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 .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 08/19/2026 are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Response to Amendment
The amendments filed on 05/20/2026 have been considered. Claims 1, 7, and 17 have been amended. Thus, claims 1-17 are pending and presented for examination.
Applicant's arguments filled on 05/20/2026 with respect to the 35 U.S.C. 101 rejections have
been fully considered and are persuasive. Thus, the 35 U.S.C. 101 rejection is withdrawn.
Applicant's arguments filled on 05/20/2026 with respect to the 35 U.S.C. 112(b) rejection have
been fully considered and are persuasive. Thus, the 35 U.S.C. 112(b) rejection is withdrawn.
Applicant's amendments filled on 05/20/2026 with respect to the claim objections have been
fully considered and are persuasive. Thus, the claim objections have been withdrawn.
Applicant's arguments filled on 05/20/2026 with respect to the 35 U.S.C. 103 rejections have
been fully considered but are moot because of the new ground of rejection. With the following argument being the exception. The response to this argument is presented below.
Response to Arguments
Applicant argues (Remarks, page 16): “Therefore, Streichert operates within a framework of ongoing monitoring and deviation analysis during normal operation, and does not teach or suggest identifying shutdown or reduced-demand operating states, nor applying machine-learning analysis specifically during such states to detect leakage.”
Examiner’s response: It appears the applicant is arguing that Streichert alone does not identify shutdown or reduced-demand operating states and does not apply its machine learning analysis during such states. The examiner agrees, but notes that Streichert was never relied upon for these features. Streichert is relied upon for detecting compressed air leaks by applying a trained ML model to measured pressure and flow rate data. In the current rejection, Koshinaka is relied upon for receiving the indication of the shutdown (deactivated) state from the supervisory control panel and for measuring the flow rate data during that state, when client demand is stopped and leakage flow becomes detectable. One cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claim 1-17 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention.
As per claims 1 and 17, these claims contain the limitation "each one of the plurality of shutdown events associated with an annotation indicative of whether the shutdown event was associated with at least one compressed air leak or no compressed air leak, wherein each annotation is automatically assigned to the respective annotated training samples by indications received from the supervisory controller." However, the specification at no time supports assigning annotations to training samples based on indications received from the supervisory controller. The indications received from the control units of the compressed air system (for example, SCADA) are described throughout the specification only as indications that one or more compressed air clients is in shutdown mode, which the leak detector uses to identify a shutdown event (see instant specification, paragraph 0086; see also paragraphs 0018 and 0046). These indications convey the shutdown status of the compressed air clients, and are never described as indicating whether a shutdown event was associated with a compressed air leak, or as being used to annotate training samples in any way. Automatic labeling is discussed in the specification: paragraph 0101 describes training samples "which are labeled automatically, as known in the art, based on statistics, weak low level classifiers, partial labeling, and/or the like," and paragraph 0128 describes shutdown or non-shutdown labels assigned automatically, again based on statistics, weak low level classifiers, partial labeling and the like. Neither paragraph makes any mention of labels or annotations assigned by indications received from a control unit or supervisory controller. The remaining discussions of annotating or labeling training samples are found in other paragraphs. However, none of these paragraphs describe the supervisory controller as the source of any annotation, or the supervisory controller indicating the presence or absence of a leak at all, let alone automatically assigning leak or no-leak annotations to training samples by indications received from the supervisory controller. This causes the claim to be new matter and therefore rejected under U.S.C. 112(a). In the next response, please indicate where support for these limitations are found within the specification.
As per claims 2-16, these claims are rejected as being dependent on a claim rejected under U.S.C. 112(a) for new matter.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claim 1, 17 rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 1, 17 recites the limitation "the ML model…". There is insufficient antecedent basis for this limitation in the claim. Suggested fix: “the at least one trained ML model”.
Claim 1, 17 recites the limitation "the plurality of shutdown events…". There is insufficient antecedent basis for this limitation in the claim. Suggested fix: “the plurality of previous shutdown events”.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Examiner’s Note: Some rejections will include an Examiner’s Note (labeled ‘EN’) to provide additional context or rationale explaining the basis for the rejection.
Claims 1-3, 5-8, 11, 13, and 16-17 are rejected under 35 U.S.C. 103 as being unpatentable over patent application US 2023/0088241 A1 Streichert et al., hereinafter “Streichert” in view of patent application US 2003/0187595 A1 Koshinaka et al., hereinafter “Koshinaka” further in view of patent application US 2020/0201950 A1 Wang et al., hereinafter “Wang”
Claim 1
Streichert teaches:
A computer implemented method of detecting leaks in a compressed air system using machine learning (ML), comprising: (Abstract, “Continuous condition monitoring of a pneumatic system, and in particular for early fault detection, is provided. The condition monitoring unit is formed with an interface to a memory in which a trained normal condition model is stored as a one-class model, which has been trained in a training phase with normal condition data and represents a normal condition of the pneumatic system.” – Examiner’s Note (EN): this denotes a pneumatic system that is equivalent to the broadest reasonable interpretation of “compressed air system” because the pneumatic system comprises actuators that function as the required “compressed air clients” consuming air delivered via valves and supply lines.)
“receiving, by the processor, pressure data and flow rate data measured by at least one pressure sensor and at least one flow sensor respectively deployed in the compressed air system…” (Para 5, “The method comprises the following method steps… Continuous acquisition of sensor data of the pneumatic system by means of a set of sensors” Para 32, “For example, one or more of the following sensors may be used: Flow meter, Pressure sensor…” Para 107, “In a step 204, sensor data of the pneumatic system 100 is continuously acquired by means of a set of sensors. The set of sensors includes at least the sensors m1 and m2 described above. In addition, several of these or other types of sensors (for example, flow sensors, pressure sensors, microphones, structure-borne sound pickups) may collect sensor data.” Para 101, “The monitoring unit 114 receives sensor data via the stroke 112 in the training and operating state of the pneumatic system 100” – EN: the condition monitoring unit 114 corresponds to the “processor” which receives the pressure data and flow rate data measured by the flow and pressure sensors deployed in the pneumatic (compressed air) system.)
“detecting, by the processor, at least one compressed air leak in the compressed air system using at least one trained ML model applied to the pressure data and the flow rate data,” (Para 49, “In a further, preferred embodiment of the invention, the calculated and output anomaly score is used for anomaly detection, in particular for leakage detection… Furthermore, the normal state data may comprise pressure signals and/or flow signals” – Examiner’s note (EN): The quote refers to a “calculated and output anomaly score” this denotes the use of a trained ML model. Earlier in the text, Streichert defines that this score comes from a trained normal state model. Para 6, “Provide a trained normal state model as a one-class model that has been trained in a training phase…” Para 9, “Determine deviations of extracted features from learned features of the normal state model…” Para 10, “Calculate an anomaly score from the determined deviations” Para 21, “In a preferred embodiment of the invention, the normal state model is a statistical model and/or machine learning model.”)
“wherein the at least one trained ML model is trained to identify a plurality of compressed air leak patterns based on pressure and flow rate data,” (Para 6-10, “Provide a trained normal state model as a one-class model that has been trained in a training phase with normal state data representing a normal state of the pneumatic system; Continuous acquisition of sensor data of the pneumatic system by means of a set of sensors; Extract features from the acquired sensor data; Determine deviations of extracted features from learned features of the normal state model using a distance metric…; Calculate an anomaly score from the determined deviations and Output of the calculated anomaly score.” Para 49, “the normal state data may comprise pressure signals and/or flow signals” Para 112, “features are extracted by means of an extractor 304, i.e. quantities are derived that provide information about the functioning of the pneumatic system… include the actuator features “reaction time extension”, “travel time extension”, “reaction time retraction”, and “travel time retraction”” – EN: this denotes the trained ML model is applied to sensor data, which includes “pressure signals and/or flow signals”, to perform “leakage detection” (para 5). The model operates by extracting specific features from this data, such as “reaction time extension” and “travel time extension”, and then “determining deviations” of these extracted features from learned norms. These determined deviations constitute the “leak patterns” because they are distinct data signatures used to identify the presence of a leak.)
“the ML model trained using… training samples including pressure data and flow rate data…” (Para 16, “These features are based on measured state data of the normal state, such as pressure, pressure curve, flow, flow curve or time stamp. The One-Class model learns these features and defines the normal state based on them. For learning the normal state, the measured normal state data of several whole production cycles are ideally used during training” Para 31, “In another embodiment, the model can be trained with state data of the normal state and mixed state data containing both normal state data and fault data.” – EN: the measured state data of “several whole production cycles” used during the training phase corresponds to a plurality of “training samples” which include pressure data and flow rate data.)
Streichert does not explicitly disclose:
“receiving, by a processor of a leak detection system, an indication of at least one shutdown event in the compressed air system from a supervisory controller operatively coupled with the compressed air system and capable of shutting down the compressed air system, the compressed air system comprising at least one compressed air client consuming compressed air delivered by at least one air compressor,” “wherein during the at least one shutdown event demand of compressed air by the at least one compressed air client is at least partially reduced, such that compressed air flow associated with air leaks becomes detectable;” “…during the at least one shutdown event;” and “…[training samples including]... data from a plurality of previous shutdown events, each one of the plurality of shutdown events associated with an [annotation] indicative of whether the shutdown event was associated with at least one compressed air leak or no compressed air leak”
However, Koshinaka teaches:
receiving, by a processor of a leak detection system, an indication of at least one shutdown event in the compressed air system from a supervisory controller operatively coupled with the compressed air system (Para 26, “The compressed air monitor system 1 of the present embodiment further includes a control panel 5 that is electrically connected to the flow meter 4 through an electric line 51. All of the valves 250-253 are electrically connected to the control panel 5 through an electric line 52 to provide the control panel 5 with information about a current open/close state of each air-driven device 31-33, i.e., a current operational state of each air-driven device 31-33.” Para 27, “The deactivated state refers to a state where all of the air-driven devices 31-33 are stopped, or deactivated.” Para 28, “Furthermore, the control panel 5 is connected to the monitor computer 6 through an electric line 53. The flow rate data and the information of operational state of the compressed air circuit 2 are transmitted from the control panel 5 to the monitor computer 6.” Para 30, “The operational state identifying means of the monitor computer 6 receives information about the current operational state from the control panel 5 and determines whether the current operational state in the compressed air circuit 2 is the activated state or the deactivated state.” – EN: the control panel 5 corresponds to the “supervisory controller” operatively coupled (electric lines 51-53) with the compressed air circuit, the monitor computer 6 corresponds to the “processor of a leak detection system”, and the transmitted information indicating the deactivated state (all air-driven devices stopped) corresponds to the received “indication of at least one shutdown event”.)
“…and capable of shutting down the compressed air system” (Para 27, “When the first valve 250 is closed to disable supply of compressed air to the other valves 251-253, the current operational state is identified as the deactivated state.” Para 55, “In a second embodiment, the control panel 5 of the compressed air monitor system 1 of the first embodiment includes an individual operating means for individually operating (i.e., activating) each of all the air-driven devices 31-33 of the compressed air circuit 2.” Para 57, “That is, flow rate data, which is indicated by the flow meter 4, is read while all of the air-driven devices 31-33 are deactivated by the individual operating means.” – EN: the control panel’s operating means deactivates (shuts down) all of the air-driven devices of the compressed air circuit, hence the supervisory controller is “capable of shutting down the compressed air system”.)
“…the compressed air system comprising at least one compressed air client consuming compressed air delivered by at least one air compressor,” (Para 6, “Generally, a plurality of air-driven devices is connected to a single supply line, which extends from a compressed air source, to form a compressed air circuit.” Para 21, “The compressed air circuit 2 includes a supply line 20 of compressed air and a plurality of air-driven devices 31-33. The supply line 20 is connected to a compressed air source. The air-driven devices 31-33 are connected to the supply line 20 and are driven by compressed air (serving as a drive source) supplied through the supply line 20.” Para 24, “The other three valves 251-253 are used to operate the corresponding air-driven devices 31-33, which are air cylinders in this instance.” – EN: the air-driven devices 31-33 (air cylinders) correspond to the “compressed air clients” consuming the compressed air, and the compressed air source that delivers the compressed air through the supply line 20 corresponds, under the broadest reasonable interpretation, to the “at least one air compressor” delivering compressed air to the at least one compressed air client.)
“wherein during the at least one shutdown event demand of compressed air by the at least one compressed air client is at least partially reduced, such that compressed air flow associated with air leaks becomes detectable;” (Para 27, “The deactivated state refers to a state where all of the air-driven devices 31-33 are stopped, or deactivated.” Para 33, “The first master flow rate M1 of the present embodiment is chosen to be a value that is about 110% of measured flow rate data, which is measured when all of the air-driven devices 31-33 are deactivated while minimizing air leakage from each air-driven device 31-33 in the compressed air circuit 2.” Para 41, “…control proceeds to step S405 where a deactivation time leakage alert, which indicates the moderate level of air leakage during the deactivation period, is outputted.” – EN: with all air-driven devices (the compressed air clients) stopped during the deactivated state, demand is reduced such that measured flow exceeding the reference master flow rate M1 is flow attributable to air leakage, i.e., the compressed air flow associated with air leaks becomes detectable.)
“…measured… during the at least one shutdown event” (Para 29, “In the present embodiment, an accumulated flow rate is measured with the flow meter 4 during each period of five minutes, and this flow rate is transmitted to the monitor computer 6 through the control panel 5 as flow rate data F at every five minute interval.” Para 43, “…a time period between points B and C corresponds to the deactivated state, i.e., the deactivation period.” Para 44, “…the flow rate data F is measured five times between points B and C…” – EN: the flow rate data is measured, and received by the monitor computer 6, during the deactivation (shutdown) period.)
and “…[training samples including]… data from a plurality of previous shutdown events, each one of the plurality of shutdown events associated with an … indicative of whether the shutdown event was associated with at least one compressed air leak or no compressed air leak” (Para 44, “…the flow rate data F is measured five times between points B and C, and each of the five measured flow rate data F between points B and C is below the first master flow rate M1. Thus, the deactivation time normal state information is outputted.” Para 47, “…each of the second to fifth flow rate data F21-F24 between points B and C is above the first master flow rate M1 but is less than a value of M1b, which indicates a 10% increase over the first master flow rate M1. Thus, the deactivation time leakage alert is outputted.” Para 52, “Practical use of the compressed air monitor system 1 allows monitoring of leakage of compressed air for a long period of time. Thus, it is possible to predict the next possible start time of leakage of the compressed air.” Para 74, “The first master flow rate is determined based on the measured flow rate data that is measured when all of the air-driven devices are deactivated under the normal state where air leakage from each air-driven device in the compressed air circuit is minimized.” – EN: over long term monitoring, flow rate data is collected for a plurality of deactivation (shutdown) periods, and each deactivation period is associated with either “deactivation time normal state information” (no compressed air leak) or a “deactivation time leakage alert” (compressed air leak), i.e., data from a plurality of previous shutdown events each one associated with a leak/no-leak determination.)
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the leak detection using machine learning system of Streichert with the compressed air circuit comprising air-driven devices supplied with compressed air from a compressed air source, the shutdown (deactivated) state indication received from the supervisory control panel, and the measurement of compressed air flow during the shutdown state of Koshinaka. The motivation for doing so would be to improve the accuracy of leak detection by performing the analysis during periods where the compressed air clients are stopped and demand is reduced, such that substantially any remaining compressed air flow is attributable to leakage, thereby making leaks distinguishable from normal consumption, and to do so continuously without waiting for periodic manual inspections. Koshinaka explicitly teaches this benefit, stating that “it is an objective of the present invention to provide a compressed air monitor system, which continuously monitors leakage of compressed air in a corresponding facility to allow a reduction in the number of inspection steps and to restrain sudden stop of the facility and wasteful consumption of compressed air in the facility.” (Koshinaka, Para 11) and “When the air leakage is detected, it is possible to take the best measures. Thus, there is no need to wait for the periodic inspection. In this way, it is possible to prevent sudden stop of the facility and to restrain wasteful consumption of the compressed air.” (Koshinaka, Para 51)
Further, the combination of Streichert and Koshinaka does not appear to explicitly disclose:
“the ML model trained using annotated training samples…” “…associated with an annotation…” and “wherein each annotation is automatically assigned to the respective annotated training samples by indications received from the supervisory controller”
However, Wang teaches:
“the ML model trained using annotated training samples…” and “…associated with an annotation…” (Para 69, “The component failure prediction system 104 may further identify labels or categories for machine learning. It will be appreciated that component failure prediction system 104 may, in some embodiments, identify labels.” Para 85, “Further, as used herein, a model training period may include a time period used to select training instances. An instance is a set of time series/event features along with the failure/non-failure of a particular component in a renewable energy asset (e.g., a wind turbine) in a specified time period.” Para 177, “In various embodiments, the WT failure forecasting engine 1316 may receive extracted features, event data, and labels (e.g., failure or no failure) for each instance and may provide the information to the model training module 506.” – EN: a label of “failure or no failure” attached to each training instance corresponds to an “annotation” indicative of whether the respective sample was associated with a fault (in the combination, at least one compressed air leak) or no fault (no compressed air leak), and the labeled instances provided to the model training module correspond to “annotated training samples” used to train the ML model.)
“wherein each annotation is automatically assigned to the respective annotated training samples by indications received from the supervisory controller” (Para 145, “Supervisory control and Data Acquisition (SCADA) is a control system architecture often used to monitor and control aspects of hardware and software systems and networks.” Para 147, “The SCADA system may further log data regarding any number of the wind turbine such as failures, health information, performance, and the like.” Para 160, “In step 1404, the WT failure data and asset data module 1306 may receive historical wind turbine component failure data and wind turbine asset metadata from one or more SCADA systems used to supervise and monitor any number of wind turbines… The historical wind turbine component failure data may include but not be limited to a turbine identifier (e.g., TurbineId), failure start time (e.g., FailureStartTime), failure end time (e.g., FailureEndTime), component, subcomponent, part, comments, and/or the like.” Para 164, “The WT cohort for model development module 1314 may also identify both healthy time window WT instances and component failure time window WT instances from the failure data for any number of components, subcomponents, parts, wind turbines, and/or cohorts (e.g. instance number 2: 303615 had generator failure during 20180101-20180115).” Para 81, “Failure event labels may be extracted from the duration of the predicted time window.” – EN: the failure/no-failure labels of the training instances are assigned automatically from the failure data and event indications logged by, and received from, the SCADA (supervisory control) system rather than by manual annotation; in the combination, the leak/no-leak indications associated with each shutdown event of Streichert in view of Koshinaka are automatically assigned as the annotations of the respective annotated training samples.)
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to modify the machine learning leak detection using data measured during shutdown events reported by the supervisory controller as taught by Streichert in view of Koshinaka to automatically annotate (label) each training sample of a previous shutdown event with the leak/no-leak indication received from the supervisory controller as taught by Wang. The motivation for doing so would be to leverage the status and failure indications already logged by the supervisory control system to automatically construct a labeled training data set at scale, thereby improving the accuracy and scalability of the trained model while eliminating the need for manual labeling of the training samples. Wang explicitly teaches this benefit, stating that “By leveraging SCADA logs and metadata using agnostic representations to derive patterns useful in machine learning, the failure prediction models may improve for accuracy and scalability.” (Wang, Para 180) and “SCADA system provide important signals for historical and present status of any number of wind turbines (WTs). However, an unmanageable number of alarms and event logs generated by a SCADA system is often ignored in wind turbine forecasting. Some embodiments of systems and method discussed herein leverages machine learning method(s) to extract a number of actionable insights from this valuable information.” (Wang, Para 149)
Claim 2
Streichert in view of Koshinaka further in view of Wang teaches all the limitations of claim 1, Streichert further teaches:
wherein the at least one trained ML model is trained to detect the at least one compressed air leak (Para 6, “Provide a trained normal state model as a one-class model that has been trained in a training phase”, Para 49, “In a further, preferred embodiment of the invention, the calculated and output anomaly score is used for anomaly detection, in particular for leakage detection”) based on analysis of the pressure data and the flow rate data (Para 8, “Extract features from the acquired sensor data;” Para 32, “For example, one or more of the following sensors may be used: Flow meter, Pressure sensor…”) accumulated during a predefined time period. (Para 69, “The time window length can be specified as a static value in the unit number of cycles or in time units, such as 10 seconds.” Para 68, “a result of the pattern recognition algorithm can be used to calculate time windows in which feature extraction is performed.”)
Claim 3
Streichert in view of Koshinaka further in view of Wang teaches all the limitations of claim 1, Streichert further teaches:
wherein the pressure data and the flow rate data are measured by at least one pressure sensor and at least one flow sensor respectively which are deployed in the compressed air system. (Para 107, “In a step 204, sensor data of the pneumatic system 100 is continuously acquired by means of a set of sensors. The set of sensors includes at least the sensors m1 and m2 described above. In addition, several of these or other types of sensors (for example, flow sensors, pressure sensors, microphones, structure-borne sound pickups) may collect sensor data.”)
Claim 5
Streichert in view of Koshinaka further in view of Wang teaches all the limitations of claim 3, Streichert further teaches:
wherein the at least one trained ML model is further trained to detect the at least one compressed air leak based on analysis of aggregated flow rate data aggregating flow rate data measured by a plurality of flow sensors deployed to measure the air flow rate at different locations in the compressed air system. (Para 60, “Alternatively or in addition, the anomaly score can also be output locally for certain subgroups and/or functional units of a pneumatic system… This involves processing a large number of sensor signals from different sensors, which is advantageous for the efficiency and scalability of anomaly detection and rectification.” – EN: this denotes the use of multiple sensors (including flow meters [para 33]) across various functional units/sub-groups to calculate anomaly scores [anomaly score is derived from the trained ML model]. Under BRI, “processing a large number of sensor signals” (which include flow sensors) to output a score maps to “aggregating flow rate data” from sensors at “different locations”.)
Claim 6
Streichert in view of Koshinaka further in view of Wang teaches all the limitations of claim 1, Streichert further teaches:
wherein the at least one trained ML model is further trained to detect the at least one compressed air leak based on analysis of an averaged air flow rate averaging the measured air flow rate to compensate for compressor load and unload periods. (Para 75, “the extracted features can comprise statistical characteristics and in particular mean values… of the sensor data” Para 84, “A differentiator for determining deviations of extracted features from learned features of the normal state model” – EN: this denotes using mean values of sensor data which include flow sensor data as noted in (Para 32) to train a model. Additionally, averaging data over a production cycle results in compensating for fluctuations such as load and unload periods within that cycle.)
Claim 7
Streichert in view of Koshinaka further in view of Wang teaches all the limitations of claim 1, Streichert further teaches:
wherein the at least one trained ML model is further trained to detect the at least one compressed air leak based on reference pressure and/or reference air flow defining demand of compressed air in the compressed air system during operating conditions in which no compressed air leaks are present. (Para 6, “Provide a trained normal state model as a one-class model that has been trained in a training phase with normal state data representing a normal state of the pneumatic system.” Para 14, “In the normal state, there are no anomalies, leaks or other faults, and the production process is functioning perfectly.” Para 15, “A pneumatic system can have more than one operating state… In this case, the normal state of the pneumatic system must be learned for each operating state during the training phase.” – EN: the normal state data (which comprises pressure signals and/or flow signals per Para 49) constitutes reference pressure/flow data measured while “there are no anomalies, leaks or other faults”, i.e., reference data defining the demand of compressed air during operating conditions in which no compressed air leaks are present, learned for each operating state.)
Claim 8
Streichert in view of Koshinaka further in view of Wang teaches all the limitations of claim 1, Streichert further teaches:
wherein the at least one trained ML model is further trained to detect the at least one compressed air leak based on normalized flow rate data in which the air flow rate is normalized according to the pressure. (Para 113, “The extracted features can be normalized to simplify their representation in an n-dimensional space. This is particularly advantageous if the features derived from the sensor data contain different physical quantities and/or magnitudes (for example, pressure and time) that are to be further processed together.” – EN: this denotes extracting features from flow and pressure. And these features are “normalized… to simplify their representation… advantageous if the features… contain different physical quantities… for example, pressure”. By normalizing flow features and pressure features into a common “n-dimensional space” so they can be “processed together”, the flow rate is being normalized according to or in relation to the pressure data and its magnitude. In the BRI of the claim, this maps to normalizing the flow data in a way that accounts for pressure.)
Claim 11
Streichert in view of Koshinaka further in view of Wang teaches all the limitations of claim 1, Streichert further teaches:
“…identified by the at least one trained ML model which is further trained to detect … based on measured pressure data and flow rate data” (Para 21, “In a preferred embodiment of the invention, the normal state model is a statistical model and/or machine learning model.” Para 15, “A pneumatic system can have more than one operating state… In this case, the normal state of the pneumatic system must be learned for each operating state during the training phase.” Para 16, “These features are based on measured state data of the normal state, such as pressure, pressure curve, flow, flow curve or time stamp. The One-Class model learns these features” – EN: this denotes the use of a trained ML model that identifies specific operating states of the system and is learned based on pressure and flow rate data.)
Streichert does not explicitly disclose:
“wherein the at least one shutdown event is identified” and “at least one shutdown event”
However, Koshinaka teaches:
“wherein the at least one shutdown event is identified” and “at least one shutdown event” (Para 27, “The deactivated state refers to a state where all of the air-driven devices 31-33 are stopped, or deactivated.” Para 30, “The operational state identifying means of the monitor computer 6 receives information about the current operational state from the control panel 5 and determines whether the current operational state in the compressed air circuit 2 is the activated state or the deactivated state.” Para 72, “The master flow rate (i.e., the second master flow rate) used in the activated state is preferably greater than the master flow rate (i.e., the first master flow rate) used in the deactivated state.” – EN: the deactivated state, where all air-driven devices are stopped, corresponds to the “shutdown event” which is identified by the operational state identifying means; further, the activated and deactivated states are characterized by different (master) flow rate levels, i.e., the shutdown state is distinguishable based on the measured flow rate data.)
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the machine learning model that is trained based on pressure and flow rate data of Streichert with the identification of the shutdown (deactivated) events of Koshinaka. The motivation for doing so would be to automatically identify the shutdown periods from the measured data itself, thereby ensuring the leak analysis is applied to the correct periods of reduced demand even without, or in addition to, the state information provided by the control panel. Koshinaka explicitly teaches that the operational states are readily identifiable from the flow rate levels because the states are characterized by different master flow rates, stating that “With the above arrangement, it is easy to categorize the operational states of the air-driven devices, and it is easy to identify the current operational state of the air-driven devices.” (Koshinaka, Para 73)
Claim 13
Streichert in view of Koshinaka further in view of Wang teaches all the limitations of claim 1, Koshinaka further teaches:
wherein the at least one shutdown event is identified according to an indication received from at least one control unit of the compressed air system. (Para 26, “All of the valves 250-253 are electrically connected to the control panel 5 through an electric line 52 to provide the control panel 5 with information about a current open/close state of each air-driven device 31-33, i.e., a current operational state of each air-driven device 31-33.” Para 29, “…when the flow rate data is transmitted from the control panel 5 to the monitor computer 6, information about the current operational state, i.e., activated state or deactivated state is also transmitted from the control panel 5 to the monitor computer 6.” Para 30, “The operational state identifying means of the monitor computer 6 receives information about the current operational state from the control panel 5 and determines whether the current operational state in the compressed air circuit 2 is the activated state or the deactivated state.” – EN: the control panel 5 corresponds to the “control unit” of the compressed air system, and the operational state information (deactivated state, i.e., shutdown event) received from the control panel 5 corresponds to the “indication received from at least one control unit”. This limitation is taught by Koshinaka as set forth in the rejection of claim 1 above, and the motivation to combine is the same as set forth in the rejection of claim 1.)
Claim 16
Streichert in view of Koshinaka further in view of Wang teaches all the limitations of claim 1, Streichert further teaches:
“further comprising transmitting at least one alert reporting the at least one detected compressed air leak.” (Para 59, “when a configurable threshold value of the anomaly score is exceeded, the operator of the pneumatic system is alerted. This can happen, for example, by means of a warning message”)
Claim 17
Streichert teaches:
A system for detecting leaks in a compressed air system using machine learning (ML), comprising: (Para 97, “The present invention relates to a method and a device for monitoring the condition of pneumatic systems, in particular for detecting anomalies such as leaks.” Para 98, “FIG. 1 shows an overview illustration of a pneumatic system 100 with a condition monitoring unit 114.” Para 100, “The condition monitoring unit 114 includes, for example, models and their parameters, training and inference algorithms, training data, state data, meta-parameters, and configuration parameters (not shown).” – Examiner’s Note (EN): this denotes a pneumatic system that is equivalent to the broadest reasonable interpretation of “compressed air system” because the pneumatic system comprises actuators that function as the required “compressed air clients” consuming air delivered via valves and supply lines.)
“at least one pressure sensor and at least one flow sensor deployed in the compressed air system to measure pressure data and flow rate data, respectively;” (Para 32, “For example, one or more of the following sensors may be used: Flow meter, Pressure sensor…” Para 107, “In a step 204, sensor data of the pneumatic system 100 is continuously acquired by means of a set of sensors. The set of sensors includes at least the sensors m1 and m2 described above. In addition, several of these or other types of sensors (for example, flow sensors, pressure sensors, microphones, structure-borne sound pickups) may collect sensor data.”)
“at least one processor configured to execute a code, the code comprising:” (Para 79, “the corresponding functional features of the method are formed by corresponding representational modules, in particular by hardware modules or microprocessor modules, of the system or product, and vice versa.” Para 87, “a computer program comprising instructions which, when the computer program is executed by a computer, cause the computer program to execute the method described above.”)
“code instructions to receive the pressure data and the flow rate data measured by the at least one pressure sensor and the at least one flow sensor in the compressed air system…” (Para 5, “The method comprises the following method steps… Continuous acquisition of sensor data of the pneumatic system by means of a set of sensors” Para 101, “The monitoring unit 114 receives sensor data via the stroke 112 in the training and operating state of the pneumatic system 100” – EN: the condition monitoring unit 114 corresponds to the “at least one processor” which receives the pressure data and flow rate data measured by the flow and pressure sensors deployed in the pneumatic (compressed air) system.)
“code instructions to detect at least one compressed air leak in the compressed air system using at least one trained ML model applied to the pressure data and the flow rate data,” (Para 49, “In a further, preferred embodiment of the invention, the calculated and output anomaly score is used for anomaly detection, in particular for leakage detection… Furthermore, the normal state data may comprise pressure signals and/or flow signals” – Examiner’s note (EN): The quote refers to a “calculated and output anomaly score” this denotes the use of a trained ML model. Earlier in the text, Streichert defines that this score comes from a trained normal state model. Para 6, “Provide a trained normal state model as a one-class model that has been trained in a training phase…” Para 9, “Determine deviations of extracted features from learned features of the normal state model…” Para 10, “Calculate an anomaly score from the determined deviations” Para 21, “In a preferred embodiment of the invention, the normal state model is a statistical model and/or machine learning model.”)
“wherein the at least one trained ML model is trained to identify a plurality of compressed air leak patterns based on pressure and flow rate data,” (Para 6-10, “Provide a trained normal state model as a one-class model that has been trained in a training phase with normal state data representing a normal state of the pneumatic system; Continuous acquisition of sensor data of the pneumatic system by means of a set of sensors; Extract features from the acquired sensor data; Determine deviations of extracted features from learned features of the normal state model using a distance metric…; Calculate an anomaly score from the determined deviations and Output of the calculated anomaly score.” Para 49, “the normal state data may comprise pressure signals and/or flow signals” Para 112, “features are extracted by means of an extractor 304, i.e. quantities are derived that provide information about the functioning of the pneumatic system… include the actuator features “reaction time extension”, “travel time extension”, “reaction time retraction”, and “travel time retraction”” – EN: this denotes the trained ML model is applied to sensor data, which includes “pressure signals and/or flow signals”, to perform “leakage detection” (para 5). The model operates by extracting specific features from this data, such as “reaction time extension” and “travel time extension”, and then “determining deviations” of these extracted features from learned norms. These determined deviations constitute the “leak patterns” because they are distinct data signatures used to identify the presence of a leak.)
“the ML model trained using… training samples including pressure data and flow rate data…” (Para 16, “These features are based on measured state data of the normal state, such as pressure, pressure curve, flow, flow curve or time stamp. The One-Class model learns these features and defines the normal state based on them. For learning the normal state, the measured normal state data of several whole production cycles are ideally used during training” Para 31, “In another embodiment, the model can be trained with state data of the normal state and mixed state data containing both normal state data and fault data.” – EN: the measured state data of “several whole production cycles” used during the training phase corresponds to a plurality of “training samples” which include pressure data and flow rate data.)
Streichert does not explicitly disclose:
“code instructions to receive an indication of at least one shutdown event in the compressed air system from a supervisory controller operatively coupled with the compressed air system and capable of shutting down the compressed air system, the compressed air system comprising at least one compressed air client consuming compressed air delivered by at least one air compressor,” “wherein during the at least one shutdown event demand of compressed air by the at least one compressed air client is at least partially reduced such that compressed air flow attributable to leaks becomes detectable relative to background client consumption;” “…during the at least one shutdown event;” and “…[training samples including]... from a plurality of previous shutdown events, each one of the plurality of shutdown events associated with an … indicative of whether the shutdown event was associated with at least one compressed air leak or no compressed air leak”
However, Koshinaka teaches:
code instructions to receive an indication of at least one shutdown event in the compressed air system from a supervisory controller operatively coupled with the compressed air system (Para 26, “The compressed air monitor system 1 of the present embodiment further includes a control panel 5 that is electrically connected to the flow meter 4 through an electric line 51. All of the valves 250-253 are electrically connected to the control panel 5 through an electric line 52 to provide the control panel 5 with information about a current open/close state of each air-driven device 31-33, i.e., a current operational state of each air-driven device 31-33.” Para 27, “The deactivated state refers to a state where all of the air-driven devices 31-33 are stopped, or deactivated.” Para 28, “Furthermore, the control panel 5 is connected to the monitor computer 6 through an electric line 53. The flow rate data and the information of operational state of the compressed air circuit 2 are transmitted from the control panel 5 to the monitor computer 6.” Para 30, “The operational state identifying means of the monitor computer 6 receives information about the current operational state from the control panel 5 and determines whether the current operational state in the compressed air circuit 2 is the activated state or the deactivated state.” – EN: the control panel 5 corresponds to the “supervisory controller” operatively coupled (electric lines 51-53) with the compressed air circuit, the monitor computer 6 corresponds to the “at least one processor”, and the transmitted information indicating the deactivated state (all air-driven devices stopped) corresponds to the received “indication of at least one shutdown event”.)
“…and capable of shutting down the compressed air system” (Para 27, “When the first valve 250 is closed to disable supply of compressed air to the other valves 251-253, the current operational state is identified as the deactivated state.” Para 55, “In a second embodiment, the control panel 5 of the compressed air monitor system 1 of the first embodiment includes an individual operating means for individually operating (i.e., activating) each of all the air-driven devices 31-33 of the compressed air circuit 2.” Para 57, “That is, flow rate data, which is indicated by the flow meter 4, is read while all of the air-driven devices 31-33 are deactivated by the individual operating means.” – EN: the control panel’s operating means deactivates (shuts down) all of the air-driven devices of the compressed air circuit, hence the supervisory controller is “capable of shutting down the compressed air system”.)
“…the compressed air system comprising at least one compressed air client consuming compressed air delivered by at least one air compressor,” (Para 6, “Generally, a plurality of air-driven devices is connected to a single supply line, which extends from a compressed air source, to form a compressed air circuit.” Para 21, “The compressed air circuit 2 includes a supply line 20 of compressed air and a plurality of air-driven devices 31-33. The supply line 20 is connected to a compressed air source. The air-driven devices 31-33 are connected to the supply line 20 and are driven by compressed air (serving as a drive source) supplied through the supply line 20.” Para 24, “The other three valves 251-253 are used to operate the corresponding air-driven devices 31-33, which are air cylinders in this instance.” – EN: the air-driven devices 31-33 (air cylinders) correspond to the “compressed air clients” consuming the compressed air, and the compressed air source that delivers the compressed air through the supply line 20 corresponds, under the broadest reasonable interpretation, to the “at least one air compressor” delivering compressed air to the at least one compressed air client.)
“wherein during the at least one shutdown event demand of compressed air by the at least one compressed air client is at least partially reduced such that compressed air flow attributable to leaks becomes detectable relative to background client consumption;” (Para 27, “The deactivated state refers to a state where all of the air-driven devices 31-33 are stopped, or deactivated.” Para 32, “The above three-step determination process is performed using two master flow rates, i.e., a first master flow rate M1, which is used as the reference value during the deactivated state, and a second master flow rate M2, which is used as the reference value during the activated state. When the flow rate data F, which is the current measured value, is below a corresponding one of the first master flow rate M1 and the second master flow rate M2, the normal state information is outputted. When the flow rate data F exceeds a corresponding one of the first master flow rate M1 and the second master flow rate M2 by an amount that is equal to or less than 10% of a corresponding one of the first master flow rate M1 and the second master flow rate M2, the leakage alert is outputted.” Para 33, “The first master flow rate M1 of the present embodiment is chosen to be a value that is about 110% of measured flow rate data, which is measured when all of the air-driven devices 31-33 are deactivated while minimizing air leakage from each air-driven device 31-33 in the compressed air circuit 2.” Para 41, “…control proceeds to step S405 where a deactivation time leakage alert, which indicates the moderate level of air leakage during the deactivation period, is outputted.” – EN: with all air-driven devices (the compressed air clients) stopped during the deactivated state, demand is reduced; the master flow rates M1/M2 are reference values derived from the background flow level measured in the respective operational state of the clients while leakage is minimized, and measured flow exceeding the respective reference value is determined to be leakage, i.e., the compressed air flow attributable to leaks becomes detectable relative to the background client consumption level.)
“…measured… during the at least one shutdown event” (Para 29, “In the present embodiment, an accumulated flow rate is measured with the flow meter 4 during each period of five minutes, and this flow rate is transmitted to the monitor computer 6 through the control panel 5 as flow rate data F at every five minute interval.” Para 43, “…a time period between points B and C corresponds to the deactivated state, i.e., the deactivation period.” Para 44, “…the flow rate data F is measured five times between points B and C…” – EN: the flow rate data is measured, and received by the monitor computer 6, during the deactivation (shutdown) period.)
and “…[training samples including]… data from a plurality of previous shutdown events, each one of the plurality of shutdown events associated with an … indicative of whether the shutdown event was associated with at least one compressed air leak or no compressed air leak” (Para 44, “…the flow rate data F is measured five times between points B and C, and each of the five measured flow rate data F between points B and C is below the first master flow rate M1. Thus, the deactivation time normal state information is outputted.” Para 47, “…each of the second to fifth flow rate data F21-F24 between points B and C is above the first master flow rate M1 but is less than a value of M1b, which indicates a 10% increase over the first master flow rate M1. Thus, the deactivation time leakage alert is outputted.” Para 52, “Practical use of the compressed air monitor system 1 allows monitoring of leakage of compressed air for a long period of time. Thus, it is possible to predict the next possible start time of leakage of the compressed air.” Para 74, “The first master flow rate is determined based on the measured flow rate data that is measured when all of the air-driven devices are deactivated under the normal state where air leakage from each air-driven device in the compressed air circuit is minimized.” – EN: over long term monitoring, flow rate data is collected for a plurality of deactivation (shutdown) periods, and each deactivation period is associated with either “deactivation time normal state information” (no compressed air leak) or a “deactivation time leakage alert” (compressed air leak), i.e., data from a plurality of previous shutdown events each one associated with a leak/no-leak determination.)
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the leak detection using machine learning system of Streichert with the compressed air circuit comprising air-driven devices supplied with compressed air from a compressed air source, the shutdown (deactivated) state indication received from the supervisory control panel, and the measurement of compressed air flow during the shutdown state of Koshinaka. The motivation for doing so would be to improve the accuracy of leak detection by performing the analysis during periods where the compressed air clients are stopped and demand is reduced, such that substantially any remaining compressed air flow is attributable to leakage, thereby making leaks distinguishable from normal consumption, and to do so continuously without waiting for periodic manual inspections. Koshinaka explicitly teaches this benefit, stating that “it is an objective of the present invention to provide a compressed air monitor system, which continuously monitors leakage of compressed air in a corresponding facility to allow a reduction in the number of inspection steps and to restrain sudden stop of the facility and wasteful consumption of compressed air in the facility.” (Koshinaka, Para 11) and “When the air leakage is detected, it is possible to take the best measures. Thus, there is no need to wait for the periodic inspection. In this way, it is possible to prevent sudden stop of the facility and to restrain wasteful consumption of the compressed air.” (Koshinaka, Para 51)
Further, the combination of Streichert and Koshinaka does not appear to explicitly disclose:
“the ML model trained using annotated training samples…” “…associated with an annotation…” and “wherein each annotation is automatically assigned to the respective annotated training samples by indications received from the supervisory controller”
However, Wang teaches:
“the ML model trained using annotated training samples…” and “…associated with an annotation…” (Para 69, “The component failure prediction system 104 may further identify labels or categories for machine learning. It will be appreciated that component failure prediction system 104 may, in some embodiments, identify labels.” Para 85, “Further, as used herein, a model training period may include a time period used to select training instances. An instance is a set of time series/event features along with the failure/non-failure of a particular component in a renewable energy asset (e.g., a wind turbine) in a specified time period.” Para 177, “In various embodiments, the WT failure forecasting engine 1316 may receive extracted features, event data, and labels (e.g., failure or no failure) for each instance and may provide the information to the model training module 506.” – EN: a label of “failure or no failure” attached to each training instance corresponds to an “annotation” indicative of whether the respective sample was associated with a fault (in the combination, at least one compressed air leak) or no fault (no compressed air leak), and the labeled instances provided to the model training module correspond to “annotated training samples” used to train the ML model.)
“wherein each annotation is automatically assigned to the respective annotated training samples by indications received from the supervisory controller” (Para 145, “Supervisory control and Data Acquisition (SCADA) is a control system architecture often used to monitor and control aspects of hardware and software systems and networks.” Para 147, “The SCADA system may further log data regarding any number of the wind turbine such as failures, health information, performance, and the like.” Para 160, “In step 1404, the WT failure data and asset data module 1306 may receive historical wind turbine component failure data and wind turbine asset metadata from one or more SCADA systems used to supervise and monitor any number of wind turbines… The historical wind turbine component failure data may include but not be limited to a turbine identifier (e.g., TurbineId), failure start time (e.g., FailureStartTime), failure end time (e.g., FailureEndTime), component, subcomponent, part, comments, and/or the like.” Para 164, “The WT cohort for model development module 1314 may also identify both healthy time window WT instances and component failure time window WT instances from the failure data for any number of components, subcomponents, parts, wind turbines, and/or cohorts (e.g. instance number 2: 303615 had generator failure during 20180101-20180115).” Para 81, “Failure event labels may be extracted from the duration of the predicted time window.” – EN: the failure/no-failure labels of the training instances are assigned automatically from the failure data and event indications logged by, and received from, the SCADA (supervisory control) system rather than by manual annotation; in the combination, the leak/no-leak indications associated with each shutdown event of Streichert in view of Koshinaka are automatically assigned as the annotations of the respective annotated training samples.)
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to modify the machine learning leak detection using data measured during shutdown events reported by the supervisory controller as taught by Streichert in view of Koshinaka to automatically annotate (label) each training sample of a previous shutdown event with the leak/no-leak indication received from the supervisory controller as taught by Wang. The motivation for doing so would be to leverage the status and failure indications already logged by the supervisory control system to automatically construct a labeled training data set at scale, thereby improving the accuracy and scalability of the trained model while eliminating the need for manual labeling of the training samples. Wang explicitly teaches this benefit, stating that “By leveraging SCADA logs and metadata using agnostic representations to derive patterns useful in machine learning, the failure prediction models may improve for accuracy and scalability.” (Wang, Para 180) and “SCADA system provide important signals for historical and present status of any number of wind turbines (WTs). However, an unmanageable number of alarms and event logs generated by a SCADA system is often ignored in wind turbine forecasting. Some embodiments of systems and method discussed herein leverages machine learning method(s) to extract a number of actionable insights from this valuable information.” (Wang, Para 149)
Claim 4 is rejected under 35 U.S.C. 103 as being unpatentable over patent application US 2023/0088241 A1 Streichert et al., hereinafter “Streichert” in view of patent application US 2003/0187595 A1 Koshinaka et al., hereinafter “Koshinaka” further in view of patent application US 2020/0201950 A1 Wang et al., hereinafter “Wang” further in view of patent application US 2008/0314122 A1 Hunaidi et al.
Claim 4
Streichert in view of Koshinaka further in view of Wang teaches all the limitations of claim 3, Streichert further teaches:
wherein the pressure data used for detecting the at least one compressed air leak is measured by at least one… pressure sensor… deployed in the compressed air system. (Para 107, “In a step 204, sensor data of the pneumatic system 100 is continuously acquired by means of a set of sensors. The set of sensors includes at least the sensors m1 and m2 described above. In addition, several of these or other types of sensors (for example, flow sensors, pressure sensors, microphones, structure-borne sound pickups) may collect sensor data.”)
Streichert in view of Koshinaka further in view of Wang does not explicitly disclose:
“…at least one selected closest pressure sensor which is located at a shortest distance from the at least one flow sensor among a plurality of pressure sensors…”
However, Hunaidi teaches:
“…at least one selected closest pressure sensor which is located at a shortest distance from the at least one flow sensor among a plurality of pressure sensors…” (Figure 2, EN: this denotes a flow meter 32 and multiple pressure gauges 34, 36, and 38. Among the plurality of pressure gauges, pressure gauge 36 corresponds to the “selected closest pressure sensor”. See Para 44, “Pressure sensor 36 senses pressure at the flow meter 32.” Therefore, the reference explicitly teaches the pressure being measured is from the closest flow meter.)
PNG
media_image1.png
423
625
media_image1.png
Greyscale
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the leak detection using machine learning system and the supervisory controller reported shutdown events of Streichert in view of Koshinaka further in view of Wang with the selection of pressure data acquired from the pressure sensor that is closest to the flow sensor of Hunaidi. The motivation for doing so would be to ensure the pressure data correlates accurately with the flow data, as Hunaidi teaches that pressure is not constant along the length of the pipe but rather varies based on location. “Discrepancy in predicted leak location is believed to be due to variation of acoustic velocity along the pipe… negative pressure in the pipe… is believed to be highest near the free end… and becomes less severe in the direction of the pumping station” (Hunaidi, Para 64)
Claims 9-10 are rejected under 35 U.S.C. 103 as being unpatentable over patent application US 2023/0088241 A1 Streichert et al., hereinafter “Streichert” in view of patent application US 2003/0187595 A1 Koshinaka et al., hereinafter “Koshinaka” further in view of patent application US 2020/0201950 A1 Wang et al., hereinafter “Wang” further in view of non-patent literature Ebin John Daniel (“ANALYZING COMPRESSED AIR DEMAND TRENDS TO DEVELOP A METHOD TO CALCULATE LEAKS IN A COMPRESSED AIR LINE USING TIME SERIES PRESSURE MEASUREMENTS”, hereinafter “Daniel”)
Claim 9
Streichert in view of Koshinaka further in view of Wang teaches all the limitations of claim 1, Daniel teaches:
further comprising estimating an amount of leaked compressed air by: (Page 26, “By utilizing the pressure difference measured across the piping… a continuous leak measurement algorithm can be developed to account for the volumetric loss of air, Qleaks, through said leaks.”) receiving a plurality of air flow rates measured in the compressed air system for a plurality of corresponding pressure levels, (Page 30, “The manufacturer of the ultrasound gun will provide a data table showing the instantaneous volumetric flow rate of the compressed air in relation to the decibel readings. The ultrasound gun used for the purposes of this paper is provided in Table 2.2 [40]. The inner cells of Table 2.2 are the instantaneous volumetric flow rates in m3/min determined by the corresponding decibel reading from the first column, to the corresponding pressure from the first row.”) computing a flow to pressure relation based on the plurality of measured air flow rates and the plurality of corresponding pressure levels, the flow to pressure relation is indicative of the at least one compressed air leak which induces a pressure dependent air flow rate, and (Page 50, “As discussed in section 2.4, the volumetric flow rate of the air through the piping is dependent on the source pressure and end pressure, the relationship was found to be as shown in equation 2.18.”)
PNG
media_image2.png
70
570
media_image2.png
Greyscale
inferring the amount of leaked compressed air based on the flow to pressure relation which is indicative of a pressure dependent air flow rate induced by the at least one air leak. (Page 53, “Figure 3.1, shows the instantaneous leak rates, qleaks, calculated at every second using equation 2.18 and compared to the total air rate consumed”)
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the leak detection using machine learning system and the supervisory controller reported shutdown events of Streichert in view of Koshinaka further in view of Wang with estimation of leaked air by computing a flow to pressure relation of Daniel. The motivation for doing so would be to obtain a more accurate estimate of the air loss by accounting for the pressure dependent nature of the leaks, as Daniel teaches that a benefit of utilizing such a relationship is that “the variations in the volumetric flow rate of air lost through the orifices will be accounted for which has usually been overlooked in traditional methods and provide a higher degree of accounting detail” (Daniel, page 23)
Claim 10
Streichert in view of Koshinaka further in view of Wang further in view of Daniel teaches all the limitations of claim 9, Streichert further teaches:
further comprising actively controlling the at least one air compressor to deliver compressed air to induce the plurality of pressure levels in the compressed air system. (Para 19, “In this case, the throttle valve and/or another suitable controller can be used to regulate the supply pressure of the supply lines, resulting in different travel times and/or reaction times with otherwise constant components.” – EN: this denotes using a “controller” to “regulate the supply pressure” to create different operating states (pressure levels). Under BRI, regulating supply pressure via a controller to induce specific levels maps to “actively controlling” the air supply/compressor system to induce those levels.)
Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over patent application US 2023/0088241 A1 Streichert et al., hereinafter “Streichert” in view of patent application US 2003/0187595 A1 Koshinaka et al., hereinafter “Koshinaka” further in view of patent application US 2020/0201950 A1 Wang et al., hereinafter “Wang” further in view of non-patent literature Ma et al. (“Digital twin and big data-driven sustainable smart manufacturing based on information management systems for energy-intensive industries”, hereinafter “Ma”)
Claim 12
Streichert in view of Koshinaka further in view of Wang teaches all the limitations of claim 11, Streichert further teaches:
wherein the at least one trained ML model is further trained to… (As established in claim 11, Streichert teaches the use of a machine learning model trained to learn specific operating states)
Koshinaka further teaches:
“…at least one potential shutdown event detected…” and a flow rate threshold associated with each operational state (Para 27, “The deactivated state refers to a state where all of the air-driven devices 31-33 are stopped, or deactivated.” Para 32, “The above three-step determination process is performed using two master flow rates, i.e., a first master flow rate M1, which is used as the reference value during the deactivated state, and a second master flow rate M2, which is used as the reference value during the activated state.” Para 72, “The master flow rate (i.e., the second master flow rate) used in the activated state is preferably greater than the master flow rate (i.e., the first master flow rate) used in the deactivated state.” Para 73, “With the above arrangement, it is easy to categorize the operational states of the air-driven devices, and it is easy to identify the current operational state of the air-driven devices.” – EN: the deactivated state corresponds to the “potential shutdown event” detected as set forth in the rejections of claims 1 and 11 above, and the master flow rates constitute flow rate thresholds (“a certain threshold”) that characterize and distinguish the operational states, the deactivated (shutdown) state being associated with the lower flow rate level.)
Streichert in view of Koshinaka further in view of Wang does not explicitly disclose:
“…filter out [at least one potential shutdown event detected] while the air flow rate in the compressed air system exceeds a certain threshold.”
However, Ma teaches:
“…filter out [at least one potential shutdown event detected] while the air flow rate in the compressed air system exceeds a certain threshold.” (Section 5.2.2, “The air compressor stopped working between 16:00 to 6:00 the next day. [valid shutdown event - air consumption stopped as part of normal operation cycle, in this case it was due to shift rotation] … However, during the three nights… especially on the night of March 25, which showed a large amount of air consumption. [exceeding threshold]… Did the company require overtime production?” – EN: this denotes analyzing a standard “shutdown” window (16:00-6:00). When the system detects that the air flow is too high (“large amount of consumption”), it does not treat that period as a shutdown, instead it “filters” it from the shutdown category by identifying it as “overtime production”. Ma is relied upon only for this filtering determination; the detection of the potential shutdown events and the flow rate thresholds are taught by Koshinaka as set forth above, and the trained ML model performing the filtering is taught by Streichert.)
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to train the leak detection machine learning model of Streichert, which identifies the shutdown (deactivated) events of Koshinaka as set forth in the rejection of claim 11 above, to filter out potential shutdown events whose air flow rate exceeds a threshold as taught by Ma. The motivation for doing so would be to ensure the accuracy of leak detection analysis by preventing the system from combining real production usage (such as overtime) with leakage during expected shutdown windows. Ma explicitly teaches analyzing flow rates during these windows to identify anomalies, noting that “a large amount of air consumption” (Ma, section 5.2.2) during a standard shutdown time prompted the determination of “overtime production” rather than a leak. Ma Section 5.2.2, “However, during the three nights from 16:00 on March 25 to 6:00 on March 28, compressed air was used erratically on the 2nd floor, especially on the night of March 25, which showed a large amount of air consumption. Did the company require overtime production?”
Claims 14-15 are rejected under 35 U.S.C. 103 as being unpatentable over patent application US 2023/0088241 A1 Streichert et al., hereinafter “Streichert” in view of patent application US 2003/0187595 A1 Koshinaka et al., hereinafter “Koshinaka” further in view of patent application US 2020/0201950 A1 Wang et al., hereinafter “Wang” further in view of non-patent literature Ravichandran et al. (“Ensemble-based machine learning approach for improved leak detection in water mains”, hereinafter “Ravichandran”)
Claim 14
Streichert in view of Koshinaka further in view of Wang teaches all the limitations of claim 1, Streichert further teaches:
“wherein the at least one trained ML model is trained to identify at least one of the plurality of compressed air leak patterns using a plurality of training samples comprising pressure data and flow rate data” (Para 6-10, “Provide a trained normal state model as a one-class model that has been trained in a training phase with normal state data representing a normal state of the pneumatic system; Continuous acquisition of sensor data of the pneumatic system by means of a set of sensors; Extract features from the acquired sensor data; Determine deviations of extracted features from learned features of the normal state model using a distance metric…; Calculate an anomaly score from the determined deviations and Output of the calculated anomaly score.” Para 49, “the normal state data may comprise pressure signals and/or flow signals” Para 112, “features are extracted by means of an extractor 304, i.e. quantities are derived that provide information about the functioning of the pneumatic system… include the actuator features “reaction time extension”, “travel time extension”, “reaction time retraction”, and “travel time retraction”” – EN: this denotes the trained ML model is applied to sensor data, which includes “pressure signals and/or flow signals”, to perform “leakage detection” (para 5). The model operates by extracting specific features from this data, such as “reaction time extension” and “travel time extension”, and then “determining deviations” of these extracted features from learned norms. These determined deviations constitute the “leak patterns” because they are distinct data signatures used to identify the presence of a leak.) “measured in at least one (Para 17, “For example, a pneumatic system can be a single pneumatic actuator or a plurality of actuators. The plurality of actuators can be operated independently of each other. A plurality of actuators may be arranged on a valve island that can control multiple valves at once.” Para 98, “FIG. 1 shows an overview illustration of a pneumatic system 100 with a condition monitoring unit 114” – Examiner’s note: this denotes a pneumatic system that is equivalent to the broadest reasonable interpretation of “compressed air system” because the pneumatic system comprises actuators that function as the required “compressed air clients” consuming air delivered via valves and supply lines.) “not having compressed air leaks” (Para 14, “The normal condition model describes the normal condition of the pneumatic system. In the normal state, there are no anomalies, leaks or other faults, and the production process is functioning perfectly.”)
Streichert in view of Koshinaka further in view of Wang does not explicitly disclose:
“ML model is trained (…) in at least one another (…) system having at least one (…) leak” -- (Examiner’s note: Streichert does not teach the ML model being trained on data from other systems having leaks.)
However, Ravichandran teaches:
“ML model is trained (…) in at least one another (…) system (Abstract, “The training and validation data sets have been collected over several months from multiple cities across North America.”) having at least one (…) leak” (Page 316 - Case Study, “The operational data set includes 13,861 negative samples and 54 leak cases observed over 3 months.”)
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the leak detection using machine learning system and the supervisory controller reported shutdown events of Streichert in view of Koshinaka further in view of Wang to train the ML model by using leak data from other systems as disclosed by Ravichandran. The motivation for doing so would be in order “For adequate performance, the training data set must… ensure sufficient diversity in the data set.” (Page 310, Ravichandran).
Claim 15
Streichert in view of Koshinaka further in view of Wang teaches all the limitations of claim 1, Streichert further teaches:
“wherein the at least one trained ML model is trained to identify at least one of the plurality of compressed air leak patterns using a plurality of training samples comprising pressure data and flow rate data measured in the at least one compressed air system…” (EN: This limitation is similar to claim 14 with the difference being “in the… system”. Streichert teaches this in Para 19, “Experiments have shown that the characteristics of the normal state can depend strongly on the settings of a throttle between a valve and an actuator. A normal state model can therefore be specific to an actuator.” – By stating that the model must be “specific to” the throttle of an actuator, this denotes that the training data must originate from that specific system rather than a different one.) “…while not having compressed air leaks.” (Para 14, “The normal condition model describes the normal condition of the pneumatic system. In the normal state, there are no anomalies, leaks or other faults, and the production process is functioning perfectly.”)
Streichert in view of Koshinaka further in view of Wang does not explicitly disclose:
“ML model is trained … in … system while having at least one … leak”
However, Ravichandran teaches:
“ML model is trained … in … system while having at least one … leak” (Page 316 - Case Study, “The operational data set includes 13,861 negative samples and 54 leak cases observed over 3 months.”)
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art to combine the leak detection using machine learning system and the supervisory controller reported shutdown events of Streichert in view of Koshinaka further in view of Wang to train the model with leak data as disclosed by Ravichandran. The motivation for doing so would be in order “For adequate performance, the training data set must… ensure sufficient diversity in the data set.” (Page 310, Ravichandran).
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to NAYMUR RAHMAN ALI whose telephone number is (571)272-0007. The examiner can normally be reached Mon-Fri. 9:30-6:30 pm.
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, Alexey Shmatov can be reached at (571)270-3428. 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.
/NAYMUR RAHMAN ALI/Examiner, Art Unit 2123
/ALEXEY SHMATOV/Supervisory Patent Examiner, Art Unit 2123
Prosecution Insights