradation curves; a global pressure drop across the entire set of filter arrangements or a sub–group thereof can be measured; a central control unit 43 can be suitably programmed for performing the above described prognostic method), wherein the aspirating air filter system includes a filter (ALLEGORICO, ¶¶ [0033]-[0035] with FIGS. 1-2: the inlet air filter system 13 contains one or more filter arrangements (e.g., filter arrangement 25 in an upstream volume 21, and filter arrangements 29 , 31 , 33 in a downstream volume 23) to prevent or limit ingress of particulate matter, such as dust or salt, or other impurities, which is fluidly coupled through a clean air duct 14 to the suction side of the compressor);
fit a machine learning model to the initial data set (ALLEGORICO, ¶¶ [0044]-[0053] with FIGS. 3-4: the larger the number of available reference degradation curves, the more accurate the prediction of the residual useful life will be; if an insufficient number of reference degradation curves is available, or if a larger number of such reference degradation curves is desirable, artificial reference degradation curves can be generated, starting from a relatively small number of experimental curves; generating artificial reference degradation curves can start from observing that each reference degradation curve can be fitted e.g. with a mathematical model, i.e., a mathematical curve fitting the actual reference degradation curve can be defined; an exponential model can be used, which can be defined as follows: Δp = a + b·exp(cT) (1), where a, b and c are fitting coefficients and T is time; to generate artificial reference degradation curves starting from the seven experimental reference degradation curves plotted in FIG. 3, coefficients (a), (b) and (c) can be manipulated and combined to obtain other sets of coefficients which, once introduced in formula (1) fit artificially generated reference degradation curves; a further mathematical models can be based e.g. on the following equation : Δp = a + bT + cT2 (1a); another mathematical setting model can be based on two equations:
∆
p
=
a
+
b
T
,
T
≤
T
1
a
+
b
∙
e
-
c
T
,
T
>
T
1
(1b); ¶ [0074]: all measured data of pressure drop across the filter arrangement, collected during the operation period thereof, can be used for fitting the curve and estimate the trend thereof towards final life);
determine, based on the machine learning model, a remaining useful life of the filter (ALLEGORICO, ¶¶ [0037]-[0039]: a filter arrangement requires to be changed when the degradation parameter achieves a threshold value, i.e., if the pressure differential, i.e., the pressure loss across the filter arrangement reaches a threshold value; each filter arrangement has a residual useful life (hereunder also indicated as RUL), which can be expressed in operating hours available before the degradation parameter reaches the threshold value; the residual useful life of the filter arrangement can thus be defined in general terms as the available operation time before the level of filter saturation will be such as to cause the pressure differential to reach the threshold value; ¶ [0041]: the residual useful life of a filter arrangement is estimated on the basis of a prognostic approach, which uses sets of predetermined experimentally obtained and/or artificially generated reference profiles of filter degradation, also named reference degradation curves; ¶¶ [0054]-[0063] with FIGS. 5-6: FIG . 5 shows four experimentally or artificially generated degradation curves C1, C2, C3, C4; a threshold value Δp_alarm of the pressure drop across the filter arrangement is reported on the vertical axis; upon reaching the threshold value Δp_alarm the filter arrangement requires to be changed; the intersection of the curves C1-C4 with the horizontal "ALARM" line corresponding to the threshold value defines the residual useful life of a filter which behaves according to the respective degradation curve; curve CF is the actual degradation curve of the filter arrangement, the degradation whereof shall be predicted; at time Ti the future trend of the degradation curve CF shall be predicted, in order to obtain an estimation of the residual useful life of the filter arrangement, which is the coordinate on the abscissa of the point of intersection between the curve CF and the "ALARM line"; the predictive method can be based on a data-driven similarity-based approach; at any instant in time (Ti) the shape of the curve CF is predicted based on the shapes of the reference degradation curves Cj (j = 1-4 in the example shown in FIG . 5); the similarity of the curve CF with the reference degradation curves Cj is calculated and used to estimate the shape of the curve CF from the time instant Ti to the end of its useful life; a first step of the predictive method can comprise the calculation of the Euclidean distance between curve CF and the reference degradation curve Cj in a time interval or time frame [Ti-n, Ti]; the calculation can be performed on a sequence of observations which includes n points, i.e., n measurements of the actual pressure loss across the filter arrangement; next, each Euclidean distance dj calculated with equation (2) can be transformed into a similarity measure using a Gaussian kernel; the final residual useful life of the filter arrangement at a time instant Ti is then obtained by a similarity weighted sum based on residual useful life values of the set of used reference degradation curves; ¶ [0068]-[0069]: a global pressure drop measured across the entire set of filter arrangements or a sub–group can be used to determine a global residual useful life of the entire group of filter arrangements or a sub-group thereof; the above described process is repeated sequentially runtime, such that at each instant in time an updated estimation of RUL (Ti) can be obtained; each time the calculation is performed, a different time frame can be used; the residual useful life of the filter arrangement is re-calculated each time on the basis of a different buffer of data , such that at each calculation the most significant portion of the reference degradation curves Cj is used);
generate an alert to replace the filter in response to the remaining useful life of the filter exceeding a threshold amount (ALLEGORICO, ¶¶ [0037]-[0039]: a filter arrangement requires to be changed when the degradation parameter achieves a threshold value, i.e. if the pressure differential, i.e. the pressure loss across the filter arrangement reaches a threshold value; ¶ [0055] with FIG. 5: a threshold value Δp_alarm of the pressure drop across the filter arrangement is reported on the vertical axis; upon reaching the threshold value Δp_alarm the filter arrangement requires to be changed; the intersection of the curves C1-C4 with the horizontal "ALARM" line corresponding to the threshold value defines the residual useful life of a filter which behaves according to the respective degradation curve); and
.
ALLEGORICO fails to explicitly disclose wherein (1) an aspirating air filter system is an aspirating smoke detector (for filter life prediction); (2) the operational data is received from the aspirating smoke detector via a control panel of an alarm system; and (2) transmit the alert to the control panel of the alarm system.
Knox teaches a system and method for indicating a time at which the filter requires replacing or filter end-of-life is signaled (Knox, ¶¶ [0092]-[0094]), wherein an aspirating air filter system is an aspirating smoke detector (Knox, ¶¶ [0005]-[0008]: aspirated smoke detection systems using optical scatter detectors to detect the amount of scattered light and hence is able to provide an output signal indicative of the amount of smoke particles or other pollutant particles within the sample flow; a difficulty arises in operation of aspirated smoke detector systems of the above kind in that most atmospheres where smoke or fire detection is required contain dust which may interfere with operation of the system; a filter may therefore be incorporated into the system for the purpose of keeping dust away from sensitive optical Surfaces and to prevent dust from artificially affecting the detection of particles indicative of the presence of fire and/or smoke; over time a filter used to reduce dust transmittance into the detection chamber will eventually fill with dust, which may prevent passage therethrough of not only dust particles, but also smoke particles. This causes the effective sensitivity of the detector system to drop; for this reason it is desirable to be able to detect filter blocking before it causes problems in smoke detection; ¶¶ [0030], [0035], and [0061]-[0094]: determine particle transmittance of a filter of an aspirated particle detector system; the flow rate is a significant factor in the determination of a filter's Smoke particle transmittance; detecting a level of first particles having a size indicative of Smoke particles and which particles are sus pended in air passing through the detection system; determining the flow rate of air passing through the detection system; determining an integrated smoke hours value by integrating the detected level of first particles over time; determining an estimated smoke particle transmittance of the filter in accordance with an operation comprising multiplying the integrated smoke hours value with the determined flow rate; the "integrated smoke hours" value defined above is, generally, a measure of cumulative filter blockage over time by smoke like particles and that general measurement is referred to hereinafter as "“integrated smoke hours"; particle transmittance as referred to herein is defined as the ratio of detectable particle level output by a filter to the detectable particle level input to the filter; produce or flag a filter warning or fault condition when the estimated transmittance reaches a threshold at which a predetermined reduction of the transmittance of the filter may be deemed to indicate an unacceptable degradation in filter performance; ¶¶ [0117]-[0118] with FIG. 1: an aspirated Smoke detector 2; filter 25, such as a volume foam filter having pores, will accumulate particles within the pores over its life) (Knox, ¶¶ [0015], [0017]-[0018], and [0092]-[0094]: indicating a first level filter warning when the estimated Smoke particle transmittance is less than or equal to the first threshold value; indicating a second level filter warning when the estimated Smoke particle transmittance is less than or equal to the second threshold value; produce or flag a filter warning or fault condition when the estimated transmittance reaches a threshold at which a predetermined reduction of the transmittance of the filter may be deemed to indicate an unacceptable degradation in filter performance); and the operational data is received from the aspirating smoke detector via a control system (Knox, ¶¶ [0058]-[0094]: integral function in Eqn. 1 applies to measurements where the flow rate of air in a detector system is not taken into account; integral function in Eqn. 2 applies to a measurement involving constant flow rate; the "integrated smoke hours" value defined above is, generally, a measure of cumulative filter blockage over time by Smoke like particles and that general measurement is referred to hereinafter as "integrated smoke hours"; ¶¶ [0118]-[0119]: blocked pores will not let dust or all smoke particles through, but may still let air through at flow rates and with pressure drops that are very close to the initial conditions, thus making it impractical to detect a filter which is substantially blocked to smoke by monitoring airflow or pressure drop alone; flow sensors are typically used in aspirated smoke detectors to recognize failures of the aspirator (fan) and to recognize gross failures of the sampling pipe network such as breakage or blockage of sampling holes; flow sensors, however, cannot determine when a filter has become significantly blocked due to the trans mission of smoke particles as air will continue to pass largely unhindered through the filter medium, even if a significant proportion of the particles within the air passing through the filter are not transmitted; ¶ [0122] with FIG. 1: the controller circuit 16 may control alarm apparatus, such as a suitable display 18 to indicate the level of detected smoke, based on the light level detected by the detector 12; ¶ [0126] with FIG. 5: integrating circuitry 166 may be incorporated into the controller to integrate the measured or recorded signal of the smoke like particles over time).
ALLEGORICO and Knox are analogous art because they are from the same field of endeavor, a system and method for indicating a time at which the filter requires replacing or filter end-of-life is signaled. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to apply the teaching of Knox to ALLEGORICO. Motivation for doing so would expand life prediction method to different types of system or devices.
ALLEGORICO in view of Knox fails to explicitly disclose to wherein (1) the operational data is received from the smoke detector via a control panel of alarm system; and (2) transmit the alert to the control panel of the alarm system.
Lennartz a system and a method relating to smoke detectors (Lennartz, ¶ [0001]), wherein the operational data is received from the smoke detector via a control panel of alarm system (Lennartz, ¶ [0002]: the alarm systems typically include an alarm panel which receives and monitors signals from a host of peripheral devices, including keypads, various sensors and warning devices; ¶ [0004]: each alarm system typically has a number of sensors which report to the alarm panel; ¶ [0036] with FIG. 1: an alarm panel 4 which basically receives and processes signals from the various peripheral devices 8, including sensors 10, key pads 12 and sounders 14; the alarm pane 4 also cooperates with an external communication network generally shown as 6 for communicating with a central station); and transmit the alert to the control panel of the alarm system (Lennartz, ¶ [0002]: the control panels, upon receiving notice of an alarm condition typically report to a remote central station over a telephone line or other communication path; ¶ [0022]: reporting alarm conditions to a control panel over a wired network; ¶ [0009]: monitoring the performance characteristics of smoke detectors, and the value of transmitting the assessment of the performance of the smoke detector to an alarm control panel, or to a portable device; ¶¶ [0029]-[0030]: a separate computer contains a log of the operating characteristics of each smoke detector and assesses changes in the operating characteristics for possible preventative service of smoke detectors where changes in the operating characteristics are indicative of potential inadequate performance; the separate computer analyzes the operating characteristics for possible conditions which can be rectified by cleaning of the smoke detectors; ¶ [0038] with FIG.1: smoke detectors used in association with this type of alarm System, are subject to decreasing performance due to age and decreasing performance due to environmental contaminations such as dust, etc.; for this reason, it is known to test alarm systems and in particular, test smoke alarm detectors on a scheduled basis; to assist in this type of routine inspection and evaluation, smoke detectors, in addition to reporting alarm conditions to the alarm panel 4, can report performance evaluation characteristics, either to the alarm panel or to a separate device; ¶ [0050]: this type of transmission in a weak RF signal, can be used for both wireless Smoke detectors which transmit RF Signals as well as hard wired smoke detectors which normally communicate over hard wires to the alarm panel).
ALLEGORICO in view of Knox, and Lennartz are analogous art because they are from the same field of endeavor, a system and a method relating to smoke detectors. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to apply the teaching of Lennartz to ALLEGORICO in view of Knox. Motivation for doing so would reduce false alarms and to provide.
Claim 2
ALLEGORICO in view of Knox and Lennartz discloses all the elements as stated in Claim 1 and further discloses to: log additional operational data of the aspirating smoke detector for a second time period; and append the additional operational data to the initial data set to generate an appended data set (ALLEGORICO, ¶¶ [0056]-[0066] and [0069] with FIGS. 5-6: at time Ti the future trend of the degradation curve CF shall be predicted, in order to obtain an estimation of the residual useful life of the filter arrangement; at any instant in time (Ti) the shape of the curve CF is predicted based on the shapes of the reference degradation curves Cj (j = 1-4 in the example shown in FIG . 5); the similarity of the curve CF with the reference degradation curves Cj is calculated and used to estimate the shape of the curve CF from the time instant Ti to the end of its useful life; the calculation can be performed on a sequence of observations which includes n points, i.e., n measurements of the actual pressure loss across the filter arrangement;; the time interval or time frame [Ti-n, Ti] can be pre-set and constant throughout the execution of the predictive method; in other embodiments, the time interval or time frame [Ti-n; Ti] can be modified; at each time instant Ti the degradation value detected by the differential pressure measuring arrangement is stored together with the previously detected values and the buffer of data (xi-n, xi) are used, together with the stored data defining the reference degradation curves Cj, in equations (2), (4) and (5) to calculate the estimated residual useful life of the filter arrangement; the above described process is repeated sequentially runtime, such that at each instant in time an updated estimation of RUL(Ti) can be obtained; each time the calculation is performed, a different time frame can be used; ¶ [0074]: all measured data of pressure drop across the filter arrangement, collected during the operation period thereof, can be used for fitting the curve and estimate the trend thereof towards final life).
Claim 3
ALLEGORICO in view of Knox and Lennartz discloses all the elements as stated in Claim 2 and further discloses to: refit the machine learning model to the appended data set; and determine, based on the refit machine learning model, a revised remaining useful life of the filter (ALLEGORICO, ¶¶ [0056]-[0066] and [0069] with FIGS. 5-6: at time Ti the future trend of the degradation curve CF shall be predicted, in order to obtain an estimation of the residual useful life of the filter arrangement; at any instant in time (Ti) the shape of the curve CF is predicted based on the shapes of the reference degradation curves Cj (j = 1-4 in the example shown in FIG . 5); the similarity of the curve CF with the reference degradation curves Cj is calculated and used to estimate the shape of the curve CF from the time instant Ti to the end of its useful life; the calculation can be performed on a sequence of observations which includes n points, i.e., n measurements of the actual pressure loss across the filter arrangement;; the time interval or time frame [Ti-n, Ti] can be pre-set and constant throughout the execution of the predictive method; in other embodiments, the time interval or time frame [Ti-n; Ti] can be modified; at each time instant Ti the degradation value detected by the differential pressure measuring arrangement is stored together with the previously detected values and the buffer of data (xi-n, xi) are used, together with the stored data defining the reference degradation curves Cj, in equations (2), (4) and (5) to calculate the estimated residual useful life of the filter arrangement; the above described process is repeated sequentially runtime, such that at each instant in time an updated estimation of RUL(Ti) can be obtained; each time the calculation is performed, a different time frame can be used; ¶ [0074]: all measured data of pressure drop across the filter arrangement, collected during the operation period thereof, can be used for fitting the curve and estimate the trend thereof towards final life).
Claim 4
ALLEGORICO in view of Knox and Lennartz discloses all the elements as stated in Claim 2 and further discloses wherein the second time period is shorter than the first time period. (ALLEGORICO, ¶¶ [0056]-[0066] and [0069] with FIGS. 5-6: at time Ti the future trend of the degradation curve CF shall be predicted, in order to obtain an estimation of the residual useful life of the filter arrangement; at any instant in time (Ti) the shape of the curve CF is predicted based on the shapes of the reference degradation curves Cj (j = 1-4 in the example shown in FIG . 5); the similarity of the curve CF with the reference degradation curves Cj is calculated and used to estimate the shape of the curve CF from the time instant Ti to the end of its useful life; the calculation can be performed on a sequence of observations which includes n points, i.e., n measurements of the actual pressure loss across the filter arrangement; the time interval or time frame [Ti-n, Ti] can be pre-set and constant throughout the execution of the predictive method; in other embodiments, the time interval or time frame [Ti-n; Ti] can be modified; i.e., time fame can be same or different (larger or smaller); at each time instant Ti the degradation value detected by the differential pressure measuring arrangement is stored together with the previously detected values and the buffer of data (xi-n, xi) are used, together with the stored data defining the reference degradation curves Cj, in equations (2), (4) and (5) to calculate the estimated residual useful life of the filter arrangement; the above described process is repeated sequentially runtime, such that at each instant in time an updated estimation of RUL(Ti) can be obtained; each time the calculation is performed, a different time frame can be used)
Claim 5
ALLEGORICO in view of Knox and Lennartz discloses all the elements as stated in Claim 1 and further discloses wherein the machine learning model is a linear polynomial model (ALLEGORICO, ¶¶ [0052]-[0053]: a further mathematical models can be based e.g. on the following equation : Δp = a + bT + cT2 (1a); another mathematical setting model can be based on two equations:
∆
p
=
a
+
b
T
,
T
≤
T
1
a
+
b
∙
e
-
c
T
,
T
>
T
1
(1b); i.e., when c= 0, equation (1) become a linear polynomial model as shown in upper formular of equation (1b) when T is less or equal to T1).
Claim 6
ALLEGORICO in view of Knox and Lennartz discloses all the elements as stated in Claim 5 and further discloses to determine a slope value of the linear polynomial model (ALLEGORICO, ¶¶ [0045]-[0053] with FIGS. 3-4: generating artificial reference degradation curves can start from observing that each reference degradation curve can be fitted e.g. with a mathematical model, i.e., a mathematical curve fitting the actual reference degradation curve can be defined; an exponential model can be used, which can be defined as follows: Δp = a + b·exp(cT) (1), where a, b and c are fitting coefficients and T is time; to generate artificial reference degradation curves starting from the seven experimental reference degradation curves plotted in FIG. 3, coefficients (a), (b) and (c) can be manipulated and combined to obtain other sets of coefficients which, once introduced in formula (1) fit artificially generated reference degradation curves; a further mathematical models can be based e.g. on the following equation : Δp = a + bT + cT2 (1a); another mathematical setting model can be based on two equations:
∆
p
=
a
+
b
T
,
T
≤
T
1
a
+
b
∙
e
-
c
T
,
T
>
T
1
(1b)).
Claim 7
ALLEGORICO in view of Knox and Lennartz discloses all the elements as stated in Claim 6 and further discloses to determine an error in the remaining useful life of the filter using the slope value (ALLEGORICO, ¶¶ [0045]-[0053] with FIGS. 3-4: generating artificial reference degradation curves can start from observing that each reference degradation curve can be fitted e.g. with a mathematical model, i.e., a mathematical curve fitting the actual reference degradation curve can be defined; an exponential model can be used, which can be defined as follows: Δp = a + b·exp(cT) (1), where a, b and c are fitting coefficients and T is time; to generate artificial reference degradation curves starting from the seven experimental reference degradation curves plotted in FIG. 3, coefficients (a), (b) and (c) can be manipulated and combined to obtain other sets of coefficients which, once introduced in formula (1) fit artificially generated reference degradation curves; a further mathematical models can be based e.g. on the following equation : Δp = a + bT + cT2 (1a); another mathematical setting model can be based on two equations:
∆
p
=
a
+
b
T
,
T
≤
T
1
a
+
b
∙
e
-
c
T
,
T
>
T
1
(1b); ¶¶ [0054]-[0066] with FIGS. 5-6: FIG . 5 shows four experimentally or artificially generated degradation curves C1, C2, C3, C4; a threshold value Δp_alarm of the pressure drop across the filter arrangement is reported on the vertical axis; upon reaching the threshold value Δp_alarm the filter arrangement requires to be changed; the intersection of the curves C1-C4 with the horizontal "ALARM" line corresponding to the threshold value defines the residual useful life of a filter which behaves according to the respective degradation curve; curve CF is the actual degradation curve of the filter arrangement, the degradation whereof shall be predicted; at time Ti the future trend of the degradation curve CF shall be predicted, in order to obtain an estimation of the residual useful life of the filter arrangement, which is the coordinate on the abscissa of the point of intersection between the curve CF and the "ALARM line"; the predictive method can be based on a data-driven similarity-based approach; at any instant in time (Ti) the shape of the curve CF is predicted based on the shapes of the reference degradation curves Cj (j = 1-4 in the example shown in FIG . 5); the similarity of the curve CF with the reference degradation curves Cj is calculated and used to estimate the shape of the curve CF from the time instant Ti to the end of its useful life; a first step of the predictive method can comprise the calculation of the Euclidean distance between curve CF and the reference degradation curve Cj in a time interval or time frame [Ti-n, Ti]; the calculation can be performed on a sequence of observations which includes n points, i.e., n measurements of the actual pressure loss across the filter arrangement; next, each Euclidean distance dj calculated with equation (2) can be transformed into a similarity measure using a Gaussian kernel; the final residual useful life of the filter arrangement at a time instant Ti is then obtained by a similarity weighted sum based on residual useful life values of the set of used reference degradation curves; ¶¶ [0070]-[0079] with FIGS. 5-7: pure data-driven, similarity-based prognostic approach can be modified or corrected when the filter arrangement approaches the end of its useful life; correction is obtained by combining the data-driven approach with a physics-based approach, wherein the estimation of the residual useful life is obtained by fitting the measured data representing curve CF (FIGS. 5, 6) with a fitting equation, e.g. a linear or a quadratic fitting model; in other words, a regression is applied to the measured degradation parameter data to obtain an estimation of the trend of curve CF towards the end of the filter life; different regression methods, either linear or non-linear, can be used; a hybrid method is thus obtained which is partially data-driven and partially physics-based; the hybrid method more reliably ensures convergence of the prediction error to 0; equations (1), (1a) or (1b) or other linear on non-linear regressions can be used to generate a model of the degradation curve CF, based on a buffer of data obtained by measurement during the period of operation of the filter arrangement; all measured data of pressure drop across the filter arrangement, collected during the operation period thereof, can be used for fitting the curve and estimate the trend thereof towards final life; in other embodiments; only a reduced frame of the collected data, e.g., those relating to the most recent period of use of the filter arrangement, can be used in the calculation; the time interval used can e.g. be approximately 1/10 of the expected total useful life of the filter; measurements of the last 1000 hours or so can be used in the regression; a hybrid prognostic or estimation method can be designed, wherein the physics-based prediction becomes gradually predominant over the data-driven prediction, as the final life of the filter arrangement is approached; to ensure a smooth transition between the data-driven method and the physics-based method, a combination of the two methods can be used, based on the following formula: RUL(Ti) = f·RULSB(Ti) + (1-f) ·RULPB(Ti) (5) wherein
f
=
m
e
d
i
a
n
0
;
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
;
1
, and wherein RULSB is the residual useful life estimated on the basis of the pure data-driven, similarity-based approach [i.e., equations (2), (3), (4)]; and RULPR is the residual useful life estimated on the basis of the regression only, i.e., on the pure physics-based approach, Δp_lim is a limit value of the pressure drop across the filter, which is near the threshold value Δp_alarm that determines the end of the useful filter life, see FIGS . 5 and 6 ; and x is the pressure drop value, i.e., the degradation parameter, measured at a given time instant T; the two estimated values are combined using the parameter (f), which takes the median (intermediate) value between 0,
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
, and 1; FIG. 7 shows the value of parameter (f) as a function of x; the degradation parameter, i.e., the pressure drop or pressure differential across the filter arrangement is constantly or intermittently measured by the respective differential pressure measuring arrangement (41A , 41B , 41C or 41D); as far as the measured degradation parameter x is below Δp_lim, the parameter ( f ) takes the value "1", since this is the median value between 0,
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
, and 1; as soon as the measured degradation parameter x reaches and then becomes greater than Δp_lim, the parameter (f) becomes
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
; the value of (f) gradually increases while the degradation parameter x increases from Δp_lim, towards Δp_alarm; consequently, the weight of the physics-based estimation of the residual useful life RULPR of the filter arrangement becomes gradually predominant over the data driven, i.e., similarity- based estimation RULSE; by combining the two methods in the hybrid method summarized in equation (6) the prediction error converges to 0, this resulting in an extremely accurate prognostic method for determining the residual useful life of the filter arrangement; i.e., the error of final residual useful life is depending on the coefficients of equations (1a) or (1b), where the coefficient b is the slope).
Claim 8
ALLEGORICO in view of Knox and Lennartz discloses all the elements as stated in Claim 1 and further discloses wherein the machine learning model is a quadratic polynomial model (ALLEGORICO, ¶ [0052]: a further mathematical models can be based e.g. on the following equation : Δp = a + bT + cT2 (1a)).
Claim 9
ALLEGORICO in view of Knox and Lennartz discloses all the elements as stated in Claim 8 and further discloses to determine a constant value of the quadratic polynomial model (ALLEGORICO, ¶¶ [0045]-[0053] with FIGS. 3-4: generating artificial reference degradation curves can start from observing that each reference degradation curve can be fitted e.g. with a mathematical model, i.e., a mathematical curve fitting the actual reference degradation curve can be defined; an exponential model can be used, which can be defined as follows: Δp = a + b·exp(cT) (1), where a, b and c are fitting coefficients and T is time; to generate artificial reference degradation curves starting from the seven experimental reference degradation curves plotted in FIG. 3, coefficients (a), (b) and (c) can be manipulated and combined to obtain other sets of coefficients which, once introduced in formula (1) fit artificially generated reference degradation curves; a further mathematical models can be based e.g. on the following equation : Δp = a + bT + cT2 (1a); another mathematical setting model can be based on two equations:
∆
p
=
a
+
b
T
,
T
≤
T
1
a
+
b
∙
e
-
c
T
,
T
>
T
1
(1b)).
Claim 10
ALLEGORICO in view of Knox and Lennartz discloses all the elements as stated in Claim 9 and further discloses to determine an error in the remaining useful life of the filter using the constant value (ALLEGORICO, ¶¶ [0045]-[0053] with FIGS. 3-4: generating artificial reference degradation curves can start from observing that each reference degradation curve can be fitted e.g. with a mathematical model, i.e., a mathematical curve fitting the actual reference degradation curve can be defined; an exponential model can be used, which can be defined as follows: Δp = a + b·exp(cT) (1), where a, b and c are fitting coefficients and T is time; to generate artificial reference degradation curves starting from the seven experimental reference degradation curves plotted in FIG. 3, coefficients (a), (b) and (c) can be manipulated and combined to obtain other sets of coefficients which, once introduced in formula (1) fit artificially generated reference degradation curves; a further mathematical models can be based e.g. on the following equation : Δp = a + bT + cT2 (1a); another mathematical setting model can be based on two equations:
∆
p
=
a
+
b
T
,
T
≤
T
1
a
+
b
∙
e
-
c
T
,
T
>
T
1
(1b); ¶¶ [0054]-[0066] with FIGS. 5-6: FIG . 5 shows four experimentally or artificially generated degradation curves C1, C2, C3, C4; a threshold value Δp_alarm of the pressure drop across the filter arrangement is reported on the vertical axis; upon reaching the threshold value Δp_alarm the filter arrangement requires to be changed; the intersection of the curves C1-C4 with the horizontal "ALARM" line corresponding to the threshold value defines the residual useful life of a filter which behaves according to the respective degradation curve; curve CF is the actual degradation curve of the filter arrangement, the degradation whereof shall be predicted; at time Ti the future trend of the degradation curve CF shall be predicted, in order to obtain an estimation of the residual useful life of the filter arrangement, which is the coordinate on the abscissa of the point of intersection between the curve CF and the "ALARM line"; the predictive method can be based on a data-driven similarity-based approach; at any instant in time (Ti) the shape of the curve CF is predicted based on the shapes of the reference degradation curves Cj (j = 1-4 in the example shown in FIG . 5); the similarity of the curve CF with the reference degradation curves Cj is calculated and used to estimate the shape of the curve CF from the time instant Ti to the end of its useful life; a first step of the predictive method can comprise the calculation of the Euclidean distance between curve CF and the reference degradation curve Cj in a time interval or time frame [Ti-n, Ti]; the calculation can be performed on a sequence of observations which includes n points, i.e., n measurements of the actual pressure loss across the filter arrangement; next, each Euclidean distance dj calculated with equation (2) can be transformed into a similarity measure using a Gaussian kernel; the final residual useful life of the filter arrangement at a time instant Ti is then obtained by a similarity weighted sum based on residual useful life values of the set of used reference degradation curves; ¶¶ [0070]-[0079] with FIGS. 5-7: pure data-driven, similarity-based prognostic approach can be modified or corrected when the filter arrangement approaches the end of its useful life; correction is obtained by combining the data-driven approach with a physics-based approach, wherein the estimation of the residual useful life is obtained by fitting the measured data representing curve CF (FIGS. 5, 6) with a fitting equation, e.g. a linear or a quadratic fitting model; in other words, a regression is applied to the measured degradation parameter data to obtain an estimation of the trend of curve CF towards the end of the filter life; different regression methods, either linear or non-linear, can be used; a hybrid method is thus obtained which is partially data-driven and partially physics-based; the hybrid method more reliably ensures convergence of the prediction error to 0; equations (1), (1a) or (1b) or other linear on non-linear regressions can be used to generate a model of the degradation curve CF, based on a buffer of data obtained by measurement during the period of operation of the filter arrangement; all measured data of pressure drop across the filter arrangement, collected during the operation period thereof, can be used for fitting the curve and estimate the trend thereof towards final life; in other embodiments; only a reduced frame of the collected data, e.g., those relating to the most recent period of use of the filter arrangement, can be used in the calculation; the time interval used can e.g. be approximately 1/10 of the expected total useful life of the filter; measurements of the last 1000 hours or so can be used in the regression; a hybrid prognostic or estimation method can be designed, wherein the physics-based prediction becomes gradually predominant over the data-driven prediction, as the final life of the filter arrangement is approached; to ensure a smooth transition between the data-driven method and the physics-based method, a combination of the two methods can be used, based on the following formula: RUL(Ti) = f·RULSB(Ti) + (1-f) ·RULPB(Ti) (5) wherein
f
=
m
e
d
i
a
n
0
;
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
;
1
, and wherein RULSB is the residual useful life estimated on the basis of the pure data-driven, similarity-based approach [i.e., equations (2), (3), (4)]; and RULPR is the residual useful life estimated on the basis of the regression only, i.e., on the pure physics-based approach, Δp_lim is a limit value of the pressure drop across the filter, which is near the threshold value Δp_alarm that determines the end of the useful filter life, see FIGS . 5 and 6 ; and x is the pressure drop value, i.e., the degradation parameter, measured at a given time instant T; the two estimated values are combined using the parameter (f), which takes the median (intermediate) value between 0,
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
, and 1; FIG. 7 shows the value of parameter (f) as a function of x; the degradation parameter, i.e., the pressure drop or pressure differential across the filter arrangement is constantly or intermittently measured by the respective differential pressure measuring arrangement (41A , 41B , 41C or 41D); as far as the measured degradation parameter x is below Δp_lim, the parameter ( f ) takes the value "1", since this is the median value between 0,
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
, and 1; as soon as the measured degradation parameter x reaches and then becomes greater than Δp_lim, the parameter (f) becomes
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
; the value of (f) gradually increases while the degradation parameter x increases from Δp_lim, towards Δp_alarm; consequently, the weight of the physics-based estimation of the residual useful life RULPR of the filter arrangement becomes gradually predominant over the data driven, i.e., similarity- based estimation RULSE; by combining the two methods in the hybrid method summarized in equation (6) the prediction error converges to 0, this resulting in an extremely accurate prognostic method for determining the residual useful life of the filter arrangement; i.e., the error of final residual useful life is depending on the coefficients of equations (1a) or (1b), where the coefficient a is the constant).
Claim 11
ALLEGORICO in view of Knox and Lennartz discloses all the elements as stated in Claim 1 and further discloses wherein the computing device is a fire system gateway device (Knox, ¶ [0004]: fire protection and suppressant systems may operate by detecting the presence of smoke and other airborne pollutants or, in general, particles; ¶ [0112] with FIG. 1: a smoke and fire detection system; ¶ [0126] with FIG. 5: the controller 16 may also comprise circuitry 164 for recording the steady signal indicative of fire hazard smoke particles and other non-fire hazard smoke like particles; ¶ [0131]: continually monitoring of the condition of a filter and alleviates the need for excessively frequent testing and maintenance of a smoke or fire detection system; accordingly, the risk of failure of the detection system to operate in the event of fire is reduced).
Claims 12-13 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over ALLEGORICO in view of Knox, HERSHEY et al. (US 2017/0286572 A1, pub. date: 10/05/2017), hereinafter HERSHEY, and Lennartz.
Independent Claim 12
ALLEGORICO discloses a system for filter life prediction for an aspirating air filter system (ALLEGORICO, ¶¶ [0003], [0005], and [0007]: during long periods of operation the filter media may become saturated with particulate matter, which subsequently obstructs or impedes the flow of air and creates a significant pressure drop between the upstream and downstream side of the filter media; an estimation of the residual useful life of a filter arrangement would be of great assistance in suitably programming a maintenance intervention of this kind; provide an accurate and efficient estimation of the residual useful life of a filter arrangement; ¶¶ [0033]-[0035] and [0067] with FIGS. 1-2: the inlet air filter system 13 contains one or more filter arrangements (e.g., filter arrangement 25 in an upstream volume 21, and filter arrangements 29 , 31 , 33 in a downstream volume 23) to prevent or limit ingress of particulate matter, such as dust or salt, or other impurities, which is fluidly coupled through a clean air duct 14 to the suction side of the compressor; residual useful life estimation system configured for estimating the residual useful life of the filter arrangements 25, 29, 31, 33), comprising:
an aspirating air filter system, wherein the aspirating air filter system includes a filter (ALLEGORICO, ¶¶ [0033]-[0035] with FIGS. 1-2: the inlet air filter system 13 contains one or more filter arrangements (e.g., filter arrangement 25 in an upstream volume 21, and filter arrangements 29 , 31 , 33 in a downstream volume 23) to prevent or limit ingress of particulate matter, such as dust or salt, or other impurities, which is fluidly coupled through a clean air duct 14 to the suction side of the compressor) and a sensor (ALLEGORICO, ¶¶ [0039]-[0041] with FIG. 2: each filter arrangement 25 , 29 , 31 , 33 is provided with its own residual useful life prognostic system; each prognostic system can comprise a differential pressure measuring arrangement, configured for measuring a pressure differential across the respective filter arrangement; in FIG . 2, a plurality of filter degradation measuring systems 41A , 41B , 41C , 41D are schematically shown, one for each filter arrangement 25 , 29 , 31 , 33; each filter degradation measuring systems can be comprised of a differential pressure measuring arrangement; differential pressure measuring arrangements can be configured to measure the total differential pressure across two or more sequentially arranged filter arrangements;); and a computing device coupled to the aspirating air filter system via a control (ALLEGORICO, ¶ [0067] with FIG. 2: a central control unit 43 can be suitably programmed for performing the above described prognostic method; measured values of the degradation parameter can be stored in a storage unit 45, together with data defining the reference degradation curves; a generic interface 47, e.g. a monitor, which provides the operator with information on the predicted residual useful life of the various filter arrangements) is configured to:
generate a air filter system from a physical-based model of the aspirating air filter system; calibrate the air filter system; receive, from the sensor via the control air filter system; determine a remaining useful life of the filter by running the calibrated digital twin model using the real-time operational data (ALLEGORICO, ¶ [0036]: the accumulated particulate matter obstructs the passage of air through the filter arrangement, thus increasing the pressure loss, i.e., the pressure differential across the filter arrangement; the pressure loss, i.e., the pressure drop across the filter arrangement can thus be used as a degradation parameter, which provides information on the degradation of the filter arrangement; ¶¶ [0039]-[0042] with FIG. 2: each filter arrangement 25 , 29 , 31 , 33 is provided with its own residual useful life prognostic system; each prognostic system can comprise a differential pressure measuring arrangement, configured for measuring a pressure differential across the respective filter arrangement; in FIG . 2, a plurality of filter degradation measuring systems 41A , 41B , 41C , 41D are schematically shown, one for each filter arrangement 25 , 29 , 31 , 33; each filter degradation measuring systems can be comprised of a differential pressure measuring arrangement; differential pressure measuring arrangements can be configured to measure the total differential pressure across two or more sequentially arranged filter arrangements; reference profiles of filter degradation are defined as set of data representing a filter degradation parameter as a function of time, which are predetermined "a priori" with respect to the actual measurement of the degradation parameter; each reference profile of filter degradation or degradation curve can therefore be represented as a curve of pressure loss across the filter arrangement versus time; the pressure loss can be expressed in mmH2O or other pressure unit of measurement; ¶ [0043] with FIG. 3: represent the experimental data measured on seven different arrangements, e.g. operating in different environmental conditions and/or in different operating conditions; ¶ [0011]: the reference degradation curves are predetermined and can be obtained by data on existing filter arrangements; reference degradation curves can be generated starting from a limited number of experimentally obtained curves; in particular, the predetermined reference curves can be based on historical data of degradation of the filter arrangement; ¶¶ [0055]-[0056] with FIGS. 2 and 5: a threshold value Δp_alarm of the pressure drop across the filter arrangement is reported on the vertical axis ; upon reaching the threshold value Δp_ alarm the filter arrangement requires to be changed; the pressure drop Δp (i.e., the degradation parameter) across the filter arrangement has been measured by the respective pressure measuring arrangement, e.g., 41A , 41B , 41C or 41D (FIG. 2), between the start of the filter operation (T = 0) and the actual time Ti; at time Ti the future trend of the degradation curve CF shall be predicted, in order to obtain an estimation of the residual useful life of the filter arrangement, which is the coordinate on the abscissa of the point of intersection between the curve CF and the "ALARM line"; ¶¶ [0064]-[0068]: use matrices of data where the pressure drop values for the reference degradation curves Cj are stored; each reference degradation curve Cj can thus be defined by a set of coordinates (Ti, Xj,i); at each time instant Ti the degradation value detected by the differential pressure measuring arrangement is stored together with the previously detected values and the buffer of data (xi-n, xi) are used, together with the stored data defining the reference degradation curves Cj, in equations (2), (4) and (5) to calculate the estimated residual useful life of the filter arrangement; measured values of the degradation parameter can be stored in a storage unit 45, together with data defining the reference degradation curves; a global pressure drop across the entire set of filter arrangements or a sub–group thereof can be measured; a central control unit 43 can be suitably programmed for performing the above described prognostic method; ¶¶ [0044]-[0053] with FIGS. 3-4: the larger the number of available reference degradation curves, the more accurate the prediction of the residual useful life will be; if an insufficient number of reference degradation curves is available, or if a larger number of such reference degradation curves is desirable, artificial reference degradation curves can be generated, starting from a relatively small number of experimental curves; generating artificial reference degradation curves can start from observing that each reference degradation curve can be fitted e.g. with a mathematical model, i.e., a mathematical curve fitting the actual reference degradation curve can be defined; an exponential model can be used, which can be defined as follows: Δp = a + b·exp(cT) (1), where a, b and c are fitting coefficients and T is time; to generate artificial reference degradation curves starting from the seven experimental reference degradation curves plotted in FIG. 3, coefficients (a), (b) and (c) can be manipulated and combined to obtain other sets of coefficients which, once introduced in formula (1) fit artificially generated reference degradation curves; a further mathematical models can be based e.g. on the following equation : Δp = a + bT + cT2 (1a); another mathematical setting model can be based on two equations:
∆
p
=
a
+
b
T
,
T
≤
T
1
a
+
b
∙
e
-
c
T
,
T
>
T
1
(1b); ¶ [0074]: all measured data of pressure drop across the filter arrangement, collected during the operation period thereof, can be used for fitting the curve and estimate the trend thereof towards final life; ¶¶ [0037]-[0039]: a filter arrangement requires to be changed when the degradation parameter achieves a threshold value, i.e., if the pressure differential, i.e., the pressure loss across the filter arrangement reaches a threshold value; each filter arrangement has a residual useful life (hereunder also indicated as RUL), which can be expressed in operating hours available before the degradation parameter reaches the threshold value; the residual useful life of the filter arrangement can thus be defined in general terms as the available operation time before the level of filter saturation will be such as to cause the pressure differential to reach the threshold value; ¶ [0041]: the residual useful life of a filter arrangement is estimated on the basis of a prognostic approach, which uses sets of predetermined experimentally obtained and/or artificially generated reference profiles of filter degradation, also named reference degradation curves; ¶¶ [0054]-[0063] with FIGS. 5-6: FIG . 5 shows four experimentally or artificially generated degradation curves C1, C2, C3, C4; a threshold value Δp_alarm of the pressure drop across the filter arrangement is reported on the vertical axis; upon reaching the threshold value Δp_alarm the filter arrangement requires to be changed; the intersection of the curves C1-C4 with the horizontal "ALARM" line corresponding to the threshold value defines the residual useful life of a filter which behaves according to the respective degradation curve; curve CF is the actual degradation curve of the filter arrangement, the degradation whereof shall be predicted; at time Ti the future trend of the degradation curve CF shall be predicted, in order to obtain an estimation of the residual useful life of the filter arrangement, which is the coordinate on the abscissa of the point of intersection between the curve CF and the "ALARM line"; the predictive method can be based on a data-driven similarity-based approach; at any instant in time (Ti) the shape of the curve CF is predicted based on the shapes of the reference degradation curves Cj (j = 1-4 in the example shown in FIG . 5); the similarity of the curve CF with the reference degradation curves Cj is calculated and used to estimate the shape of the curve CF from the time instant Ti to the end of its useful life; a first step of the predictive method can comprise the calculation of the Euclidean distance between curve CF and the reference degradation curve Cj in a time interval or time frame [Ti-n, Ti]; the calculation can be performed on a sequence of observations which includes n points, i.e., n measurements of the actual pressure loss across the filter arrangement; next, each Euclidean distance dj calculated with equation (2) can be transformed into a similarity measure using a Gaussian kernel; the final residual useful life of the filter arrangement at a time instant Ti is then obtained by a similarity weighted sum based on residual useful life values of the set of used reference degradation curves; ¶ [0068]-[0069]: a global pressure drop measured across the entire set of filter arrangements or a sub–group can be used to determine a global residual useful life of the entire group of filter arrangements or a sub-group thereof; the above described process is repeated sequentially runtime, such that at each instant in time an updated estimation of RUL (Ti) can be obtained; each time the calculation is performed, a different time frame can be used; the residual useful life of the filter arrangement is re-calculated each time on the basis of a different buffer of data , such that at each calculation the most significant portion of the reference degradation curves Cj is used; ¶¶ [0070]-[0079] with FIGS. 5-7: pure data-driven, similarity-based prognostic approach can be modified or corrected when the filter arrangement approaches the end of its useful life; correction is obtained by combining the data-driven approach with a physics-based approach, wherein the estimation of the residual useful life is obtained by fitting the measured data representing curve CF (FIGS. 5, 6) with a fitting equation, e.g. a linear or a quadratic fitting model; in other words, a regression is applied to the measured degradation parameter data to obtain an estimation of the trend of curve CF towards the end of the filter life; different regression methods, either linear or non-linear, can be used; a hybrid method is thus obtained which is partially data-driven and partially physics-based; the hybrid method more reliably ensures convergence of the prediction error to 0; equations (1), (1a) or (1b) or other linear on non-linear regressions can be used to generate a model of the degradation curve CF, based on a buffer of data obtained by measurement during the period of operation of the filter arrangement; all measured data of pressure drop across the filter arrangement, collected during the operation period thereof, can be used for fitting the curve and estimate the trend thereof towards final life; in other embodiments; only a reduced frame of the collected data, e.g., those relating to the most recent period of use of the filter arrangement, can be used in the calculation; the time interval used can e.g. be approximately 1/10 of the expected total useful life of the filter; measurements of the last 1000 hours or so can be used in the regression; a hybrid prognostic or estimation method can be designed, wherein the physics-based prediction becomes gradually predominant over the data-driven prediction, as the final life of the filter arrangement is approached; to ensure a smooth transition between the data-driven method and the physics-based method, a combination of the two methods can be used, based on the following formula: RUL(Ti) = f·RULSB(Ti) + (1-f) ·RULPB(Ti) (5) wherein
f
=
m
e
d
i
a
n
0
;
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
;
1
, and wherein RULSB is the residual useful life estimated on the basis of the pure data-driven, similarity-based approach [i.e., equations (2), (3), (4)]; and RULPR is the residual useful life estimated on the basis of the regression only, i.e., on the pure physics-based approach, Δp_lim is a limit value of the pressure drop across the filter, which is near the threshold value Δp_alarm that determines the end of the useful filter life, see FIGS . 5 and 6 ; and x is the pressure drop value, i.e., the degradation parameter, measured at a given time instant T; the two estimated values are combined using the parameter (f), which takes the median (intermediate) value between 0,
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
, and 1; FIG. 7 shows the value of parameter (f) as a function of x; the degradation parameter, i.e., the pressure drop or pressure differential across the filter arrangement is constantly or intermittently measured by the respective differential pressure measuring arrangement (41A , 41B , 41C or 41D); as far as the measured degradation parameter x is below Δp_lim, the parameter ( f ) takes the value "1", since this is the median value between 0,
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
, and 1; as soon as the measured degradation parameter x reaches and then becomes greater than Δp_lim, the parameter (f) becomes
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
; the value of (f) gradually increases while the degradation parameter x increases from Δp_lim, towards Δp_alarm; consequently, the weight of the physics-based estimation of the residual useful life RULPR of the filter arrangement becomes gradually predominant over the data driven, i.e., similarity- based estimation RULSE; by combining the two methods in the hybrid method summarized in equation (6) the prediction error converges to 0, this resulting in an extremely accurate prognostic method for determining the residual useful life of the filter arrangement);
generate an alert to replace the filter in response to the remaining useful life of the filter exceeding a threshold amount (ALLEGORICO, ¶¶ [0037]-[0039]: a filter arrangement requires to be changed when the degradation parameter achieves a threshold value, i.e. if the pressure differential, i.e. the pressure loss across the filter arrangement reaches a threshold value; ¶ [0055] with FIG. 5: a threshold value Δp_alarm of the pressure drop across the filter arrangement is reported on the vertical axis; upon reaching the threshold value Δp_alarm the filter arrangement requires to be changed; the intersection of the curves C1-C4 with the horizontal "ALARM" line corresponding to the threshold value defines the residual useful life of a filter which behaves according to the respective degradation curve); and
.
ALLEGORICO fails to explicitly disclose wherein (1) an aspirating air filter system is an aspirating smoke detector (for filter life prediction); (2) a computing device coupled to the aspirating smoke detector via a control panel of an alarm system; (3) a model is a digital twin model from a reduced order model; (4) receive, from the sensor via the control panel of the alarm system, operational data; and (5) transmit the alert to the control panel of the alarm system.
Knox teaches a system and method for indicating a time at which the filter requires replacing or filter end-of-life is signaled (Knox, ¶¶ [0092]-[0094]), wherein an aspirating air filter system is an aspirating smoke detector (Knox, ¶¶ [0005]-[0008]: aspirated smoke detection systems using optical scatter detectors to detect the amount of scattered light and hence is able to provide an output signal indicative of the amount of smoke particles or other pollutant particles within the sample flow; a difficulty arises in operation of aspirated smoke detector systems of the above kind in that most atmospheres where smoke or fire detection is required contain dust which may interfere with operation of the system; a filter may therefore be incorporated into the system for the purpose of keeping dust away from sensitive optical Surfaces and to prevent dust from artificially affecting the detection of particles indicative of the presence of fire and/or smoke; over time a filter used to reduce dust transmittance into the detection chamber will eventually fill with dust, which may prevent passage therethrough of not only dust particles, but also smoke particles. This causes the effective sensitivity of the detector system to drop; for this reason it is desirable to be able to detect filter blocking before it causes problems in smoke detection; ¶¶ [0030], [0035], and [0061]-[0094]: determine particle transmittance of a filter of an aspirated particle detector system; the flow rate is a significant factor in the determination of a filter's Smoke particle transmittance; detecting a level of first particles having a size indicative of Smoke particles and which particles are sus pended in air passing through the detection system; determining the flow rate of air passing through the detection system; determining an integrated smoke hours value by integrating the detected level of first particles over time; determining an estimated smoke particle transmittance of the filter in accordance with an operation comprising multiplying the integrated smoke hours value with the determined flow rate; the "integrated smoke hours" value defined above is, generally, a measure of cumulative filter blockage over time by smoke like particles and that general measurement is referred to hereinafter as "“integrated smoke hours"; particle transmittance as referred to herein is defined as the ratio of detectable particle level output by a filter to the detectable particle level input to the filter; produce or flag a filter warning or fault condition when the estimated transmittance reaches a threshold at which a predetermined reduction of the transmittance of the filter may be deemed to indicate an unacceptable degradation in filter performance; ¶¶ [0117]-[0118] with FIG. 1: an aspirated Smoke detector 2; filter 25, such as a volume foam filter having pores, will accumulate particles within the pores over its life); a computing device coupled to the aspirating smoke detector via a control system; and receive, from the sensor via the control system, operational data (Knox, ¶¶ [0058]-[0094]: integral function in Eqn. 1 applies to measurements where the flow rate of air in a detector system is not taken into account; integral function in Eqn. 2 applies to a measurement involving constant flow rate; the "integrated smoke hours" value defined above is, generally, a measure of cumulative filter blockage over time by Smoke like particles and that general measurement is referred to hereinafter as "integrated smoke hours"; ¶¶ [0118]-[0119]: blocked pores will not let dust or all smoke particles through, but may still let air through at flow rates and with pressure drops that are very close to the initial conditions, thus making it impractical to detect a filter which is substantially blocked to smoke by monitoring airflow or pressure drop alone; flow sensors are typically used in aspirated smoke detectors to recognize failures of the aspirator (fan) and to recognize gross failures of the sampling pipe network such as breakage or blockage of sampling holes; flow sensors, however, cannot determine when a filter has become significantly blocked due to the trans mission of smoke particles as air will continue to pass largely unhindered through the filter medium, even if a significant proportion of the particles within the air passing through the filter are not transmitted; ¶ [0122] with FIG. 1: the controller circuit 16 may control alarm apparatus, such as a suitable display 18 to indicate the level of detected smoke, based on the light level detected by the detector 12; ¶ [0126] with FIG. 5: integrating circuitry 166 may be incorporated into the controller to integrate the measured or recorded signal of the smoke like particles over time)
ALLEGORICO and Knox are analogous art because they are from the same field of endeavor, a system and method for indicating a time at which the filter requires replacing or filter end-of-life is signaled. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to apply the teaching of Knox to ALLEGORICO. Motivation for doing so would expand life prediction method to different types of system or devices.
ALLEGORICO in view of Knox fails to explicitly disclose wherein (1) a computing device coupled to the smoke detector via a control panel of an alarm system; (2) a model is a digital twin model from a reduced order model; (3) receive, from the sensor via the control panel of the alarm system, operational data; and (4) transmit the alert to the control panel of the alarm system.
HERSHEY teaches a system and a method relating to predicting/estimating remaining useful life (HERSHEY, ¶¶ [0001] and [0035]-[0036]), wherein a model is a digital twin model from a reduced order model (HERSHEY, ¶¶ [0037], [0041]-[0045], and [0059] with FIG. 2B: a digital twin may estimate a remaining useful life of a twinned physical system using sensors, communications, modeling, history, and computation; a Per Asset digital twin may comprise a model of the structural components, their physical functions, and/or their interactions; a Per Asset digital twin may also track and perform calculations associated with estimating a system's remaining useful life; a Per Asset digital twin may be configured to function as a continually tuned digital twin, a digital twin that is continually updated as its twinned physical system is on-operation; an adaptable digital twin is designed to adapt to new scenarios and new system configurations and may be transferred to another system or class of systems, and/or one of a plurality of interacting digital twins that are scalable over an asset class and may be broadened to not only model a twinned physical system but also provide control over the asset; the outputs from the digital twin 250 may include a continually updated estimate of the twinned physical system's Remaining Useful Life ("RUL"); ¶¶ [0098]-[0104]: computation times needed to solve exact equations may exceed the time required for a result in order to monitor, protect, and/or effectively prognosticate concerning a twinned physical system; for this reason, it may be desirable to use computational approximations by employing such techniques as linearization, reduced order modeling, fuzzy logic, and/or neural network; in the case of Reduced Order Modeling ("ROM"), software for evaluating damage and predicting RUL or the time to failure of a twinned physical system may be formed by appropriate extractions from full digital twin code; these extractions may in tum be reduced in complexity by approximations; an additional approach in using a ROM digital twin is to use a discrete event simulation approach and essentially adjust the granularity of the time increments used in running the models; the ROM of a Digital Twin ("RO MDT") may an approximation of the ideal digital twin and the approximations may represent the physical models, their integration, and/or the complete state spaces of the components).
ALLEGORIC and HERSHEY are analogous art because they are from the same field of endeavor, a system and a method relating to predicting/estimating remaining useful life. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to apply the teaching of HERSHEY to ALLEGORICO in view of Knox. Motivation for doing so would (1) facilitate .
ALLEGORICO in view of Knox and HERSHEY fails to explicitly disclose to wherein (1) a computing device coupled to the smoke detector via a control panel of an alarm system; (2) receive, from the sensor via the control panel of the alarm system, operational data; and (3) transmit the alert to the control panel of the alarm system.
Lennartz a system and a method relating to smoke detectors (Lennartz, ¶ [0001]), wherein a computing device coupled to the smoke detector via a control panel of an alarm system; receive, from the sensor via the control panel of the alarm system, operational data (Lennartz, ¶ [0002]: the alarm systems typically include an alarm panel which receives and monitors signals from a host of peripheral devices, including keypads, various sensors and warning devices; ¶ [0004]: each alarm system typically has a number of sensors which report to the alarm panel; ¶ [0036] with FIG. 1: an alarm panel 4 which basically receives and processes signals from the various peripheral devices 8, including sensors 10, key pads 12 and sounders 14; the alarm pane 4 also cooperates with an external communication network generally shown as 6 for communicating with a central station); and transmit the alert to the control panel of the alarm system (Lennartz, ¶ [0002]: the control panels, upon receiving notice of an alarm condition typically report to a remote central station over a telephone line or other communication path; ¶ [0022]: reporting alarm conditions to a control panel over a wired network; ¶ [0009]: monitoring the performance characteristics of smoke detectors, and the value of transmitting the assessment of the performance of the smoke detector to an alarm control panel, or to a portable device; ¶¶ [0029]-[0030]: a separate computer contains a log of the operating characteristics of each smoke detector and assesses changes in the operating characteristics for possible preventative service of smoke detectors where changes in the operating characteristics are indicative of potential inadequate performance; the separate computer analyzes the operating characteristics for possible conditions which can be rectified by cleaning of the smoke detectors; ¶ [0038] with FIG.1: smoke detectors used in association with this type of alarm System, are subject to decreasing performance due to age and decreasing performance due to environmental contaminations such as dust, etc.; for this reason, it is known to test alarm systems and in particular, test smoke alarm detectors on a scheduled basis; to assist in this type of routine inspection and evaluation, smoke detectors, in addition to reporting alarm conditions to the alarm panel 4, can report performance evaluation characteristics, either to the alarm panel or to a separate device; ¶ [0050]: this type of transmission in a weak RF signal, can be used for both wireless Smoke detectors which transmit RF Signals as well as hard wired smoke detectors which normally communicate over hard wires to the alarm panel).
ALLEGORICO in view of Knox and HERSHEY, and Lennartz are analogous art because they are from the same field of endeavor, a system and a method relating to an aspirating smoke detection system. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to apply the teaching of Lennartz to ALLEGORICO in view of Knox and HERSHEY. Motivation for doing so would reduce false alarms and to provide early indication of deterioration for preventative maintenance (Lennartz, ¶¶ [0005], [0011], [0029], [0050], and [0054]).
Claim 13
ALLEGORICO in view of Knox, HERSHEY, and Lennartz discloses all the elements as stated in Claim 12 and further discloses to convert a predefined computational fluid dynamics (CFD) model of the aspirating smoke detector into the reduced order model (HERSHEY, ¶¶ [0052] and [0098]-[0104] with FIG. 2A: a digital twin may have two functions: monitoring a twinned physical system and performing prognostics on it; another function of a digital twin may comprise a limited or total control of the twinned physical system; a digital twin of a twinned physical system consists of (1) one or more sensors sensing the values of designated parameters of the twinned physical system and (2) an ultra-realistic computer model of all of the subject system's multiple elements and their interactions under a spectrum of conditions; this may be implemented using a computer model having substantial number of degrees of freedom and may be associated with an integration of complex physical models for computational fluid dynamics 202, structural dynamics 204, thermodynamic modeling 206, stress analysis modeling 210, and/or a fatigue cracking model 208; such an approach may be associated with, e.g., a Unified Physics Model ("UPM"); computation times needed to solve exact equations may exceed the time required for a result in order to monitor, protect, and/or effectively prognosticate concerning a twinned physical system; for this reason, it may be desirable to use computational approximations by employing such techniques as linearization, reduced order modeling, fuzzy logic, and/or neural network; in the case of Reduced Order Modeling ("ROM"), software for evaluating damage and predicting RUL or the time to failure of a twinned physical system may be formed by appropriate extractions from full digital twin code; these extractions may in tum be reduced in complexity by approximations; an additional approach in using a ROM digital twin is to use a discrete event simulation approach and essentially adjust the granularity of the time increments used in running the models; the ROM of a Digital Twin ("RO MDT") may an approximation of the ideal digital twin and the approximations may represent the physical models, their integration, and/or the complete state spaces of the components).
Claim 15
ALLEGORICO in view of Knox, HERSHEY, and Lennartz discloses all the elements as stated in Claim 12 and further discloses wherein the initial data set includes at least one of: a flow rate of gas through the filter; a pressure differential over the filter; and smoke particulate detected in the aspirating smoke detector (ALLEGORICO, ¶ [0003]: during long periods of operation the filter media may become saturated with particulate matter , which subsequently obstructs or impedes the flow of air and creates a significant pressure drop between the upstream and downstream side of the filter media; ¶ [0006]: existing methods for estimating the residual useful life of filters require an accurate model of the degradation phenomenon and a number of difficult-to-measure or unknown parameters, such as the air flow rate, environmental data, etc.; ¶ [0013]-[0014]: the degradation parameter is the pressure loss, i.e. the pressure drop across the filter arrangement; the higher the pressure drop across the filter arrangement, the higher the degradation of the filter; the pressure drop across said single filter system is detected; ¶¶ [0036]-[0037] and [0039]-[0042] with FIG. 2: the accumulated particulate matter obstructs the passage of air through the filter arrangement, thus increasing the pressure loss, i.e. the pressure differential across the filter arrangement; the pressure loss, i.e. the pressure drop across the filter arrangement can thus be used as a degradation parameter, which provides information on the degradation of the filter arrangement; a filter arrangement requires to be changed when the degradation parameter achieves a threshold value, i.e. if the pressure differential, i.e. the pressure loss across the filter arrangement reaches a threshold value; each prognostic system can comprise a differential pressure measuring arrangement , configured for measuring a pressure differential across the respective filter arrangement).
Claim 14 is rejected under 35 U.S.C. 103 as being unpatentable over ALLEGORICO in view of Knox, HERSHEY, and Lennartz as applied to Claim 13 above, and further in view of Straw et al. ("Predictive Digital Twin for Performance and Integrity", Offshore Technology Conference, Houston, Texas, May 2022, pp. 1-14), hereinafter Straw.
Claim 14
ALLEGORICO in view of Knox, HERSHEY, and Lennartz discloses all the elements as stated in Claim 13 and further discloses that putting sensors, and even intelligence, into basic parts may expand the number of dimensions of any particular system so that no two systems will stay strictly identical as they age through different operational, control transient, and/or environmental conditions (HERSHEY, ¶ [0071]).
ALLEGORICO in view of Knox, HERSHEY, and Lennartz fails to explicitly disclose wherein the predefined CFD model is a transient model.
Straw discloses a system and a method relating to predictive approaches using digital twin (Straw, Abstract in Page 1), wherein the predefined CFD model is a transient model (Straw, Section "Building a reliable system simulation or ROM" with FIGS. 12-14 in Pages 12-14: a system simulation may often be built without the need to train or tune it using higher fidelity approaches; it depends on what is required to be taken from the data obtained and the system complexity; for geometrically or operationally complex engineering systems, a system simulation may need to be trained to obtain accurate data; the first of the four CFD cases in FIG. 12 was used to train the system simulation; in this case it was used to tune local heat transfer coefficients and thermal characteristics where complex geometrical features exist in the system or insulation design; using this training data, the other three cases were used to validate the system simulation to ensure it was able to predict temperatures accurately across all the conditions the equipment may experience; when the same tuned system simulation was applied to the remaining three cases, an excellent correlation can be as shown in Figure 14; very good agreement is gained across all locations between CFD and system simulation data; this same correlation is obtained at other locations through the system; in this way the transient thermal response of the complete subsea jumper was captured, and the system simulation provides a real-time prediction of the temperatures allowing risk of hydrate formation to be assessed).
ALLEGORICO in view of Knox and Straw are analogous art because they are from the same field of endeavor, a system and a method relating to predictive approaches using digital twin. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to apply the teaching of Straw to ALLEGORICO in view of Knox, HERSHEY, and Lennartz. Motivation for doing so would completely capture real-world physical behavior in simulation (.
Claims 16-20 are rejected under 35 U.S.C. 103 as being unpatentable over ALLEGORICO in view of Knox, Adams et al. ("Hybrid Digital Twins: A Primer on Combining Physics-Based and Data Analytics Approaches", IEEE Software, Vol. 39, 02/14/2022, pp. 47-52), hereinafter Adams, and Lennartz.
Independent Claim 16
ALLEGORICO discloses a method for filter life prediction for an aspirating air filter system (ALLEGORICO, ¶¶ [0003], [0005], and [0007]: during long periods of operation the filter media may become saturated with particulate matter, which subsequently obstructs or impedes the flow of air and creates a significant pressure drop between the upstream and downstream side of the filter media; an estimation of the residual useful life of a filter arrangement would be of great assistance in suitably programming a maintenance intervention of this kind; provide an accurate and efficient estimation of the residual useful life of a filter arrangement; ¶¶ [0033]-[0035] and [0067] with FIGS. 1-2: the inlet air filter system 13 contains one or more filter arrangements (e.g., filter arrangement 25 in an upstream volume 21, and filter arrangements 29 , 31 , 33 in a downstream volume 23) to prevent or limit ingress of particulate matter, such as dust or salt, or other impurities, which is fluidly coupled through a clean air duct 14 to the suction side of the compressor; residual useful life estimation system configured for estimating the residual useful life of the filter arrangements 25, 29, 31, 33), comprising:
receive, by a computing device, operational data from the aspirating air filter system, wherein the operational data is received from the aspirating air filter system via a control (ALLEGORICO, ¶ [0067] with FIG. 2: a central control unit 43 can be suitably programmed for performing the above described prognostic method; measured values of the degradation parameter can be stored in a storage unit 45, together with data defining the reference degradation curves; a generic interface 47, e.g. a monitor, which provides the operator with information on the predicted residual useful life of the various filter arrangements), the operational data from the aspirating air filter system to generate an initial data set (ALLEGORICO, ¶ [0036]: the accumulated particulate matter obstructs the passage of air through the filter arrangement, thus increasing the pressure loss, i.e., the pressure differential across the filter arrangement; the pressure loss, i.e., the pressure drop across the filter arrangement can thus be used as a degradation parameter, which provides information on the degradation of the filter arrangement; ¶¶ [0039]-[0042] with FIG. 2: each filter arrangement 25 , 29 , 31 , 33 is provided with its own residual useful life prognostic system; each prognostic system can comprise a differential pressure measuring arrangement, configured for measuring a pressure differential across the respective filter arrangement; in FIG . 2, a plurality of filter degradation measuring systems 41A , 41B , 41C , 41D are schematically shown, one for each filter arrangement 25 , 29 , 31 , 33; each filter degradation measuring systems can be comprised of a differential pressure measuring arrangement; differential pressure measuring arrangements can be configured to measure the total differential pressure across two or more sequentially arranged filter arrangements; reference profiles of filter degradation are defined as set of data representing a filter degradation parameter as a function of time, which are predetermined "a priori" with respect to the actual measurement of the degradation parameter; each reference profile of filter degradation or degradation curve can therefore be represented as a curve of pressure loss across the filter arrangement versus time; the pressure loss can be expressed in mmH2O or other pressure unit of measurement; ¶ [0043] with FIG. 3: represent the experimental data measured on seven different arrangements, e.g. operating in different environmental conditions and/or in different operating conditions; ¶ [0011]: the reference degradation curves are predetermined and can be obtained by data on existing filter arrangements; reference degradation curves can be generated starting from a limited number of experimentally obtained curves; in particular, the predetermined reference curves can be based on historical data of degradation of the filter arrangement; ¶¶ [0055]-[0056] with FIGS. 2 and 5: a threshold value Δp_alarm of the pressure drop across the filter arrangement is reported on the vertical axis ; upon reaching the threshold value Δp_ alarm the filter arrangement requires to be changed; the pressure drop Δp (i.e., the degradation parameter) across the filter arrangement has been measured by the respective pressure measuring arrangement, e.g., 41A , 41B , 41C or 41D (FIG. 2), between the start of the filter operation (T = 0) and the actual time Ti; at time Ti the future trend of the degradation curve CF shall be predicted, in order to obtain an estimation of the residual useful life of the filter arrangement, which is the coordinate on the abscissa of the point of intersection between the curve CF and the "ALARM line"; ¶¶ [0064]-[0068]: use matrices of data where the pressure drop values for the reference degradation curves Cj are stored; each reference degradation curve Cj can thus be defined by a set of coordinates (Ti, Xj,i); at each time instant Ti the degradation value detected by the differential pressure measuring arrangement is stored together with the previously detected values and the buffer of data (xi-n, xi) are used, together with the stored data defining the reference degradation curves Cj, in equations (2), (4) and (5) to calculate the estimated residual useful life of the filter arrangement; measured values of the degradation parameter can be stored in a storage unit 45, together with data defining the reference degradation curves; a global pressure drop across the entire set of filter arrangements or a sub–group thereof can be measured; a central control unit 43 can be suitably programmed for performing the above described prognostic method);
fitting, by the computing device, a machine learning model to the initial data set (ALLEGORICO, ¶¶ [0044]-[0053] with FIGS. 3-4: the larger the number of available reference degradation curves, the more accurate the prediction of the residual useful life will be; if an insufficient number of reference degradation curves is available, or if a larger number of such reference degradation curves is desirable, artificial reference degradation curves can be generated, starting from a relatively small number of experimental curves; generating artificial reference degradation curves can start from observing that each reference degradation curve can be fitted e.g. with a mathematical model, i.e., a mathematical curve fitting the actual reference degradation curve can be defined; an exponential model can be used, which can be defined as follows: Δp = a + b·exp(cT) (1), where a, b and c are fitting coefficients and T is time; to generate artificial reference degradation curves starting from the seven experimental reference degradation curves plotted in FIG. 3, coefficients (a), (b) and (c) can be manipulated and combined to obtain other sets of coefficients which, once introduced in formula (1) fit artificially generated reference degradation curves; a further mathematical models can be based e.g. on the following equation : Δp = a + bT + cT2 (1a); another mathematical setting model can be based on two equations:
∆
p
=
a
+
b
T
,
T
≤
T
1
a
+
b
∙
e
-
c
T
,
T
>
T
1
(1b); ¶ [0074]: all measured data of pressure drop across the filter arrangement, collected during the operation period thereof, can be used for fitting the curve and estimate the trend thereof towards final life);
determining, by the computing device based on the machine learning model, a first predicted remaining useful life of a filter of the aspirating air filter system (ALLEGORICO, ¶¶ [0037]-[0039]: a filter arrangement requires to be changed when the degradation parameter achieves a threshold value, i.e., if the pressure differential, i.e., the pressure loss across the filter arrangement reaches a threshold value; each filter arrangement has a residual useful life (hereunder also indicated as RUL), which can be expressed in operating hours available before the degradation parameter reaches the threshold value; the residual useful life of the filter arrangement can thus be defined in general terms as the available operation time before the level of filter saturation will be such as to cause the pressure differential to reach the threshold value; ¶ [0041]: the residual useful life of a filter arrangement is estimated on the basis of a prognostic approach, which uses sets of predetermined experimentally obtained and/or artificially generated reference profiles of filter degradation, also named reference degradation curves; ¶¶ [0054]-[0063] with FIGS. 5-6: FIG . 5 shows four experimentally or artificially generated degradation curves C1, C2, C3, C4; a threshold value Δp_alarm of the pressure drop across the filter arrangement is reported on the vertical axis; upon reaching the threshold value Δp_alarm the filter arrangement requires to be changed; the intersection of the curves C1-C4 with the horizontal "ALARM" line corresponding to the threshold value defines the residual useful life of a filter which behaves according to the respective degradation curve; curve CF is the actual degradation curve of the filter arrangement, the degradation whereof shall be predicted; at time Ti the future trend of the degradation curve CF shall be predicted, in order to obtain an estimation of the residual useful life of the filter arrangement, which is the coordinate on the abscissa of the point of intersection between the curve CF and the "ALARM line"; the predictive method can be based on a data-driven similarity-based approach; at any instant in time (Ti) the shape of the curve CF is predicted based on the shapes of the reference degradation curves Cj (j = 1-4 in the example shown in FIG . 5); the similarity of the curve CF with the reference degradation curves Cj is calculated and used to estimate the shape of the curve CF from the time instant Ti to the end of its useful life; a first step of the predictive method can comprise the calculation of the Euclidean distance between curve CF and the reference degradation curve Cj in a time interval or time frame [Ti-n, Ti]; the calculation can be performed on a sequence of observations which includes n points, i.e., n measurements of the actual pressure loss across the filter arrangement; next, each Euclidean distance dj calculated with equation (2) can be transformed into a similarity measure using a Gaussian kernel; the final residual useful life of the filter arrangement at a time instant Ti is then obtained by a similarity weighted sum based on residual useful life values of the set of used reference degradation curves; ¶ [0068]-[0069]: a global pressure drop measured across the entire set of filter arrangements or a sub–group can be used to determine a global residual useful life of the entire group of filter arrangements or a sub-group thereof; the above described process is repeated sequentially runtime, such that at each instant in time an updated estimation of RUL (Ti) can be obtained; each time the calculation is performed, a different time frame can be used; the residual useful life of the filter arrangement is re-calculated each time on the basis of a different buffer of data , such that at each calculation the most significant portion of the reference degradation curves Cj is used);
calibrating, by the computing device, a air filter system using the initial data set; determining, by the computing device, a second predicted remaining useful life of the filter by running the calibrated digital twin model using real-time operational data of the aspirating smoke detector air filter system; and determining, by the computing device, a remaining useful life of the filter from the first predicted remaining useful life and the second predicted remaining useful life (ALLEGORICO, ¶¶ [0070]-[0079] with FIGS. 5-7: pure data-driven, similarity-based prognostic approach can be modified or corrected when the filter arrangement approaches the end of its useful life; correction is obtained by combining the data-driven approach with a physics-based approach, wherein the estimation of the residual useful life is obtained by fitting the measured data representing curve CF (FIGS. 5, 6) with a fitting equation, e.g. a linear or a quadratic fitting model; in other words, a regression is applied to the measured degradation parameter data to obtain an estimation of the trend of curve CF towards the end of the filter life; different regression methods, either linear or non-linear, can be used; a hybrid method is thus obtained which is partially data-driven and partially physics-based; the hybrid method more reliably ensures convergence of the prediction error to 0; equations (1), (1a) or (1b) or other linear on non-linear regressions can be used to generate a model of the degradation curve CF, based on a buffer of data obtained by measurement during the period of operation of the filter arrangement; all measured data of pressure drop across the filter arrangement, collected during the operation period thereof, can be used for fitting the curve and estimate the trend thereof towards final life; in other embodiments; only a reduced frame of the collected data, e.g., those relating to the most recent period of use of the filter arrangement, can be used in the calculation; the time interval used can e.g. be approximately 1/10 of the expected total useful life of the filter; measurements of the last 1000 hours or so can be used in the regression; a hybrid prognostic or estimation method can be designed, wherein the physics-based prediction becomes gradually predominant over the data-driven prediction, as the final life of the filter arrangement is approached; to ensure a smooth transition between the data-driven method and the physics-based method, a combination of the two methods can be used, based on the following formula: RUL(Ti) = f·RULSB(Ti) + (1-f) ·RULPB(Ti) (5) wherein
f
=
m
e
d
i
a
n
0
;
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
;
1
, and wherein RULSB is the residual useful life estimated on the basis of the pure data-driven, similarity-based approach [i.e., equations (2), (3), (4)]; and RULPR is the residual useful life estimated on the basis of the regression only, i.e., on the pure physics-based approach, Δp_lim is a limit value of the pressure drop across the filter, which is near the threshold value Δp_alarm that determines the end of the useful filter life, see FIGS . 5 and 6 ; and x is the pressure drop value, i.e., the degradation parameter, measured at a given time instant T; the two estimated values are combined using the parameter (f), which takes the median (intermediate) value between 0,
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
, and 1; FIG. 7 shows the value of parameter (f) as a function of x; the degradation parameter, i.e., the pressure drop or pressure differential across the filter arrangement is constantly or intermittently measured by the respective differential pressure measuring arrangement (41A , 41B , 41C or 41D); as far as the measured degradation parameter x is below Δp_lim, the parameter ( f ) takes the value "1", since this is the median value between 0,
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
, and 1; as soon as the measured degradation parameter x reaches and then becomes greater than Δp_lim, the parameter (f) becomes
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
; the value of (f) gradually increases while the degradation parameter x increases from Δp_lim, towards Δp_alarm; consequently, the weight of the physics-based estimation of the residual useful life RULPR of the filter arrangement becomes gradually predominant over the data driven, i.e., similarity- based estimation RULSE; by combining the two methods in the hybrid method summarized in equation (6) the prediction error converges to 0, this resulting in an extremely accurate prognostic method for determining the residual useful life of the filter arrangement);
generating, by the computing device, an alert to replace the filter in response to the remaining useful life exceeding a threshold amount (ALLEGORICO, ¶¶ [0037]-[0039]: a filter arrangement requires to be changed when the degradation parameter achieves a threshold value, i.e. if the pressure differential, i.e. the pressure loss across the filter arrangement reaches a threshold value; ¶ [0055] with FIG. 5: a threshold value Δp_alarm of the pressure drop across the filter arrangement is reported on the vertical axis; upon reaching the threshold value Δp_alarm the filter arrangement requires to be changed; the intersection of the curves C1-C4 with the horizontal "ALARM" line corresponding to the threshold value defines the residual useful life of a filter which behaves according to the respective degradation curve); and
.
ALLEGORICO fails to explicitly disclose wherein (1) an aspirating air filter system is an aspirating smoke detector (for filter life prediction); (2) the operational data is received from the aspirating smoke detector via a control panel of an alarm system; (3) a model is a twain model; and (4) transmitting, by the computing device, the alert to the control panel of the alarm system.
Knox teaches a system and method for indicating a time at which the filter requires replacing or filter end-of-life is signaled (Knox, ¶¶ [0092]-[0094]), wherein an aspirating air filter system is an aspirating smoke detector (Knox, ¶¶ [0005]-[0008]: aspirated smoke detection systems using optical scatter detectors to detect the amount of scattered light and hence is able to provide an output signal indicative of the amount of smoke particles or other pollutant particles within the sample flow; a difficulty arises in operation of aspirated smoke detector systems of the above kind in that most atmospheres where smoke or fire detection is required contain dust which may interfere with operation of the system; a filter may therefore be incorporated into the system for the purpose of keeping dust away from sensitive optical Surfaces and to prevent dust from artificially affecting the detection of particles indicative of the presence of fire and/or smoke; over time a filter used to reduce dust transmittance into the detection chamber will eventually fill with dust, which may prevent passage therethrough of not only dust particles, but also smoke particles. This causes the effective sensitivity of the detector system to drop; for this reason it is desirable to be able to detect filter blocking before it causes problems in smoke detection; ¶¶ [0030], [0035], and [0061]-[0094]: determine particle transmittance of a filter of an aspirated particle detector system; the flow rate is a significant factor in the determination of a filter's Smoke particle transmittance; detecting a level of first particles having a size indicative of Smoke particles and which particles are sus pended in air passing through the detection system; determining the flow rate of air passing through the detection system; determining an integrated smoke hours value by integrating the detected level of first particles over time; determining an estimated smoke particle transmittance of the filter in accordance with an operation comprising multiplying the integrated smoke hours value with the determined flow rate; the "integrated smoke hours" value defined above is, generally, a measure of cumulative filter blockage over time by smoke like particles and that general measurement is referred to hereinafter as "“integrated smoke hours"; particle transmittance as referred to herein is defined as the ratio of detectable particle level output by a filter to the detectable particle level input to the filter; produce or flag a filter warning or fault condition when the estimated transmittance reaches a threshold at which a predetermined reduction of the transmittance of the filter may be deemed to indicate an unacceptable degradation in filter performance; ¶¶ [0117]-[0118] with FIG. 1: an aspirated Smoke detector 2; filter 25, such as a volume foam filter having pores, will accumulate particles within the pores over its life); and the operational data is received from the aspirating smoke detector via a control system (Knox, ¶¶ [0058]-[0094]: integral function in Eqn. 1 applies to measurements where the flow rate of air in a detector system is not taken into account; integral function in Eqn. 2 applies to a measurement involving constant flow rate; the "integrated smoke hours" value defined above is, generally, a measure of cumulative filter blockage over time by Smoke like particles and that general measurement is referred to hereinafter as "integrated smoke hours"; ¶¶ [0118]-[0119]: blocked pores will not let dust or all smoke particles through, but may still let air through at flow rates and with pressure drops that are very close to the initial conditions, thus making it impractical to detect a filter which is substantially blocked to smoke by monitoring airflow or pressure drop alone; flow sensors are typically used in aspirated smoke detectors to recognize failures of the aspirator (fan) and to recognize gross failures of the sampling pipe network such as breakage or blockage of sampling holes; flow sensors, however, cannot determine when a filter has become significantly blocked due to the trans mission of smoke particles as air will continue to pass largely unhindered through the filter medium, even if a significant proportion of the particles within the air passing through the filter are not transmitted; ¶ [0122] with FIG. 1: the controller circuit 16 may control alarm apparatus, such as a suitable display 18 to indicate the level of detected smoke, based on the light level detected by the detector 12; ¶ [0126] with FIG. 5: integrating circuitry 166 may be incorporated into the controller to integrate the measured or recorded signal of the smoke like particles over time).
ALLEGORICO and Knox are analogous art because they are from the same field of endeavor, a system and method for indicating a time at which the filter requires replacing or filter end-of-life is signaled. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to apply the teaching of Knox to ALLEGORICO. Motivation for doing so would expand life prediction method to different types of system or devices.
ALLEGORICO in view of Knox fails to explicitly disclose wherein (1) the operational data is received from the smoke detector via a control panel of alarm system; (2) a model is a twain model; and (3) transmitting, by the computing device, the alert to the control panel of the alarm system.
Adams teaches a system and a method relating to predicting remaining useful life using hybrid models (Adams, Title in Page 47; Subsection "Coexistence" in Page 49), wherein a model is a twain model (Adams, Pages 47-48: two popular approaches to building digital twins are pure data-based and physics/simulation-based methods; present a framework for hybrid digital twins that combines the strengths of the two approaches, sharing results and demonstrating applicability to a flow network; a digital twin is a model of a physical system whose behavior adheres as closely as possible to the behavior of the actual physical system; both methods can be applied in a single digital twin to leverage all of the available engineering and physics knowledge as well as experimental observations that will show deviations from or omissions in the physical modeling; each of the approaches has different strengths and weaknesses, and in combination the two complement each other; present a framework explaining how the multiple available techniques can be utilized to create what we call a hybrid digital twin; the hybrid digital twin leverages the best available technologies from both physics simulation and data modeling; methods require only sufficient data to train and validate the model; typically, data models are incapable of giving underlying reasons for why the behavior occurred; physics-based models are built from physical equations; consequently, they can predict a physically relevant output for any physically realistic set of inputs; a physics model can simulate what-if scenarios and be used to analyze root causes; physics models are data independent, scalable to a large parameter variability, and extensible; physics-based models require domain expertise; as a system ages and possibly degrades, the as-designed model behavior no longer matches the actual operating conditions; additionally, physics-based models miss any behavior that is not explicitly accounted for in the physical equations; in brownfield scenarios, there is often a lack of both sensor information, making a data modeling approach difficult, and engineering data sheets or component drawings, making a physics-based approach difficult; in such cases, the best approach is to leverage both resources — whatever data and engineering knowledge are available — to build a digital twin from which we can glean useful predictions; building a unique digital twin for each member in a fleet is not scalable; instead, the best approach is to build a single representative digital twin that can be quickly calibrated to match the unique behavior and environment of each member of the fleet; key areas where the hybrid approach is required: low data, scalable deployment, and configurability; Section "A Hybrid Digital Twin Framework" with FIGS. 1-2 in Pages 48-50: a hybrid digital twin is built from three key ingredients: physical models, data, and ML algorithms; Figure 1 depicts how either a bottom-up or top-down approach can be used to create a digital twin; in a bottom-up approach, create a model with the full physical equations of a system; the main benefits are a high-fidelity representation of the physical system behavior and a reuse of models that may already be present from design teams; the top-down approach starts with more general information about a system, such as process and instrumentation diagrams and performance curves; this approach is useful when a twin needs to be built quickly or from an automated workflow, a twin needs to simulate quickly, or more detailed information to build the twin is not available; both techniques converge in the orange-colored region of Figure 1, where ML techniques can be used to supplement the physics-based model by creating data models, calibrating the physical models, or creating reduced-order models (ROMs); at the most basic level, physics-based models, data models, and ROMs can coexist in a single digital twin, each performing independent functions; physical models can be used to improve data models; e.g., when experimental data are lacking, supplement the training of data models with simulated data from a physics model; likewise, data can be used to enhance physics models; typically, this is done by using data to learn better model parameters; model parameters give physical models the flexibility to accurately describe behavior over a wide range of possible system configurations; the last approach shown in Figure 2 is to integrate physics and data in a unified model; compare the predictions of a calibrated physics model to the experimental data and learn the residual; this tight integration between physics and data modeling has several benefits; the digital twin can be deployed faster because it doesn’t require extensive data collection and modeling or deep investments in physics modeling; additionally, the approach is scalable).
ALLEGORICO in view of Knox, and Adams are analogous art because they are from the same field of endeavor, a system and a method relating to predicting remaining useful life using hybrid models. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to apply the teaching of Adams to ALLEGORICO in view of Knox. Motivation for doing so would provide the best possible representation of a system given all available information, which can make accurate prediction in a scalable way (.
ALLEGORICO in view of Knox and Adams fails to explicitly disclose wherein (1) the operational data is received from the smoke detector via a control panel of alarm system; and (2) transmitting, by the computing device, the alert to the control panel of the alarm system.
Lennartz a system and a method relating to smoke detectors (Lennartz, ¶ [0001]), wherein the operational data is received from the smoke detector via a control panel of alarm system (Lennartz, ¶ [0002]: the alarm systems typically include an alarm panel which receives and monitors signals from a host of peripheral devices, including keypads, various sensors and warning devices; ¶ [0004]: each alarm system typically has a number of sensors which report to the alarm panel; ¶ [0036] with FIG. 1: an alarm panel 4 which basically receives and processes signals from the various peripheral devices 8, including sensors 10, key pads 12 and sounders 14; the alarm pane 4 also cooperates with an external communication network generally shown as 6 for communicating with a central station); and transmitting, by the computing device, the alert to the control panel of teh alarm system (Lennartz, ¶ [0002]: the control panels, upon receiving notice of an alarm condition typically report to a remote central station over a telephone line or other communication path; ¶ [0022]: reporting alarm conditions to a control panel over a wired network; ¶ [0009]: monitoring the performance characteristics of smoke detectors, and the value of transmitting the assessment of the performance of the smoke detector to an alarm control panel, or to a portable device; ¶¶ [0029]-[0030]: a separate computer contains a log of the operating characteristics of each smoke detector and assesses changes in the operating characteristics for possible preventative service of smoke detectors where changes in the operating characteristics are indicative of potential inadequate performance; the separate computer analyzes the operating characteristics for possible conditions which can be rectified by cleaning of the smoke detectors; ¶ [0038] with FIG.1: smoke detectors used in association with this type of alarm System, are subject to decreasing performance due to age and decreasing performance due to environmental contaminations such as dust, etc.; for this reason, it is known to test alarm systems and in particular, test smoke alarm detectors on a scheduled basis; to assist in this type of routine inspection and evaluation, smoke detectors, in addition to reporting alarm conditions to the alarm panel 4, can report performance evaluation characteristics, either to the alarm panel or to a separate device; ¶ [0050]: this type of transmission in a weak RF signal, can be used for both wireless Smoke detectors which transmit RF Signals as well as hard wired smoke detectors which normally communicate over hard wires to the alarm panel).
ALLEGORICO in view of Knox and Adams, and Lennartz are analogous art because they are from the same field of endeavor, a system and a method relating to an aspirating smoke detection system. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filling date of the claimed invention to apply the teaching of Lennartz to ALLEGORICO in view of Knox and Adams. Motivation for doing so would reduce false alarms and to provide early indication of deterioration for preventative maintenance (Lennartz, ¶¶ [0005], [0011], [0029], [0050], and [0054]).
Claim 17
ALLEGORICO in view of Knox, Adams, and Lennartz discloses all the elements as stated in Claim 16 and further discloses determining the remaining useful life of the filter includes determining an average between the first predicted remaining useful life and the second predicted remaining useful life (ALLEGORICO, ¶¶ [0070]-[0079] with FIGS. 5-7: pure data-driven, similarity-based prognostic approach can be modified or corrected when the filter arrangement approaches the end of its useful life; correction is obtained by combining the data-driven approach with a physics-based approach, wherein the estimation of the residual useful life is obtained by fitting the measured data representing curve CF (FIGS. 5, 6) with a fitting equation, e.g. a linear or a quadratic fitting model; in other words, a regression is applied to the measured degradation parameter data to obtain an estimation of the trend of curve CF towards the end of the filter life; different regression methods, either linear or non-linear, can be used; a hybrid method is thus obtained which is partially data-driven and partially physics-based; the hybrid method more reliably ensures convergence of the prediction error to 0; equations (1), (1a) or (1b) or other linear on non-linear regressions can be used to generate a model of the degradation curve CF, based on a buffer of data obtained by measurement during the period of operation of the filter arrangement; all measured data of pressure drop across the filter arrangement, collected during the operation period thereof, can be used for fitting the curve and estimate the trend thereof towards final life; in other embodiments; only a reduced frame of the collected data, e.g., those relating to the most recent period of use of the filter arrangement, can be used in the calculation; the time interval used can e.g. be approximately 1/10 of the expected total useful life of the filter; measurements of the last 1000 hours or so can be used in the regression; a hybrid prognostic or estimation method can be designed, wherein the physics-based prediction becomes gradually predominant over the data-driven prediction, as the final life of the filter arrangement is approached; to ensure a smooth transition between the data-driven method and the physics-based method, a combination of the two methods can be used, based on the following formula: RUL(Ti) = f·RULSB(Ti) + (1-f) ·RULPB(Ti) (5) wherein
f
=
m
e
d
i
a
n
0
;
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
;
1
, and wherein RULSB is the residual useful life estimated on the basis of the pure data-driven, similarity-based approach [i.e., equations (2), (3), (4)]; and RULPR is the residual useful life estimated on the basis of the regression only, i.e., on the pure physics-based approach, Δp_lim is a limit value of the pressure drop across the filter, which is near the threshold value Δp_alarm that determines the end of the useful filter life, see FIGS . 5 and 6 ; and x is the pressure drop value, i.e., the degradation parameter, measured at a given time instant T; the two estimated values are combined using the parameter (f), which takes the median (intermediate) value between 0,
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
, and 1; FIG. 7 shows the value of parameter (f) as a function of x; the degradation parameter, i.e., the pressure drop or pressure differential across the filter arrangement is constantly or intermittently measured by the respective differential pressure measuring arrangement (41A , 41B , 41C or 41D); as far as the measured degradation parameter x is below Δp_lim, the parameter ( f ) takes the value "1", since this is the median value between 0,
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
, and 1; as soon as the measured degradation parameter x reaches and then becomes greater than Δp_lim, the parameter (f) becomes
∆
p
_
a
l
a
r
m
∆
p
_
l
i
m
; the value of (f) gradually increases while the degradation parameter x increases from Δp_lim, towards Δp_alarm; consequently, the weight of the physics-based estimation of the residual useful life RULPR of the filter arrangement becomes gradually predominant over the data driven, i.e., similarity- based estimation RULSE; by combining the two methods in the hybrid method summarized in equation (6) the prediction error converges to 0, this resulting in an extremely accurate prognostic method for determining the residual useful life of the filter arrangement; Claim 4 and ¶¶ [0012], [0015], and [0063]: calculating the residual useful life of the filter arrangement as a weighted sum of the residual useful life values of the reference degradation curves, the sum being weighted by the similarity measures).
Claim 18
ALLEGORICO in view of Knox, Adams, and Lennartz discloses all the elements as stated in Claim 16 and further discloses wherein determining the remaining useful life of the filter includes determining a time range of remaining useful life of the filter (ALLEGORICO, ¶¶ [0062]-[0063]: a similarity weight measure is calculated using a Gaussian kernel, wherein h is the bandwidth of the kernel, which can range between 0.1 and 5 or between 0.2 and 3; the final residual useful life of the filter arrangement at a time instant Ti is then obtained by a similarity weighted sum based on residual useful life values of the set of used reference degradation curves) (Knox, ¶¶ [0015], [0017]-[0018], and [0091]-[0094]: indicating a first level filter warning when the estimated Smoke particle transmittance is less than or equal to the first threshold value; indicating a second level filter warning when the estimated Smoke particle transmittance is less than or equal to the second threshold value; produce or flag a filter warning or fault condition when the estimated transmittance reaches a threshold at which a predetermined reduction of the transmittance of the filter may be deemed to indicate an unacceptable degradation in filter performance).
Claim 19
ALLEGORICO in view of Knox, Adams, and Lennartz discloses all the elements as stated in Claim 16 and further discloses logging, by the computing device, additional operational data; appending, by the computing device, the additional operational data to the initial data set to generate an appended data set; refitting, by the computing device, the machine learning model to the appended data set; and determining, by the computing device, a revised remaining useful life of the filter (ALLEGORICO, ¶ [0067] with FIG. 2: a central control unit 43 can be suitably programmed for performing the above described prognostic method; measured values of the degradation parameter can be stored in a storage unit 45, together with data defining the reference degradation curves; a generic interface 47, e.g. a monitor, which provides the operator with information on the predicted residual useful life of the various filter arrangements) (ALLEGORICO, ¶ [0036]: the accumulated particulate matter obstructs the passage of air through the filter arrangement, thus increasing the pressure loss, i.e., the pressure differential across the filter arrangement; the pressure loss, i.e., the pressure drop across the filter arrangement can thus be used as a degradation parameter, which provides information on the degradation of the filter arrangement; ¶¶ [0039]-[0041] with FIG. 2: each filter arrangement 25 , 29 , 31 , 33 is provided with its own residual useful life prognostic system; each prognostic system can comprise a differential pressure measuring arrangement, configured for measuring a pressure differential across the respective filter arrangement; in FIG . 2, a plurality of filter degradation measuring systems 41A , 41B , 41C , 41D are schematically shown, one for each filter arrangement 25 , 29 , 31 , 33; each filter degradation measuring systems can be comprised of a differential pressure measuring arrangement; differential pressure measuring arrangements can be configured to measure the total differential pressure across two or more sequentially arranged filter arrangements; reference profiles of filter degradation are defined as set of data representing a filter degradation parameter as a function of time, which are predetermined "a priori" with respect to the actual measurement of the degradation parameter; each reference profile of filter degradation or degradation curve can therefore be represented as a curve of pressure loss across the filter arrangement versus time; ¶ [0043] with FIG. 3: represent the experimental data measured on seven different arrangements, e.g. operating in different environmental conditions and/or in different operating conditions; ¶ [0011]: the reference degradation curves are predetermined and can be obtained by data on existing filter arrangements; reference degradation curves can be generated starting from a limited number of experimentally obtained curves; in particular, the predetermined reference curves can be based on historical data of degradation of the filter arrangement; ¶ [0056] with FIGS. 2 and 5: the pressure drop Δp (i.e., the degradation parameter) across the filter arrangement has been measured by the respective pressure measuring arrangement, e.g., 41A , 41B , 41C or 41D (FIG. 2), between the start of the filter operation (T = 0) and the actual time Ti; ¶¶ [0064]-[0068]: use matrices of data where the pressure drop values for the reference degradation curves Cj are stored; each reference degradation curve Cj can thus be defined by a set of coordinates (Ti, Xj,i); at each time instant Ti the degradation value detected by the differential pressure measuring arrangement is stored together with the previously detected values and the buffer of data (xi-n, xi) are used, together with the stored data defining the reference degradation curves Cj, in equations (2), (4) and (5) to calculate the estimated residual useful life of the filter arrangement; measured values of the degradation parameter can be stored in a storage unit 45, together with data defining the reference degradation curves; a global pressure drop across the entire set of filter arrangements or a sub–group thereof can be measured) (ALLEGORICO, ¶¶ [0044]-[0053] with FIGS. 3-4: the larger the number of available reference degradation curves, the more accurate the prediction of the residual useful life will be; if an insufficient number of reference degradation curves is available, or if a larger number of such reference degradation curves is desirable, artificial reference degradation curves can be generated, starting from a relatively small number of experimental curves; generating artificial reference degradation curves can start from observing that each reference degradation curve can be fitted e.g. with a mathematical model, i.e., a mathematical curve fitting the actual reference degradation curve can be defined; an exponential model can be used, which can be defined as follows: Δp = a + b·exp(cT) (1), where a, b and c are fitting coefficients and T is time; to generate artificial reference degradation curves starting from the seven experimental reference degradation curves plotted in FIG. 3, coefficients (a), (b) and (c) can be manipulated and combined to obtain other sets of coefficients which, once introduced in formula (1) fit artificially generated reference degradation curves; a further mathematical models can be based e.g. on the following equation : Δp = a + bT + cT2 (1a); another mathematical setting model can be based on two equations:
∆
p
=
a
+
b
T
,
T
≤
T
1
a
+
b
∙
e
-
c
T
,
T
>
T
1
(1b); ¶ [0074]: all measured data of pressure drop across the filter arrangement, collected during the operation period thereof, can be used for fitting the curve and estimate the trend thereof towards final life) (ALLEGORICO, ¶¶ [0037]-[0039]: a filter arrangement requires to be changed when the degradation parameter achieves a threshold value, i.e., if the pressure differential, i.e., the pressure loss across the filter arrangement reaches a threshold value; each filter arrangement has a residual useful life (hereunder also indicated as RUL), which can be expressed in operating hours available before the degradation parameter reaches the threshold value; the residual useful life of the filter arrangement can thus be defined in general terms as the available operation time before the level of filter saturation will be such as to cause the pressure differential to reach the threshold value; ¶ [0041]: the residual useful life of a filter arrangement is estimated on the basis of a prognostic approach, which uses sets of predetermined experimentally obtained and/or artificially generated reference profiles of filter degradation, also named reference degradation curves; ¶¶ [0054]-[0063] with FIGS. 5-6: FIG . 5 shows four experimentally or artificially generated degradation curves C1, C2, C3, C4; a threshold value Δp_alarm of the pressure drop across the filter arrangement is reported on the vertical axis; upon reaching the threshold value Δp_alarm the filter arrangement requires to be changed; the intersection of the curves C1-C4 with the horizontal "ALARM" line corresponding to the threshold value defines the residual useful life of a filter which behaves according to the respective degradation curve; curve CF is the actual degradation curve of the filter arrangement, the degradation whereof shall be predicted; at time Ti the future trend of the degradation curve CF shall be predicted, in order to obtain an estimation of the residual useful life of the filter arrangement, which is the coordinate on the abscissa of the point of intersection between the curve CF and the "ALARM line"; the predictive method can be based on a data-driven similarity-based approach; at any instant in time (Ti) the shape of the curve CF is predicted based on the shapes of the reference degradation curves Cj (j = 1-4 in the example shown in FIG . 5); the similarity of the curve CF with the reference degradation curves Cj is calculated and used to estimate the shape of the curve CF from the time instant Ti to the end of its useful life; a first step of the predictive method can comprise the calculation of the Euclidean distance between curve CF and the reference degradation curve Cj in a time interval or time frame [Ti-n, Ti]; the calculation can be performed on a sequence of observations which includes n points, i.e., n measurements of the actual pressure loss across the filter arrangement; next, each Euclidean distance dj calculated with equation (2) can be transformed into a similarity measure using a Gaussian kernel; the final residual useful life of the filter arrangement at a time instant Ti is then obtained by a similarity weighted sum based on residual useful life values of the set of used reference degradation curves; ¶ [0068]-[0069]: a global pressure drop measured across the entire set of filter arrangements or a sub–group can be used to determine a global residual useful life of the entire group of filter arrangements or a sub-group thereof; the above described process is repeated sequentially runtime, such that at each instant in time an updated estimation of RUL (Ti) can be obtained; each time the calculation is performed, a different time frame can be used; the residual useful life of the filter arrangement is re-calculated each time on the basis of a different buffer of data , such that at each calculation the most significant portion of the reference degradation curves Cj is used) (ALLEGORICO, ¶ [0069]: the above described process is repeated sequentially runtime, such that at each instant in time an updated estimation of RUL (Ti) can be obtained).
Response to Arguments
Applicant’s arguments filed on 03/17/2026 with respect to claims 1, 12, and 16 have been fully considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Moffa (US 2017/0032661 A1, pub. date: 02/02/2017) discloses in ABSTRACT and ¶¶ [0008]-[0009] that a system for facilitating Smoke detector performance analysis including a server configured to receive operational data from an alarm panel and to perform analytics using the operational data, wherein the operational data is associated with at least one smoke detector that is operatively connected to the alarm panel. Moffa further discloses in ¶¶ [0019]-[0030] with FIG. 1 that (1) a fire safety system 100 may include one or more smoke detectors 1101-110a (wherein “a” can be any positive integer) operatively coupled to a centralized alarm panel 120; (2) each of the smoke detectors 1101-110a may be adapted to measure a level of ambient smoke or other particulate in a surrounding environment and to generate a digital output value representing such level (e.g., a conventional measurement of smoke density or obscuration level); (3) each of the smoke detectors 1101-110a may be associated with a "baseline average value" that may be a periodically or continuously updated average of the output values of a smoke detector over time; (4) the baseline average values of the smoke detectors 1101-110a may be calculated by a processor 127 of the alarm panel 120 and may be stored in a memory 128 of the alarm panel 120; (5) each of the smoke detectors 110-110 may additionally be associated with a predefined, operator-selectable "sensitivity value" that may be stored in the memory 128 of the alarm panel 120; (6) the sensitivity value for a smoke detector may define a number of counts (e.g., 60 counts) above the baseline average value that is determined to be indicative of an alarm; (7) the sum of the sensitivity value and the baseline average value for a smoke detector may yield an "alarm threshold value” for that smoke detector that may be calculated by the processor 127 of the alarm panel 120 and stored in the memory 128 of the alarm panel 120; (8) during normal operation of the system 100, the alarm panel 120 may initiate an alarm if one or more of the smoke detectors 110-110 generate an output value that is greater than its associated alarm threshold value; (9) the alarm panel 120 may communicate alarm conditions and other data relating to the status of the alarm panel 120 and the smoke detectors 1101-110a to one or more monitoring entities 124 via an alarm reporting network 122; (10) once the sum of the baseline average value 200 and the sensitivity value 204 exceeds the maxi mum output value 206 (i.e., 255 counts) of the smoke detector, the smoke detector will lose a portion of its effective operating range since an output value equal to the maximum output value 206 will always cause the alarm panel 120 to initiate an alarm; (11) identifying which Smoke detectors in a fire Safety system are actually dirty and are in need of cleaning as well as how well they were cleaned; and (12) the operational data is transmitted from the alarm panel 120 via the data communication device 129 may be entirely separate and independent from the alarm reporting network 122.
Culp et al. (US 2022/0148403 A1, filed on 11/01/2021) discloses in ABSTRACT that (1) a detection system including a smoke detector configured to transmit a detector baseline signal to at least one of a control panel and a server (e.g., to establish at least one indoor air quality trend), and a method for monitoring indoor air quality with at least one smoke detector are provided; (2) determine whether a current condition indicates a need to trigger an alarm; and (3) measure a baseline, wherein the detector baseline signal is used (e.g., by the control panel and/or the server) to establish at least one indoor air quality (IAQ) trend. Culp further discloses in ¶¶ [0007]-[0024] that (1) determine whether a current condition indicates a need to trigger an alarm; (2) at least one of a control panel and a server configured to receive and compile detector baseline signal to establish at least one indoor air quality (IAQ) trend; (3) the control panel is configured to receive detector baseline signals from multiple smoke detectors; (4) the server is communicatively connected to at least one control panel; (5) at least one of the control panel and the server are configured to trigger a notification when an IAQ trend meets a certain criteria; (6) a mobile device communicatively connected to at least one of the control panel and the server. Culp also discloses in ¶¶ [0033]-[0038] with FIG. 1 that (1) multiple smoke detectors 10 configured to transmit detector baseline signals to at least one of a control panel 50 and a server 31; (2) the detector baseline signal may be transmitted to the control panel 50 from the smoke detector 10 before the detector baseline signal is transmitted to the server 31 or to a mobile device 40; (3) the control panel 50 may be configured to receive detector baseline signals from multiple smoke detectors 10; and (4) each
control panel 50 may be connected to numerous hazard detectors (e.g. smoke detectors 10, etc.), notification devices (e.g. horns, strobes, annunciators, etc.), alarm triggers (e.g. pull stations, call points, door alarms, etc.), and other communicatively connected infrastructure.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to HWEI-MIN LU whose telephone number is (313)446-4913. The examiner can normally be reached Mon - Fri: 9:00 AM - 6:00 PM EST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Mariela D. Reyes can be reached at (571) 270-1006. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/HWEI-MIN LU/Primary Examiner, Art Unit 2142
Prosecution Insights