DETAILED ACTION
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on Apr. 10th, 2026 has been entered.
This action is in response to the amendments filed on March 13th, 2026. A summary of this action:
Claims 1,3-8, 10-14, 16-19, 22 have been presented for examination.
Claim(s) 1, 3-5, 8, 10-13, 16-18, 22 is/are rejected under 35 U.S.C. 103 as being unpatentable over Wang, Xiaofeng, Zhenjie Zhu, and Guoliang Lu. "Multiple regression analysis for change detection in multi-sensory monitoring data with application to induction motor speed condition monitoring." Measurement Science and Technology 31.9 (2020): 095103 In view of Sridhar, Prasanna, Asad M. Madni, and Mo Jamshidi. "Hierarchical data aggregation in spatially correlated distributed sensor networks." 2006 World Automation Congress. IEEE, 2006 in further view of Wang, Teng, Guoliang Lu, and Peng Yan (hereinafter Lu). "Multi-sensors based condition monitoring of rotary machines: An approach of multidimensional time-series analysis." Measurement 134 (2019): 326-335 and in further view of Song et al., US 2021/0341910.
Claim(s) 6-7, 14, and 19 is/are rejected under 35 U.S.C. 103 as being unpatentable Wang, Xiaofeng, Zhenjie Zhu, and Guoliang Lu. "Multiple regression analysis for change detection in multi-sensory monitoring data with application to induction motor speed condition monitoring." Measurement Science and Technology 31.9 (2020): 095103 In view of Sridhar, Prasanna, Asad M. Madni, and Mo Jamshidi. "Hierarchical data aggregation in spatially correlated distributed sensor networks." 2006 World Automation Congress. IEEE, 2006 in further view of Wang, Teng, Guoliang Lu, and Peng Yan (hereinafter Lu). "Multi-sensors based condition monitoring of rotary machines: An approach of multidimensional time-series analysis." Measurement 134 (2019): 326-335 and in further view of Song et al., US 2021/0341910 and in further view of Cormode, Graham, Srikanta Tirthapura, and Bojian Xu. "Time-decaying sketches for robust aggregation of sensor data." SIAM Journal on Computing 39.4 (2010): 1309-1339.
This action is non-final
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 .
Response to Arguments/Amendments
Regarding the § 112(a) Rejection
Withdrawn in view of remarks.
Regarding the § 101 Rejection
Withdrawn in view of the amendments and supporting remarks, as the last limitation integrates a practical application in view of example 45 claims 2 and 4.
Regarding the § 102/103 Rejection
Withdrawn in view of amendments, new grounds below as necessitated by amendment.
Remarks at 14 are conclusory and do not address the particular portions of Wang and Sridhar as were relied upon in combination nor do they address the particular rationale expressly stated in the rejection for how they are combined, i.e. it’s a piecemeal attack again the two references each taken alone merely for what they are alleged to be “directed to” without pointing out how the language of the claims are distinct from the particular citations relied upon and the rationale stated in the § 103 rejection for how the references would have been taken in combination by POSITA.
Remarks for the newly amended subject are moot, as a new grounds of rejection is presented below for the newly amended subjected matter.
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.
Claim(s) 1, 3-5, 8, 10-13, 16-18, 22 is/are rejected under 35 U.S.C. 103 as being unpatentable over Wang, Xiaofeng, Zhenjie Zhu, and Guoliang Lu. "Multiple regression analysis for change detection in multi-sensory monitoring data with application to induction motor speed condition monitoring." Measurement Science and Technology 31.9 (2020): 095103 In view of Sridhar, Prasanna, Asad M. Madni, and Mo Jamshidi. "Hierarchical data aggregation in spatially correlated distributed sensor networks." 2006 World Automation Congress. IEEE, 2006 in further view of Wang, Teng, Guoliang Lu, and Peng Yan (hereinafter Lu). "Multi-sensors based condition monitoring of rotary machines: An approach of multidimensional time-series analysis." Measurement 134 (2019): 326-335 and in further view of Song et al., US 2021/0341910.
Regarding Claim 1
Wang teaches:
A computer-implemented method, comprising: (Wang, abstract and § 6)
obtaining multi-dimensional time series data from sensors that collect the multi- dimensional time series data from a system during a time …(Wang, abstract, including: “…This paper presents a novel framework, based on multiple regression analysis (MRA), for the on-line monitoring of induction motors in order to detect changes in motor rotational speed during the course of successive operations. To utilize the synergistic information of multiple sensors, a prediction model is established based on MRA…” – to clarify, see fig. 1, as discussed in part in § 2.1.1: “Let us assume that X1:j is the collected multidimensional condition data up to inspection time j, i.e. X1:j = [x11 :j; x21 :j; · · · , xi 1:j] ′ , where i is the number of sensors; [·] ′ stands for the transposition operation. Each dimension, e.g. xi 1:j = [xi 1, xi 2, xi 3, · · · , xij ] where xij is the observation value of ith sensor at time j, may be expresed as a periodic form, considering the symmetry of physical structure of motor machinery [37]….” – and see equation 1 and its accompanying description including: “where xn,v contains observations of vth phase in n+1th cycle sensed by a measurement system consisting of the number of i sensors,”
creating matrices based on the multi-dimensional time series data; (Wang, abstract and § 2.1.1 as discussed above, and see the last paragraph in § 2.1.1: “Moreover, in order to fully utilize the advantage of LASSO, a sliding-window strategy was adopted for model training [42], thus…” – and see the equation, i.e. the “sliding window” creates matrices based on the data
to be specific, see the equation 6, then see equations 1-3, then see § 2.1.2 ¶ 1, then see fig. 2 and its accompanying description, i.e. in equation 1, there is a matrix representing the data from each of “i sensors” from time 0 to time “nT+v”, wherein the “sliding-window strategy” in § 2.1.1 creates matrices from this matrix of smaller windows of time, as visible depicted in fig. 2 and its accompanying description (include seeing table 1, e.g. “The window including first 5 cycles are adopted for model training;”
in addition, in § 2.1.2, eq. 7, there would have been matrices for each sensors data for the time periods, e.g. matrices created for “0” to “nT+v” to “c”, and additional matrices for “c” to “nT+v”
determining, using a first computer-based numerical modeling method, patterns based on the matrices; (Wang, as discussed above including the abstract and § 2.1.1-2.1.2, include seeing eq. 3 -4: “Meanwhile, in order to utilize the synergistic information existing in multi-sensory data, a novel prediction strategy is presented, as shown in figure 1, and the prediction model is established by the extension of the multivariate linear model, thus where the element number of each coefficient vector b is equal to the number of sensors, i.e. b = [b1, b2, · · · , bi]′ , but where their specific values are unequal for different dimensions… On the basis of several completed cycles, the coefficients of the different rows may be estimated by minimizing a costing function in the model training process, by means of the well - known least square regression (LRS) method:…”
to clarify, then see § 2.1.2 eq. 7: “By virtue of the established MRA model, an appropriate statistical metric should be adopted to quantify temporal anomalies for decision making [21]. When the motor condition [each motor condition being an example of a pattern] changes at time c, the generation mechanism of the data changes accordingly, which means the coefficients of trained models before and after c are distinct [the patterns were determined for the coefficient determination] ; i.e. {b10 , b20 , · · · , bi 0 } changes to {b11 , b21 , · · · , bi 1 }. Such a distribution change of data may be depicted through a piecewise regression model, given as“
In addition, for a second example of this feature, also see § 2.1.1 last paragraph: “Moreover, in order to fully utilize the advantage of LASSO, a sliding-window strategy was adopted for model training [42], thus …where w is the size of the window” – as taken in view of equations 1-3, i.e. the “coefficient vector” was determined for the “sliding window” during the training, see fig. 2 to clarify – specifically the “Slid window” to the “Model establishment” [incl. determination of the “coefficient vector”] as detailed in table 1: “Step 1. The window including first 5 cycles are adopted for model training…Step 3. The window including training cycles moves forward by one cycle and the model is updated accordingly…Step 1. The window moves forward by one cycle, and the model is updated using equations (3), (5), and (6), accordingly….”, i.e. in each window there is a “coefficient vector” which represents a determined pattern
training a machine learning model using a second computer-based numerical modeling method, the machine learning model being a time series model created based on the patterns; predicting a future condition of the system using the machine learning model with current data of the system, the future condition of the system represented by a quantified state of the system; determining whether the quantified state of the system satisfies a threshold value, (Wang, abstract: “This paper presents a novel framework, based on multiple regression analysis (MRA), for the on-line monitoring of induction motors in order to detect changes in motor rotational speed during the course of successive operations. To utilize the synergistic information of multiple sensors, a prediction model is established based on MRA. By virtue of this model, the residual between model output and sensor observation is defined as a dynamic stability indicator, for the purpose of characterizing the running status of a motor.” – then see § 2.1.1-2.1.2 as discussed above, including seeing fig. 1 for the “Model establishment” step wherein this provides “prediction[s]” for time series as visibly depicted, wherein this is based on the patterns as discussed in § 2.1.2 (the “piecewise regression model…which allows for a trend-type [pattern] change… Assuming that {b10 , b20 , · · · , bi 0 } has already been estimated by means of model training, the predictions of xn,v at the inspected n+1th cycle can be calculated and denoted byˆxn,v.”)
to clarify on the training, page 3, col. 2, last paragraph: “On the basis of several completed cycles, the coefficients of the different rows may be estimated by minimizing a costing function in the model training process,…” to page 3: “Moreover, in order to fully utilize the advantage of LASSO, a sliding-window strategy was adopted for model training” and § 2.1.2: “When the motor condition changes at time c, the generation mechanism of the data changes accordingly, which means the coefficients of trained models before and after c are distinct ;”
as to the quantifying, Wang § 2.1.2: “Next, an anomaly score Qn+1 is defined, which quantifies the extent of motor deviation from normal, considering the offset of residuals and the effect of cycle length, thus”, and § 2.2: “Let us assume that the motor’s condition was inspected as normal prior to the n+1th cycle. The anomaly scores of historical cycles {Q1, Q2, · · · , Qn} will follow independent and identical distribution, and will vary within an interval ranging from zero to pre-defined limit… Once a change occurs at n+1th cycle, i.e. nT +v > c, the anomaly score Qn+1 will increase and exceed the limit, which can be detected via hypothesis testing as… If H0 is satisfied, then change decision making will take place; otherwise, no change occurs.” – see fig. 2 to further clarify, for the “On-line monitoring” followed by the “Decision making” portions
While Wang does not explicitly teach the following feature, Wang in view of Sridhar teaches: wherein the sensors are arranged in groups of sensors defined according to their relative distances to respective ones of center points in the groups of sensors, and an individual sensor in an individual group of sensors is physically closer to a respective center point of the respective individual group of sensors where the individual sensor belongs than the respective center point of any other of the groups of sensors;… wherein each matrix of the matrices is an MxN matrix where M is a number of the groups of sensors and N is a number of dimensions of the multi-dimensional time series data, the number of dimensions being a number of different types of data collected by each one of the groups of sensors; (Wang, as discussed above including §§ 2.1.2-2.1.2 for the matrices, wherein each matrix is for a single sensor, with a set of “observations” for each sensor, and how this is used in a “piecewise regression model” in § 2.1.2 as was discussed above, also Wang, abstract: “Multi-sensory configuration enables the collection of comprehensive information relating to the operating condition of machinery in use” and § 1; see Wang table 5 and fig. 5-6 which also clarify that the number of sensors was 4, and each sensor collected a different type of information, i.e. a “Sound” sensor; a “Current Sensor”, and note in fig. 5(a) that the vibration sensors collected two different types of vibration data (as clarified in fig. 5(b), note the right-most red lines as these two indicate schematically that the first vibration sensor was collecting “Stepper motor” vibration; and the second vibration was collecting “Gearbox” vibration, i.e. N = 4 = 4 types of information
taken in further view of Sridhar, abstract: “The central idea of using sensor networks for monitoring events and conditions is to exploit the distributed nature provided by tiny and low powered devices. Multiple sensors can be used collaboratively to monitor events or space more effectively than a single sensor... These sensors in general are prone to failure due to their inherent characteristics. In this paper, we propose a robust fault tolerant data aggregation scheme in sensor networks.” – then see § 1.1 including: “Parallel fused data from multiple sensors can represent decision milestones which will incur less communication cost than serially processing raw data acquired by individual sensors. It is an intractable problem to actually detect if a sensor is faulty by looking at the raw data acquired from the sensors. However, because of faulty sensors, the fused data will deviate from the actual physical value being sensed. In order to reduce the impact of faulty information prior to fusing, we propose a novel approach of weighted average aggregation of data from these sensors.” – see §2.2 to further clarify: “Clustering of randomly deployed sensor nodes based on some metric (say, distance) has the advantage of dividing the problem space into several sub-problems and solving each sub-problem for estimation; a divide-and-conquer approach…. Consider three overlapping sensing regions. The region of interest is the aggregated data obtained around the region of the intersection of these sensing regions. For a large deployment scenario, these sensing regions can be extended to cluster regions… In hierarchical structure, sensor information is fused in each cluster to produce a local estimate which is then fused to obtain a global estimate of the sensed information. Several fusion steps are needed in each cluster, however, each of these local estimates can be done in parallel. Weighted adaptation can be easily managed resulting in more reliable information from each sensor/cluster heads…. Each sensor node has a weighting factor at any instance of time t, given by wi(t). In the event of sensor failure, the proposed… In order to estimate Δwi(t), we use the concept of spatial correlation…. The sensors deployed in large numbers are thus correlated spatially within the region of events, that is, the sensor I reads the same event value (with minimal variation) as the neighboring k sensors which are closely deployed…” and see equations 2-3 - then see “Theorem 1” and “Corollary 1” in § 2.2.1
with respect to the sensors all being closest in a group, see Sridhar, fig. 1 which shows such a configuration
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings from Wang on “Multi-sensory configuration enables the collection of comprehensive information relating to the operating condition of machinery in use. However, the complex properties of multi-sensory monitoring data create serious challenges for data modelling and analysis. This paper presents a novel framework, based on multiple regression analysis (MRA), for the on-line monitoring of induction motors in order to detect changes in motor rotational speed during the course of successive operations…” (Wang, abstract) with the teachings from Sridhar on “Multiple sensors can be used collaboratively to monitor events or space more effectively than a single sensor” and “In this paper, we propose a robust fault tolerant data aggregation scheme in sensor networks” (Sridhar, abstract). The motivation to combine would have been that “Multiple sensors can be used collaboratively to monitor events or space more effectively than a single sensor. Several applications can be envisioned with sensor networks ranging from military and commercial applications to environment and earth sciences. Typical examples include: traffic monitoring of vehicles, military reconnaissance and surveillance, target tracking, cross-border infiltration, habitat monitoring and structural monitoring, to name a few. These sensors in general are prone to failure due to their inherent characteristics. In this paper, we propose a robust fault tolerant data aggregation scheme in sensor networks” (Sridhar, abstract) Also, see § 1.1, ¶ 2 for an additional motivation to combine, and § 3: “Data aggregation in sensor networks in general helps to reduce communication cost. However, a faulty reading by a sensor can represent a false estimate for the aggregated data. In this paper, we propose a robust mechanism to aggregate data from different sensors with some tolerance to faults”
While Wang, as taken in combination above, does not explicitly teach the following feature, Wang, as taken in combination above and in further view of
and in response to the quantified state of the system satisfying the threshold value, adjusting at least one system control, wherein the at least one system control controls an operating parameter of the system. (Wang, as cited above, § 2.2: “Let us assume that the motor’s condition was inspected as normal prior to the n+1th cycle. The anomaly scores of historical cycles {Q1, Q2, · · · , Qn} will follow independent and identical distribution, and will vary within an interval ranging from zero to pre-defined limit… Once a change occurs at n+1th cycle, i.e. nT +v > c, the anomaly score Qn+1 will increase and exceed the limit, which can be detected via hypothesis testing as… If H0 is satisfied, then change decision making will take place; otherwise, no change occurs.” – see fig. 2 to further clarify, for the “On-line monitoring” followed by the “Decision making” portions – also, see Wang, § 1 ¶ 1 including: “…With these applications, one of the major goals is to detect changes (e.g. faults, anomalies, and switching/transit points) in the running status of dynamic motors at an early stage, based on sensor signals [4, 5]…The capacity to detect such changes enables abnormal running behavior to be observed and highlighted so as to help users to formulate corrective schedules and/or carry out predictive maintenance [7]. Change detection is also desirable in advanced applications where appropriate actions or adaptive adjustments need to be carried out as soon as possible once a change alarm is received.”
As taken in view of Song, ¶¶ 27-28: “Having determined that the cyber - physical system is in an anomalous state , block 216 determines which sensors 104 have contributed to that determination…Block 218 then performs a responsive action... The responsive action can include diagnostics designed to acquire more information regarding the anomaly from the sensors 104. The responsive action can include sending an instruction to one or more sub - systems of the monitored system 102 , to bring the sensor readings back to a “ normal ” state . Responsive actions may also include changing a setting or state of devices associated with the respective sensors 104. As noted above, automatic responsive actions may include changing a security setting for an application or hardware component , changing an operational parameter of an application or hardware component ( for example , an operating speed ) , halting and / or restarting an application , halting and / or rebooting a hardware component , changing an environmental condition , changing a network interface's status or settings , etc.” – also, see ¶ 53 as well
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings from Wang, as modified above, on a system for change detection such as to detect anomalies/faults (Wang, as cited above) which suggests performing “adaptive adjustments” in response to a change being detected (Wang, § 1 ¶ 1) with the teachings from Song on particular responsive actions performed in response to a detected anomaly. The motivation to combine would have been that Song’s responsive actions would have automatically resolved the change alarm of Wang, thereby improving the efficiency of the system by reducing the actions required by the user in order to respond to the change alarm (Wang, § 1 ¶ 1 which suggests user involvement to respond of the change alarm and Wang “Decision Making” algorithm on page 5, which teaches it “will output an alarm to the user”).
Regarding Claim 3
Wang, in view of Sridhar, as discussed above, teaches: The method of claim 1, wherein each value in the MxN matrix is a weighted average of values of plural respective ones of the sensors in the respective individual group of sensors. (See Wang as was taken in view of Sridhar as discussed above for claim 1 incl. in Sridhar the abstract, and §§ 1.1, 2.2, and 2.2.1 as were cited above)
Regarding Claim 4
Wang, in view of Sridhar, as discussed above, teaches:
The method of claim 3, wherein respective weights of the plural respective ones of the sensors in the respective individual group of sensors are based on a closest distance to the respective center point of the respective individual group of sensors. (Wang, as discussed above including §§ 2.1.2-2.1.2 for the matrices, wherein each matrix is for a single sensor, with a set of “observations” for each sensor, and how this is used in a “piecewise regression model” in § 2.1.2 as was discussed above, also Wang, abstract: “Multi-sensory configuration enables the collection of comprehensive information relating to the operating condition of machinery in use” and § 1
as taken in view of Sridhar, abstract, § 1.1, and § 2.2 as discussed above, in particular see § 2.2: “Clustering of randomly deployed sensor nodes based on some metric (say, distance) has the advantage of dividing the problem space into several sub-problems and solving each sub-problem for estimation; a divide-and-conquer approach…. Consider three overlapping sensing regions. The region of interest is the aggregated data obtained around the region of the intersection of these sensing regions. For a large deployment scenario, these sensing regions can be extended to cluster regions… In hierarchical structure, sensor information is fused in each cluster to produce a local estimate which is then fused to obtain a global estimate of the sensed information. Several fusion steps are needed in each cluster, however, each of these local estimates can be done in parallel. Weighted adaptation can be easily managed resulting in more reliable information from each sensor/cluster heads…. Each sensor node has a weighting factor at any instance of time t, given by wi(t). In the event of sensor failure, the proposed… In order to estimate Δwi(t), we use the concept of spatial correlation…. The sensors deployed in large numbers are thus correlated spatially within the region of events, that is, the sensor I reads the same event value (with minimal variation) as the neighboring k sensors which are closely deployed…” and see equations 2-3 - then see “Theorem 1” and “Corollary 1” in § 2.2.1 – in other words, the weights are determined based on the sensors in the group being “correlated spatially” in the cluster (i.e. that the neighboring sensors are “closely deployed”, see fig. 1 to further clarify which visually depicts a center point of the cluster of the sensors)
The rationale to combine is the same as discussed above for claim 1
Regarding Claim 5
Wang teaches:
The method of claim 1, wherein the determining the patterns comprises: defining a number of windows each representing a respective period of the time; and determining a respective vector of coefficients for each one of the windows, wherein the vector of coefficients for a particular one of the windows represents a pattern between a condition of the system measured during the respective period of the time and the multidimensional-time series data collected during the respective period of the time. (Wang, abstract and §§ 2.1.1-2.1.2 as discussed above, including the “piecewise regression model” in § 2.1.2 – in particular note the determined vector of coefficients: “When the motor condition changes at time c, the generation mechanism of the data changes accordingly, which means the coefficients of trained models before and after c are distinct ; i.e. {b10 , b20 , · · · , bi 0 } changes to {b11 , b21 , · · · , bi 1 }.” (see § 2.1.1: “each coefficient vector b” in the description of eq. 3) – wherein eq. 7 provides two windows of time, see the “0 ≤ nT+v < c,” and “c ≤ nT+v”, i.e. this is a “piecewise regression model”
In addition, also see § 2.1.1 last paragraph: “Moreover, in order to fully utilize the advantage of LASSO, a sliding-window strategy was adopted for model training [42], thus …where w is the size of the window” – as taken in view of equations 1-3, i.e. the “coefficient vector” was determined for the “sliding window” during the training, see fig. 2 to clarify – specifically the “Slid window” to the “Model establishment” [incl. determination of the “coefficient vector”] as detailed in table 1: “Step 1. The window including first 5 cycles are adopted for model training…Step 3. The window including training cycles moves forward by one cycle and the model is updated accordingly…Step 1. The window moves forward by one cycle, and the model is updated using equations (3), (5), and (6), accordingly….”, i.e. in each window there is a “coefficient vector” for use in establishing the model
Regarding Claim 8
Wang teaches: The method of claim 1, wherein the first computer-based numerical modeling method is different than the second computer-based numerical modeling method. (Wang, § 2.1.1: “On the basis of several completed cycles, the coefficients of the different rows may be estimated by minimizing a costing function in the model training process, by means of the well -
known least square regression (LRS) method:” – which is an example of a first numerical modeling method used, then see § 2.1.2 which shows the created model is a “piecewise regression model” (example of a second modeling method)
to clarify on the BRI, ¶ 61: “The modeling module 220 may be programmed to use a least square method to solve for the Beta vector, although embodiments are not limited to a least square method.” And ¶ 63: “In embodiments, the modeling module 220 is programmed to use a second computer-based numerical modeling method to form a linear regression for the patterns and sensor data… For example, after determining plural Beta vectors Bl, B2, B3, .. , BN in the manner described herein, the modeling module 220 then uses those plural Beta vectors with an ARMA model to create a time series model that predicts a future Beta vector B(N+ 1).”
Regarding Claim 10.
Wang teaches: The method of claim 1, wherein the predicting comprises: predicting a future pattern using the machine learning model; and predicting a future target value of the system using the future pattern. (Wang, see § 2.1.1 last paragraph: “Moreover, in order to fully utilize the advantage of LASSO, a sliding-window strategy was adopted for model training [42], thus …where w is the size of the window” – as taken in view of equations 1-3, i.e. the “coefficient vector” was determined for the “sliding window” during the training, see fig. 2 to clarify – specifically the “Slid window” to the “Model establishment” [incl. determination of the “coefficient vector”] as detailed in table 1: “Step 1. The window including first 5 cycles are adopted for model training…Step 3. The window including training cycles moves forward by one cycle and the model is updated accordingly…Step 1. The window moves forward by one cycle, and the model is updated using equations (3), (5), and (6), accordingly….”, i.e. the prediction is predicting the future pattern (the future window) and predicting a target value in the future pattern)
Regarding Claim 11.
This is rejected under a similar rationale as claim 1 above, wherein Wang teaches: A computer program product comprising one or more computer readable storage media having program instructions collectively stored on the one or more computer readable storage media, the program instructions executable to: (Wang, abstract and § 6)
obtain multi-dimensional time series data from sensors that collect the multi-dimensional time series from a system during a time corresponding to a user-defined cycle…(Wang, abstract, including: “…This paper presents a novel framework, based on multiple regression analysis (MRA), for the on-line monitoring of induction motors in order to detect changes in motor rotational speed during the course of successive operations. To utilize the synergistic information of multiple sensors, a prediction model is established based on MRA…” – to clarify, see fig. 1, as discussed in part in § 2.1.1: “Let us assume that X1:j is the collected multidimensional condition data up to inspection time j, i.e. X1:j = [x11 :j; x21 :j; · · · , xi 1:j] ′ , where i is the number of sensors; [·] ′ stands for the transposition operation. Each dimension, e.g. xi 1:j = [xi 1, xi 2, xi 3, · · · , xij ] where xij is the observation value of ith sensor at time j, may be expresed as a periodic form, considering the symmetry of physical structure of motor machinery [37]….” – and see equation 1 and its accompanying description including: “where xn,v contains observations of vth phase in n+1th cycle sensed by a measurement system consisting of the number of i sensors,” – wherein POSITA would have inferred that the cycle of Wang was used defined by Wang as input parameter, or at least would have found it obvious to have the cycle be user defined because this would have been “making adjustable” the cycle of Wang (MPEP § 2144.04(V)(D)) – to clarify, the claim recites no particular method of how a user is to define what the cycle is, but only that it is user defined
Regarding Claim 12.
This is rejected under a similar rationale as claims 1 and 3-4 as discussed above. With respect to the recitation of “user-defined”, this is rejected under a similar rationale as the similar recitation in claim 11, i.e. it would have been inferred that the “fusion node” of Sridhar (e.g. fig. 1) was user defined; or at least it would have obvious to have made this user-defined as this would be “making adjustable” the center point (MPEP § 2144.04(V)(D)).
Regarding Claim 13.
This is rejected under a similar rationale as claim 5 above.
Regarding Claim 16.
This is rejected under a similar rationale as claim 1 above, wherein Wang teaches: A system comprising: a processor, a computer readable memory, one or more computer readable storage media, and program instructions collectively stored on the one or more computer readable storage media, the program instructions executable to: (Wang, abstract and § 6)
Regarding Claim 17.
This is rejected under a similar rationale as claims 1 and 3-4 as discussed above. With respect to the center point corresponding to a device in an environment, this would have been taught by Wang, in view of Sridhar - see Wang, fig. 5-6 and table 5 show where the sensors are and what they are, then see Sridhar, abstract: “Multiple sensors can be used collaboratively to monitor events or space more effectively than a single sensor.” And Sridhar fig. 1 – i.e. in combination, the multiple sensors would have been positioned to monitor the same space as the single sensor, and that space would correspond to the device to be monitored in Wang by the single sensor (e.g. Wang, fig. 5). The rationale to combine is the same as discussed above
Regarding Claim 18.
This is rejected under a similar rationale as claim 5 above.
Regarding Claim 22.
Wang in view of Song teaches:
The method of claim 1, wherein the threshold value represents a value below which the at least one system control of the system is adjusted to raise the quantified state of the system above the threshold value. (Wang § 2.1.2: “Next, an anomaly score Qn+1 is defined, which quantifies the extent of motor deviation from normal, considering the offset of residuals and the effect of cycle length, thus”, and § 2.2: “Let us assume that the motor’s condition was inspected as normal prior to the n+1th cycle. The anomaly scores of historical cycles {Q1, Q2, · · · , Qn} will follow independent and identical distribution, and will vary within an interval ranging from zero to pre-defined limit… Once a change occurs at n+1th cycle, i.e. nT +v > c, the anomaly score Qn+1 will increase and exceed the limit, which can be detected via hypothesis testing as… If H0 is satisfied, then change decision making will take place; otherwise, no change occurs.” – see fig. 2 to further clarify, for the “On-line monitoring” followed by the “Decision making” portions
Then see § 1 ¶ 1: “…The capacity to detect such changes enables abnormal running behavior to be observed and highlighted so as to help users to formulate corrective schedules and/or carry out predictive maintenance [7]. Change detection is also desirable in advanced applications where appropriate actions or adaptive adjustments need to be carried out as soon as possible once a change alarm is received” – thus Wang renders this obvious, because § 1 ¶ 1 clarifies once the alarm is receiving “appropriate actions or adaptive adjustments need to be carried out”, wherein POSITA would have been motivated to do so because “Change detection is also desirable in advanced applications”
As was taken in view of Song as cited above for claim 1.
Claim(s) 6-7, 14, and 19 is/are rejected under 35 U.S.C. 103 as being unpatentable Wang, Xiaofeng, Zhenjie Zhu, and Guoliang Lu. "Multiple regression analysis for change detection in multi-sensory monitoring data with application to induction motor speed condition monitoring." Measurement Science and Technology 31.9 (2020): 095103 In view of Sridhar, Prasanna, Asad M. Madni, and Mo Jamshidi. "Hierarchical data aggregation in spatially correlated distributed sensor networks." 2006 World Automation Congress. IEEE, 2006 in further view of Wang, Teng, Guoliang Lu, and Peng Yan (hereinafter Lu). "Multi-sensors based condition monitoring of rotary machines: An approach of multidimensional time-series analysis." Measurement 134 (2019): 326-335 and in further view of Song et al., US 2021/0341910 and in further view of Cormode, Graham, Srikanta Tirthapura, and Bojian Xu. "Time-decaying sketches for robust aggregation of sensor data." SIAM Journal on Computing 39.4 (2010): 1309-1339.
Regarding Claim 6
While Wang, in view of Sridhar and Song, does not explicitly teach the following feature, Wang in view of Sridhar, Song and Cormode teaches: The method of claim 1, wherein the first computer-based numerical modeling method utilizes an algorithm that includes a first factor based on attenuation of the data over the time. (Wang, as was cited above for §§ 2.1.1-2.1.2 including the use of a “piecewise regression model”, and “a sliding-window strategy”
As taken in further view of Cormode, abstract: “The sketch has the following properties which make it useful in communication-efficient aggregation in distributed streaming scenarios, such as sensor networks:… is also time decaying, so that the weight of a data item in the sketch can decrease with time according to a user-specified decay function.”
To clarify, page 1310, ¶ 2: “Lastly, we observe that in any evolving setting, recent data are more reliable than older data. We should therefore weight newer observations more heavily than older ones. This can be formalized in a variety of ways: we may only consider observations that fall within a sliding window of recent time (say, the last hour) and ignore (assign zero weight to) any that are older, or, more generally, use an arbitrary function that assigns a weight to each observation as a function of its initial weight and its age [18, 14]. A data summary should allow such decay functions to be applied and give us guarantees relative to the exact answer”
Then see page 1311, definition 1.1, in particular: “The decayed weight of an element (v,w, t, id) at time c ≥ t is f(w, c − t). An example decay function is the sliding window model [18, 22, 34], where f(w, x) is defined as follows. For some window size W, if x ≤ W, then f(w, x) = w; otherwise, f(w, x) = 0. Other popular decay functions include exponential decay f(w, x) =
w · exp(−ax) and polynomial decay f(w, x) = w · (x + 1)−a, where a is a constant.” – i.e. the “w” is an example of an attenuation factor, and the “a” defines the speed of the attenuation (note its location in the exponential decay in particular))
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings from Wang on “Multi-sensory configuration enables the collection of comprehensive information relating to the operating condition of machinery in use. However, the complex properties of multi-sensory monitoring data create serious challenges for data modelling and analysis. This paper presents a novel framework, based on multiple regression analysis (MRA), for the on-line monitoring of induction motors in order to detect changes in motor rotational speed during the course of successive operations…” (Wang, abstract) with the teachings from Cormode, abstract: “The sketch has the following properties which make it useful in communication-efficient aggregation in distributed streaming scenarios, such as sensor networks:… is also time decaying, so that the weight of a data item in the sketch can decrease with time according to a user-specified decay function.”
The motivation to combine would have been that as per page 1310, ¶ 2: “Lastly, we observe that in any evolving setting, recent data are more reliable than older data. We should therefore weight newer observations more heavily than older ones. This can be formalized in a variety of ways: we may only consider observations that fall within a sliding window of recent time (say, the last hour) and ignore (assign zero weight to) any that are older, or, more generally, use an arbitrary function that assigns a weight to each observation as a function of its initial weight and its age [18, 14]. A data summary should allow such decay functions to be applied and give us guarantees relative to the exact answer”
Regarding Claim 7
Wang in view of Sridhar and Cormode teaches: The method of claim 6, wherein the algorithm includes a second factor that defines a speed of the attenuation. (Wang, as was cited above for §§ 2.1.1-2.1.2 including the use of a “piecewise regression model”, and “a sliding-window strategy”
As taken in further view of Cormode, abstract: “The sketch has the following properties which make it useful in communication-efficient aggregation in distributed streaming scenarios, such as sensor networks:… is also time decaying, so that the weight of a data item in the sketch can decrease with time according to a user-specified decay function.”
To clarify, page 1310, ¶ 2: “Lastly, we observe that in any evolving setting, recent data are more reliable than older data. We should therefore weight newer observations more heavily than older ones. This can be formalized in a variety of ways: we may only consider observations that fall within a sliding window of recent time (say, the last hour) and ignore (assign zero weight to) any that are older, or, more generally, use an arbitrary function that assigns a weight to each observation as a function of its initial weight and its age [18, 14]. A data summary should allow such decay functions to be applied and give us guarantees relative to the exact answer”
Then see page 1311, definition 1.1, in particular: “The decayed weight of an element (v,w, t, id) at time c ≥ t is f(w, c − t). An example decay function is the sliding window model [18, 22, 34], where f(w, x) is defined as follows. For some window size W, if x ≤ W, then f(w, x) = w; otherwise, f(w, x) = 0. Other popular decay functions include exponential decay f(w, x) =
w · exp(−ax) and polynomial decay f(w, x) = w · (x + 1)−a, where a is a constant.” – i.e. the “w” is an example of an attenuation factor, and the “a” defines the speed of the attenuation (note its location in the exponential decay in particular))
The rationale is the same as discussed above for claim 6
Regarding Claim 14.
This claim is rejected under a similar rationale as claims 6-7 as discussed above.
Regarding Claim 19.
This claim is rejected under a similar rationale as claims 6-7 as discussed above.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Fradkin, US 2020/0257608. Abstract and ¶¶ 1, 3, 5-6, 14, 23-25, 33
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DAVID A. HOPKINS whose telephone number is (571)272-0537. The examiner can normally be reached Monday to Friday, 10AM to 7 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, Ryan Pitaro can be reached at (571) 272-4071. 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.
/David A Hopkins/Primary Examiner, Art Unit 2188
Prosecution Insights