DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Status of Claims
The office action is being examined in response to the application filed by the Applicant on October 30, 2023.
Claims 1-19 are pending and have been examined.
This action is made NON-FINAL.
The Examiner would like to note that this application is now being handled by examiner Ivonnemary Rivera González.
Foreign Priority
Acknowledgment is made of applicant’s claim for foreign priority under 35 U.S.C. 119 (a)-(d). The certified copy has been filed in parent Application No. CN202311192558X, filed on September, 15, 2023.
Receipt is acknowledged of certified copies of papers required by 37 CFR 1.55.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on December 4, 2023 and January 28, 2026 are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
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.
Claims 6 and 14 - 16 are 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. Because of the following claims and their element features have insufficient antecedent basis for:
Claims 14 and 16: recites "… the first abnormal probability and the second abnormal probability…" , in the last lines, respectively. However, the first and second abnormal probabilities are not claimed in claim 1 from which claim 14 depends upon. Rather, is firstly and later introduced in claim 4.
Claims 5 and 15: recites "…the outlier threshold…" , in the last lines, respectively. However, this outlier threshold is not claimed in claim 2 from which claim 5 depends upon and neither claim 1 recites this element from which claim 2 depends upon. Similarly, claim 15 which depends upon claim 14 that depends from claim 1 does not recite this element feature either.
Thus, there is insufficient antecedent basis for these limitations in their respective claims as the element features recited above were not previously claimed. Finally, the claim language is not clear. Therefore, the scope of what is claimed is not clear for these reasons and is considered to be indefinite. Also, it's not clear whether the Applicant intended claims 14-16 to depend upon Claim 4 for at least the “the first abnormal probability” and the “second abnormal probability” features. For purposes of examination, the Examiner is interpreting these set of claims as being dependent to claim 1 and have their claimed element features already disclosed in at least independent claim 1. For the same reasons stated above, claims 6 and 15 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ) based on their dependency to claims 5 and 16, respectively.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 17 – 18 are rejected under 35 U.S.C. 101 because the claimed invention is directed to non-statutory subject matter. The claim(s) do not fall within at least one of the four categories of patent eligible subject matter because the "Smart gas Internet of Things (IoT) system" disclosed that further includes a "smart gas device management platform" (from claim 17) does not fall within at least one of the four categories of patent eligible subject matter recited in 35 U.S.C. 101 (process, machine, manufacture, or composition of matter) as they are directed to "software per se" considered a "computer program" that is claimed as "a product without any structural recitations" and expresses software "as code or a set of instructions detached from any medium" which is an "idea without physical embodiment" ("software per se", see MPEP § 2106, and 2106.03, subsection I, respectively). Further, the specifications fail to limit the BRI of “smart gas device management platform” including its sub-components, and disavow the structure of the platform being hardware such as at least a “terminal device” or other “hardware devices” since it can also encompass and/or be implemented as “software-only solutions, such as an installation on an existing server or mobile device” (see ¶0020 and ¶0159, respectively; Figure 1 from Applicant disclosure). Thus, an amendment to claims 17 and 18 is suggested to clarify and specify at least the hardware structure that the “smart gas device management platform” is executed/configured by the hardware component of “a terminal device”. Finally, claim 18 is rejected due to its dependency to claim 17.
Claims 1 - 19 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more, although the claim(s) mentioned above failed step 1, all claims were further evaluated for the purpose of compact prosecution. The analysis of this claimed invention recited in the claims begins in view of independent claim 1, the most representative claim of the independent claims set 1, 17 and 19, as follows:
At Step 1: Claims 1 - 16 falls under statutory category of a process, while claims 17 - 18 are directed to a machine and claim 19 is considered an article of manufacture. Of note, the Examiner interpreted, under BRI and based on Applicant disclosure, that the “smart gas device management platform” is implemented in a “terminal device”, that based on its plain meaning is a “hardware device” with a concrete structure of at least a physical interface with keyboard/monitor or screen (Refer to ¶0020 and ¶0156 from Applicant disclosure). See MPEP § 2106 subsection I and § 2106.03.
At Step 2A Prong 1: Claim 1 (representative of claims 17 and 19) recites an abstract idea in the following limitations:
…obtaining a user feature and a pipeline network transportation feature of each of a plurality of gas users;
obtaining a first clustering result and a second clustering result by clustering the gas user based on the user feature and the pipeline network transportation feature respectively, the first clustering result and the second clustering result including one or more gas user clusters, respectively;
for any one of the gas user clusters:
determining, based on device use data and/or gas metering data of the gas user in the gas user cluster, a potential abnormal gas user; wherein the device use data includes…and a gas usage..., and the gas metering data include a cumulative gas usage value of a plurality of moments; and the potential abnormal gas user includes a first abnormal user and a second abnormal user;
determining a target abnormal user based on the first abnormal user and the second abnormal user, the first abnormal user being the potential abnormal gas user determined based on the first clustering result, and the second abnormal user being the potential abnormal gas user determined based on the second clustering result; and
sending an early warning message to the target abnormal user.
Generally, and as disclosed in the specification in ¶0004, this claimed invention provides “a method, an Internet of Things (IoT) system, and a storage medium for a smart gas abnormal data analysis for quickly and accurately determining an anomalous user and performing a timely warning.” However, the abstract idea(s) of mental processes are/is recited in claim 1 as the limitation steps can be practically be performed in the human mind or in pen and paper (See MPEP 2106.04(a)(2), subsection III). Specifically, the abstract idea is recited in the steps of “…clustering the gas user based on the user feature and the pipeline network transportation feature”, “determining…a potential abnormal gas user”, and “determining a target abnormal user based on the first abnormal user and the second abnormal user…”. Because clustering or grouping user and gas pipeline features to determine potential and target abnormal gas users at least encompasses observations, evaluations and judgements. Also, these steps can either be done with the help of physical aid such as pen and paper or can be performed by humans without or with the assistance (e.g. tool) a computer. Thus, the steps do not negate and further still reads in the mental nature of the limitation(s), when clustering and determining such user/pipeline network transportation information and potential/target abnormal gas user, as well as the concept is merely claimed to be performed on a generic computer and is merely using a computer as a tool to perform the concept of sending early warning messages to target abnormal user(s) based on their gas usage among other features (see MPEP 2106.04(a)(2)(III)(B & C)).
At Step 2A Prong 2: For independent claims 1, 17 and 19, The judicial exception(s) or abstract idea previously identified is not integrated into a practical application (see MPEP 2106.04 (d)). The claims recite the additional element(s) of a smart gas device management platform of a smart gas Internet of Things (IoT) system and a gas device (from claims 1 and 17); and non-transitory computer-readable storage medium (from claim 19). These additional elements, individually and in combination, and while considering the claims as a whole, are merely used as a tool to perform the abstract idea (See MPEP 2106.05(f)). Specifically, these steps are recited as being performed by a computer. The computer is recited at a high level of generality that is being used as a tool to perform the generic computer functions for clustering or grouping user and gas pipeline features data to determine potential/target abnormal gas users to send them an early warning message. Thus, these steps mentioned above are further describing and applying the abstract idea without placing any limits on how the technological components are being improved, while distinguishing in the claim language, the performing limitations from functions that generic computer components can perform.
Finally, the step of “sending an early warning message to the target abnormal user” in the representative claim is really nothing more than links to computer for implementing the use of ordinary capacity for economic or other tasks (e.g., to receive, store, or transmit data) or simply adding a general-purpose computer or computer components (refer to MPEP 2106.05 f (2)). Thus, in this limitation step, the computer is used to perform an abstract idea, as discussed above in Step 2A, Prong One, such that it amounts to no more than mere instructions to apply the exception using a generic computer.
Step 2B: For independent claims 1, 17 and 19, these claims do not provide an inventive concept. The recited additional elements of the claim(s) are the following: a smart gas device management platform of a smart gas Internet of Things (IoT) system and a gas device (from claims 1 and 17); and non-transitory computer-readable storage medium (from claim 19). These additional elements are not sufficient to amount significantly more than the judicial exception or abstract idea (see MPEP 2106.05). Because, as indicated in Step 2A Prong 2, these additional element(s) claimed are merely, instructions to “apply” the abstract ideas, which cannot provide an inventive concept. Thus, even when considered in combination, these additional elements represent mere instructions to implement an abstract idea or other exception on a computer, which do not provide an inventive concept at Step 2B.
For dependent claims 2-16 and 18, the same analysis is incorporated. Due to their dependency to the independent claims analyzed, these claims cover or fall under the same abstract idea(s) of mental processes. They describe additional limitations steps of:
Claims 2-16 and 18: further describes the abstract idea of the method for smart gas abnormal data analysis and how determining potential abnormal gas users includes: the generation of a plurality of histogram distributions to determine/count outlier users, different clustering parameters, frequency of times the user satisfies a number of preset conditions and outlier thresholds, weighting outlier degrees to determine abnormal probabilities, inputs including outlier user distribution map with nodes and edges, calculating reference correlation coefficients to further weight sub-difference correlations based on actual correlation coefficients and gas users’ clustering results, clustering parameters. Thus, being directed to the abstract idea group of mental processes as these steps require observations, evaluations and judgements. But also, these determining/calculating steps encompass mathematical concepts that can be performed mentally or in pen and paper and requires specific mathematical calculations.
Step 2A Prong 2 and Step 2B: For dependent claims 7 – 8 and 18, these claims recite the additional elements of: the prediction model being a machine learning model (from claim 7); a gas metering device (from claim 8); and a smart gas user platform, a smart gas service platform, a smart gas sensing network platform, and a smart gas object platform; a gas user sub-platform, a government user sub-platform, and a supervision user sub-platform; a smart gas use service sub-platform, a smart operation service sub-platform, and a smart supervision service sub-platform; a smart gas indoor device parameter management sub-platform, a smart gas pipeline network device parameter management sub-platform, and a smart gas data center; a device operation parameter monitoring and warning module and a device parameter remote management module; a device operation parameter monitoring and warning module and a device parameter remote management module; a smart gas indoor device sensing network sub-platform and a smart gas pipeline network device sensing network sub-platform and a smart gas indoor device object sub-platform and a smart gas pipeline network device object sub-platform (from claim 18). These additional elements recited are invoking computers merely used as a tool to perform or “apply” the abstract idea(s) to the existing process of obtaining/transmitting gas user and gas pipeline network transportation features, data gas metering data as well as warning messages. Thus, amounting to no more than mere instructions to “apply” the exception using a generic computer component (MPEP 2106.05(f) and (f)(2)). As for the machine learning model used to determine abnormal users, this ML model is also recited in a high level of generality that does not limits how the “determining” step is distinctively achieved from utilizing other general ML models or computer technology. See MPEP 2106.05(f). Accordingly, for the same reasons stated above, these additional element(s) claimed cannot provide an inventive concept at Step 2B.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-7 and 9 - 19 are rejected under 35 U.S.C. 103 as being unpatentable over Siebel (U.S. Patent No. 11886843 B2) in view of Gas-Theft Suspect Detection Among Boiler Room Users: A Data-Driven Approach (referred to as Xiuwen hereafter by the Examiner).
Regarding claims 1 and 17:
This independent claim set is represented by claim 1.
Siebel teaches:
obtaining a user feature and a pipeline network transportation feature of each of a plurality of gas users; (In C19 – 20; L48 – 67 and L1 – 7; Fig. 4 (402): teaches that the “signal data acquisition module 402 can be configured to determine a set of signals and associated signal values for the set of signals” which are associated “with a plurality of energy usage conditions” wherein “a specific set of signal values can be associated with a current state of a specific utility meter for a specific customer at a specific location and venue”, in accordance to features definitions and examples given in ¶0056 – 57 from Applicant disclosure. Also, refer to C14 – 15; L55 – 67 and L1 – 46 wherein the “set of signals” or “signal values” can also reflect “various conditions of energy usage” that “may relate to, for example, types of energy usage, states of energy usage, amounts of energy usage, readings of energy usage from meters, operating status of meters, states of customer accounts with energy providers, and any other considerations that directly or indirectly reflect energy provision, usage, availability, and payments” as well as each signal can “reflect a particular energy usage condition” that “can be associated not only with meters that measure the usage, but also associated with customer information, locational information, venue types, dates and times, etc.”. Further, “data received from data sources can include, but is not limited to, AMI systems data (meter data management and head end data), customer information data, customer consumption data, billing information, contract information, meter event information, outage management system (OMS) data, producer generation, workorder management (WOM) data, verified theft and malfunction data, weather, and geographic localization” which comes from “grid and utility operational systems, meter data management (MDM) systems, customer information systems (CIS), billing systems, utility customer systems, utility enterprise systems, utility energy conservation measures, rebate databases, building characteristic systems, weather data sources, third-party property management systems, industry-standard benchmark databases, etc.” )
obtaining a first clustering result and a second clustering result by clustering the gas user based on the user feature and the pipeline network transportation feature respectively, the first clustering result and the second clustering result including one or more gas user clusters, respectively; (In C21; L4 – 31; Fig. 4 (406); Fig. 6; Fig. 7 (706): teaches that “machine learning module 406 can be configured to receive or acquire new signals values for the set of signals” wherein the module can “classify the new N-dimensional representation based on the classifier model” and “at least one machine learning algorithm can facilitate mapping, based on signal values, at least some N-dimensional representations to non-technical loss”. For example, “if the classifier model indicates that the new N-dimensional representation is similar to (or sufficiently close in N-dimensional proximity to, near, clustered with, etc.) another representation that has already been classified as corresponding to non-technical loss, then the new N-dimensional representation can be classified as corresponding to non-technical loss as well” which each “N-dimensional representation” is directed to the clustering results. Refer to C23; L28 – 56 for more details about Fig. 6 that illustrates different clusters or “N-dimensional representations”.)
for any one of the gas user clusters: determining, based on device use data and/or gas metering data of the gas user in the gas user cluster, a potential abnormal gas user; (In C11; L1 – 9; Fig. 3 (304): teaches “integration services module 304 may perform data validation” and “can detect data gaps and data anomalies (e.g., statistical anomalies), identify outliers, and conduct referential integrity checks. Referential integrity checking ensures that data has the correct network of associations to enable analysis and aggregation, such as ensuring that loaded meter data is associated with a facility or, conversely, that facilities have associated meters”, in accordance to device use data definitions and examples given in ¶0064 from Applicant disclosure. Refer to C16; L15 – 25 wherein an example of a signal category is an ““Anomalous Load” signal category can include an “Active Power vs. Reactive Power Curve Analysis” signal, which relates to analyzing active and reactive power data and identifying anomalous patterns that are indicative of theft and/or malfunction. For example, signal values for the “Active Power vs. Reactive Power Curve Analysis” signal can characterize irregular variations in year-over-year consumption patterns for a given customer, which can indicate a likelihood of theft and/or malfunction.” Also, these signals can be can be generated based on a “modification to” or “permutation of” a “second signal in the set of signals” as well as based on “a combination of the second signal and a third signal.” (see C18; L39 – 50))
wherein the device use data includes a gas device and a gas usage of the gas device, and the gas metering data include a cumulative gas usage value of a plurality of moments; and the potential abnormal gas user includes a first abnormal user and a second abnormal user; (In C8; L50 – 67; Fig. 1 (104n); Fig. 3 (302): teaches that “data integrator module 302 accepts data from a broad range of data sources, including grid and operational systems such as MDM, CIS, and billing systems, as well as third-party data sources such as weather databases, building databases (e.g., Urban Planning Council database), third-party property management systems, and external benchmark databases. The imported data may include, for example, meter data (e.g., electricity consumption, water consumption, natural gas consumption) provided at minimum daily or other time intervals (e.g., 15-minute intervals), weather data (e.g., temperature, humidity at daily or other time intervals (e.g., hourly intervals), building data (e.g., square footage, occupancy, age, building type, number of floors, air conditioned square footage), aggregation definitions (hierarchy) (e.g., meters to building, buildings to city block, building's regional identification), and asset data (e.g., number and type of HVAC assets, number and type of production units (for plants)).” For cumulative gas usage, refer to C7; L30 – 33 and C11; L49 – 57 wherein “key/value stores 216 may maintain various data supporting the energy management platform 202. In an embodiment, time series data (e.g., meter readings, meter events, etc.) may be stored in the key/value store”)
sending an early warning message to the target abnormal user. (In C12; L33 – 49; Fig. 3 (306 and 310): teaches that “the stream analytic services module 310 also may include stream processing logic, which can be provided by a user of the energy management platform 102” wherein this logic can “provide a calculated result that can be persisted and used for subsequent analysis” and “provide an alert based on a calculated result”. For example, “a utility may want to receive alerts and on-the-fly analysis when there is an unexpected and significant drop or spike in load”. Thus, “Data about the unexpected load change can be rapidly recognized, analyzed, and used to send the necessary alert.” Also, in C13; L4 – 15, the “stream analytic services module 310” can send “streams of recently generated electricity consumption and demand data” into “an event queue associated with the data services module 306” that can trigger “automatic analytic processes” that can include outputs that may be “alerts and calculations that are then stored in a database and made available to designated end users as analysis results.”)
Siebel does teach the different clustering results as “N-dimensional representation” per “signal category” (see Fig. 6 and C21; L4 – 31; Siebel), such as ““Anomalous Load” signal category which can point to abnormal usage from a “given customer” (see C16; L15 – 25; Siebel). However, Siebel does not explicitly teach the ability of determining a specific target abnormal user based on a first and second abnormal user that are derived from their respective clustering results. Thus, Xiuwen teaches:
determining a target abnormal user based on the first abnormal user and the second abnormal user, the first abnormal user being the potential abnormal gas user determined based on the first clustering result, and the second abnormal user being the potential abnormal gas user determined based on the second clustering result; and (In Section 3.3, ¶1, p.9; Figs. 11 – 14: teaches the use of “deformation-based normality detection algorithm” to exclude “normal boiler rooms from abnormal ones” and for “distinguishing gas-theft suspects from irregular users, as shown in Fig. 11” a “OCSVM based anomaly detection algorithm” is used to “capture multiple characteristic factors from different data sources” (i.e. interpreted as the first and second clustering results per gas user) and “predict the detected abnormal users with the trained OCSVM to differentiate suspected users” based on three categories of features: “boiler room attribute features, gas consumption features, and temperature-gas joint features”, in accordance to target abnormal user definitions and determination examples given in ¶0072 – 77 and ¶0145 – 148 as well as a gas user definition/example in ¶0022 from Applicant disclosure. Further, in Section 3.3.2, ¶1, p.10, the mapping of “daily consumption into 5 intervals” (see equation) wherein “each dimension of the feature means the probability of corresponding interval”. Thus, “boiler room users can be clustered into two groups: the steady and the vibrating one, shown in Fig. 13a. For the steady group, their gas consumption mainly concentrates in the large-value interval while with minor probability in the small-value intervals. While for the vibrating group, they are turned down more often, so their consumption rises and falls.” Refer to Section 5, ¶1, p.18, wherein a “system for a gas group in northern China” conduct inspections based on “detection results during the 2019-2020 heating season” and utilizes the system “GasShield” and its models to obtain percentage values of abnormal users.)
It would have been obvious to one of ordinary skill in the art before the earliest effective filing date of the claimed invention to modify Siebel to provide the ability of determining a specific target abnormal user based on a first and second abnormal user that are derived from their respective clustering results, as taught by Xiuwen in order to “exclude 62 percent boiler rooms as normal users; with the anomaly detection algorithm, we can further detect 21 percent boiler rooms as gas-theft suspects with high recall” (Section 7, ¶1, p.19; Xiuwen), have “potential suspects [that] could be discovered in the early stage with higher accuracy” and to overcome “the label scarcity problem” (Section 1, bullets 3 and 5, p.3; Xiuwen), see also MPEP 2143.I.G.
Regarding claim 19:
Siebel further teaches:
A non-transitory computer-readable storage medium storing computer instructions, wherein when reading the computer instructions in the storage medium, a computer implements the method for smart gas abnormal data analysis of claim 1. (In C5; L5 – 10; Fig. 1 (102); Fig. 2 (202); Fig. 8 (802 and 304): teaches the “energy management platform 102 may be implemented as a computer system, such as a server or series of servers and other hardware (e.g., applications servers, analytic computational servers, database servers, data integrator servers, network infrastructure (e.g., firewalls, routers, communication nodes))”. See C24 - 25; L59 – 67 and L1 – 16 for more details.)
Siebel teaches all the functions from the method for smart gas abnormal data analysis of claim 1 (see claim 1 and 17 mapping above), except determining a specific target abnormal user based on a first and second abnormal user that are derived from their respective clustering results which is taught by Xiuwen as using a “OCSVM based anomaly detection algorithm” to “capture multiple characteristic factors from different data sources” (i.e. interpreted as the first and second clustering results per gas user) and “predict the detected abnormal users with the trained OCSVM to differentiate suspected users” based on three categories of features: “boiler room attribute features, gas consumption features, and temperature-gas joint features” (see Section 3.3, ¶1, p.9 and Figs. 11 – 14; Xiuwen) as well as the mapping of “daily consumption into 5 intervals” so “boiler room users can be clustered into two groups: the steady and the vibrating one, shown in Fig. 13a.” (see Section 3.3.2, ¶1, p.10; Xiuwen).
It would have been obvious to one of ordinary skill in the art before the earliest effective filing date of the claimed invention to modify Siebel to provide the ability of determining a specific target abnormal user based on a first and second abnormal user that are derived from their respective clustering results, as taught by Xiuwen in order to “exclude 62 percent boiler rooms as normal users; with the anomaly detection algorithm, we can further detect 21 percent boiler rooms as gas-theft suspects with high recall” (Section 7, ¶1, p.19; Xiuwen), have “potential suspects [that] could be discovered in the early stage with higher accuracy” and to overcome “the label scarcity problem” (Section 1, bullets 3 and 5, p.3; Xiuwen), see also MPEP 2143.I.G.
Regarding claim 2:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 1.
Siebel does not explicitly teach the abilities of specifically generating histogram distributions for the gas user clusters based on multiple preset gas use features to determine outlier users, count gas users’ times in the cluster as an outlier user and determine a first abnormal user. Thus, Xiuwen further teaches:
wherein the determining, based on device use data and/or gas metering data of the gas user in the gas user cluster, a potential abnormal gas user comprises: for one or more of the gas user clusters in the first clustering result, generating, based on a plurality of preset gas use features, a plurality of histogram distributions respectively; (In Section 3.1, ¶1 – 3, pp.4 – 5; Figs. 3 and 16: teaches that the system “detect such boiler room users having such one type of data quality issue” such as “detect data-missing users” and “zero-consumption users” (i.e. based on “missing data” and “zero consumption” conditions) which such data in illustrated in Figs. 3a – 3b as histograms distributions. See Section 5, ¶1 – 2, p.18 for more details of another histogram counting the suspicious users as shown in Fig. 16.)
for any one of the histogram distributions, determining one or more outlier users in the histogram distribution; counting a number of times for each gas user in the first clustering result being determined as the outlier user in the plurality of histogram distributions; and determining, at least based on the number of times, the first abnormal user in the gas user cluster. (In Section 3.1, ¶3, p.4 and ¶ 1 – 3, p. 5; Figs. 3 and 16: teaches an example based on the Figs. 3a – 3b and its histograms distributions, wherein “the distribution of zero rate appears two obvious plunges, the first one after 10 percent and the second one after 70 percent. The high proportion of zero readings indicate that either longtime continuous or frequent irregular shutdown has occurred. It conflicts with normal operation patterns of boiler rooms and is highly suspicious of stealing gas”. This way, “users whose data zero rate is higher than 70 percent” can be excluded. Then, it detects “users whose gas consumption fluctuates severely” with “spikes that exceed its usual gas consumption level appear in records” to “exclude users whose maximum daily gas consumption is ten times greater than the median” as well as detecting “users whose gas consumption is continuously low” wherein “its gas consumption remains comparatively lower than its former level” indicating “boiler room either operates at a low-temperature level or has fraudulent behaviors to report less gas consumption” which further, “users whose daily gas consumption is lower than half of the maximum for more than 7 days (one week)” are excluded. Finally, “data-deficient (data-missed and data-zero) boiler room users and data-abnormal (severe fluctuations and continuous low consumption) boiler room users can be quickly detected.”)
It would have been obvious to one of ordinary skill in the art before the earliest effective filing date of the claimed invention to modify Siebel to provide the abilities of specifically generating histogram distributions for the gas user clusters based on multiple preset gas use features to determine outlier users, count gas users’ times in the cluster as an outlier user and determine a first abnormal user, as taught by Xiuwen in order to “exclude 62 percent boiler rooms as normal users; with the anomaly detection algorithm, we can further detect 21 percent boiler rooms as gas-theft suspects with high recall” (Section 7, ¶1, p.19; Xiuwen), have “potential suspects [that] could be discovered in the early stage with higher accuracy” and to overcome “the label scarcity problem” (Section 1, bullets 3 and 5, p.3; Xiuwen), see also MPEP 2143.I.G.
Regarding claim 3:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 2.
Siebel further teaches:
wherein a clustering parameter corresponding to the first clustering result includes at least one of a gas device type, a user type, and a monthly usage. (In C19; L15 – 26; Fig. 6: teaches “the non-technical loss identification module 330 can generate, based on the signal values, a plurality of N-dimensional representations (e.g., points in N-dimensional space) for the plurality of energy usage conditions, where N represents the number of signals (i.e., signal quantity) in a set of signals indicative of the presence of non-technical loss. For example, if there are 150 signals, then the N-dimensional representation can have 150 dimensions. Each dimension can correspond to a respective signal. A particular energy usage condition in the plurality of energy usage conditions can be represented as a point in N-dimensional space with coordinates based on the signal values.” Thus, such “signal values” can include “at least one of an account attribute signal category, an anomalous load signal category, a calculated status signal category, a consumption on inactive signal category, a current analysis signal category, a missing data signal category, a disconnected signal category, a meter event signal category, a monthly meter anomalous load signal category, a monthly meter consumption on inactive signal category, an outage signal category, a stolen meter signal category, an unusual production signal category, a work order signal category, or a zero reads signal category” (see C18; L51 – 64). For more details per signal category, refer to C15; L47 – 59 through C18; L39 – 50.)
Regarding claim 4:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 2.
Siebel further teaches:
wherein the determining, at least based on the number of times, the first abnormal user in the gas user cluster comprises: determining the gas user whose number of times satisfies a preset number of times condition as the first abnormal user, (In C22; L28 – 40: teaches that “the results processing module 408 can generate rankings or scores for the identified meters based on their respective likelihoods of being associated with the non-technical loss. In some implementations, the likelihood for an identified meter associated with a particular energy usage condition can depend on an N-dimensional proximity between the representation associated with one energy usage condition and another representation verified as corresponding to non-technical loss. A lesser N-dimensional proximity can indicate a higher likelihood.” Refer to C16; L15 – 25 wherein an example of a signal category is an ““Anomalous Load” signal category can include an “Active Power vs. Reactive Power Curve Analysis” signal, which relates to analyzing active and reactive power data and identifying anomalous patterns that are indicative of theft and/or malfunction. For example, signal values for the “Active Power vs. Reactive Power Curve Analysis” signal can characterize irregular variations in year-over-year consumption patterns for a given customer, which can indicate a likelihood of theft and/or malfunction.”)
and determining a first abnormal probability of the first abnormal user. (In C22; L41 – 52: teaches that the “results processing module 408 can further determine that at least some of the plurality of meters meet specified ranking threshold criteria and can provide the at least some of the plurality of utility meters as candidates for investigation about potential non-technical loss. In one example, the ranking threshold criteria can specify a minimum likelihood percentage amount. In another example, the ranking threshold criteria can specify a quantity having the highest likelihoods. Those ranked meters that satisfy the ranking threshold criteria can be the meters most likely to have encountered non-technical loss, such as due to theft or malfunction”, in accordance to abnormal probability formula and examples given in ¶0102 from Applicant disclosure.)
Regarding claim 5:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 4.
Siebel does not explicitly teach the ability of having specific preset numbers of time conditions that specifically include outlier thresholds related to the outlier degree of the gas user that is determined based on the histogram distribution. However, Xiuwen further teaches:
wherein the preset number of times condition includes an outlier threshold, the outlier threshold being related to an outlier degree of the gas user when the gas user is determined as the outlier user; the outlier degree being determined based on the histogram distribution. (In Section 3.1, ¶3, p.4 and ¶ 1 – 3, p. 5; Figs. 3, 14c and 16: teaches an example based on the Figs. 3a – 3b and its histograms distributions with different outlier degrees and outlier thresholds, wherein “the distribution of zero rate appears two obvious plunges, the first one after 10 percent and the second one after 70 percent. The high proportion of zero readings indicate that either longtime continuous or frequent irregular shutdown has occurred. It conflicts with normal operation patterns of boiler rooms and is highly suspicious of stealing gas”. This way, “users whose data zero rate is higher than 70 percent” can be excluded. Then, it detects “users whose gas consumption fluctuates severely” with “spikes that exceed its usual gas consumption level appear in records” to “exclude users whose maximum daily gas consumption is ten times greater than the median” as well as detecting “users whose gas consumption is continuously low” wherein “its gas consumption remains comparatively lower than its former level” indicating “boiler room either operates at a low-temperature level or has fraudulent behaviors to report less gas consumption” which further, “users whose daily gas consumption is lower than half of the maximum for more than 7 days (one week)” are excluded. Finally, “data-deficient (data-missed and data-zero) boiler room users and data-abnormal (severe fluctuations and continuous low consumption) boiler room users can be quickly detected.” Refer to Section 3.2.3, ¶3, p.8 wherein with “the calculated ShapeVarr, we can set a threshold to judge whether a boiler room is normal or not” and when “ShapeVarr on at least one timestamps surpasses the threshold, the boiler room is judged as an anomaly” and the “threshold of ShapeVarr is set by considering both gas-theft labels and expert experience of the upper bound proportion of boiler room users who can be suspicious of stealing gas.” See Section 3.3.3, pp. 11 – 12 also for more threshold used for abnormal user(s) determinations.)
It would have been obvious to one of ordinary skill in the art before the earliest effective filing date of the claimed invention to modify Siebel to provide the ability of having specific preset numbers of time conditions that specifically include outlier thresholds related to the outlier degree of the gas user that is determined based on the histogram distribution, as taught by Xiuwen in order to “exclude 62 percent boiler rooms as normal users; with the anomaly detection algorithm, we can further detect 21 percent boiler rooms as gas-theft suspects with high recall” (Section 7, ¶1, p.19; Xiuwen), have “potential suspects [that] could be discovered in the early stage with higher accuracy” and to overcome “the label scarcity problem” (Section 1, bullets 3 and 5, p.3; Xiuwen), see also MPEP 2143.I.G.
Regarding claim 6:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 5.
Siebel at least teaches that the “set of signals can grow, shrink, and/or change over time”. For example, “signal quantity of the set of signals may be modified based on machine learning algorithms for classifying energy usage conditions” and “energy providers such as utility companies can create their own signals and provide these signals to the energy management platform 102 to be utilized in addition to or instead of the set of signals determined by the operator of the energy management platform 102” which is related to adjusting the criteria of the signal categories (see C24; L19 – 28; Siebel). Further, Siebel teaches that the “batch parallel processing analytic services module 312 may perform a substantial portion of analysis required by users of the energy management platform 102” such as “outlier analysis” (see C13; L21 – 37; Siebel) Siebel does not explicitly teach the ability of determining an outlier degree of the gas user by specifically weighing the outlier degrees with weighted value determined and being related to the preset gas use feature. However, Xiuwen further teaches:
wherein a determination of the outlier degree of the gas user when the gas user is determined as the outlier user comprises: weighting the outlier degrees when the gas user is determined as the outlier user for more than one time, determining a weighted value as the outlier degree when the gas user is determined as the outlier user, the weighting being related to the preset gas use feature. (In Section 3.3.4., ¶3, p.12: teaches that “with the extracted features, OCSVM adopt the rbf kernel function exp(−γ(∥x−x′∥)2) to learn a decision boundary. It first maps the original features into a high dimensional space corresponding to the kernel function, and then separate them from the original one using a decision boundary, which maximizes the distance from this boundary to the origin [8]. For a new sample of abnormal boiler rooms, if it falls on the same side of the decision boundary where most training data fall, it will be classified as a normal sample, otherwise as an anomaly. The optimization of OCSVM is to solve the quadratic programming problem, where tuning the parameters ν and γ”, which is directed to weighting the outlier degrees by determining a weighted value that is related to the preset gas use feature, in accordance to weights definitions and determination examples given in ¶0097 – 100 from Applicant disclosure. Refer to Section 3.3.4., pp.13 – 14 for parameters settings.)
It would have been obvious to one of ordinary skill in the art before the earliest effective filing date of the claimed invention to modify Siebel to provide the ability of determining an outlier degree of the gas user by specifically weighing the outlier degrees with weighted value determined and being related to the preset gas use feature, as taught by Xiuwen in order to “exclude 62 percent boiler rooms as normal users; with the anomaly detection algorithm, we can further detect 21 percent boiler rooms as gas-theft suspects with high recall” (Section 7, ¶1, p.19; Xiuwen), have “potential suspects [that] could be discovered in the early stage with higher accuracy” and to overcome “the label scarcity problem” (Section 1, bullets 3 and 5, p.3; Xiuwen), see also MPEP 2143.I.G.
Regarding claim 7:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 2.
Siebel does teach the application of “machine learning to the signal values to identify energy usage conditions associated with non-technical loss” wherein the “application of machine learning to the signal values may involve application of at least one machine learning algorithm to the plurality of N-dimensional representations to produce a classifier model for identifying non-technical loss. In some embodiments, the classifier model can be modified, refined, and/or improved over time” (see Fig. 6 and C24; L36 – 44; Siebel). However, Siebel does not explicitly teach the ability of determining a specific first abnormal user with a prediction model that is a ML model. Thus, Xiuwen further teaches:
wherein the determining, at least based on the number of times, the first abnormal user in the gas user cluster comprises: determining, at least based on the number of times, the first abnormal user through a prediction model, the prediction model being a machine learning model. (In Section 3.3., ¶1, p.9: teaches the application of a “One-Class Support Vector Machine” or “OCSVM based anomaly detection algorithm to capture multiple characteristic factors from different data sources” and distinguish “gas-theft suspects from irregular users” which can be “viewed as the pseudo label for designing a self-supervised or one class classification model” (see Section 3.3.4., ¶1 - 2, p.12).)
It would have been obvious to one of ordinary skill in the art before the earliest effective filing date of the claimed invention to modify Siebel to provide the ability of determining a specific first abnormal user with a prediction model that is a ML model, as taught by Xiuwen in order to “exclude 62 percent boiler rooms as normal users; with the anomaly detection algorithm, we can further detect 21 percent boiler rooms as gas-theft suspects with high recall” (Section 7, ¶1, p.19; Xiuwen), have “potential suspects [that] could be discovered in the early stage with higher accuracy” and to overcome “the label scarcity problem” (Section 1, bullets 3 and 5, p.3; Xiuwen), see also MPEP 2143.I.G.
Regarding claim 9:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 1.
Siebel does teach the different clustering results as “N-dimensional representation” per “signal category” (see Fig. 6 and C21; L4 – 31; Siebel), such as ““Anomalous Load” signal category which can point to abnormal usage from a “given customer” (see C16; L15 – 25; Siebel). However, Siebel does not explicitly teach the abilities of calculating a reference correlation coefficient based on gas metering data of a gas user and determining associated user(s) of each gas user and if it is a second abnormal user. Thus, Xiuwen further teaches:
wherein the determining, based on device use data and/or gas metering data of the gas user in the gas user cluster, a potential abnormal gas user comprises: for one of the gas user clusters in the second clustering result, for any two of the gas users in the gas user cluster, calculating, based on the gas metering data of a historical gas user, a reference correlation coefficient; (In Section 3.2, ¶1, p.5; Fig. 5: teaches an example wherein as “shown in Figs. 5a and 5b, with the integrated analysis on gas consumption data and outdoor temperature data, we find that the daily gas consumption is strongly negatively related to the daily outdoor temperature. When it becomes colder, the gas consumption will increase in the upcoming days to offer the external heat supply, and vice versa [3]. Thus, it is important to take the opposite outdoor temperature as a reference. If the gas consumption curve fits the reference well, we can infer that the boiler room is normal. While for the remaining boiler rooms, we can judge them as abnormal users” wherein the “opposite outdoor temperature” derived from the daily gas consumption of gas users is directed to the reference correlation coefficient. Also, refer to Section 3.2.3, ¶1 – 2, pp. 7 – 8 wherein “for each boiler room r, we get a pair of denoised and shifted time series of the daily gas consumption Gr and the opposite daily outdoor temperature Tr. If the curve of daily gas consumption fits the reference curve well, the boiler room can be inferred to be normal. For detecting such normal users, we define a temperature-gas shape variation ShapeVar in the Equation (1). It measures the deformation correlation between the two time series. Its two components Φ(CORT) and Diff are defined in Equations (2) and (5), which characterize the trend consistency and the value deviation respectively” which are directed having at least a reference correlation coefficient(s). Further, the “component Φ(CORT) reflects how severely the gas consumption deviates from the reference, which can represent the trend consistency” and “the component Diff portrays to what extent values of the gas consumption and the temperature diverge, namely their value deviation.”)
determining, based on the reference correlation coefficient, at least one associated user of each gas user in the gas user cluster; and determining, based on the device use data and the gas metering data of the gas user and the associated user of the gas user, whether the gas user is the second abnormal user. (In Section 3.3, ¶1, p.9; Figs. 11 – 14: teaches the use of “deformation-based normality detection algorithm” to exclude “normal boiler rooms from abnormal ones” and for “distinguishing gas-theft suspects from irregular users, as shown in Fig. 11” a “OCSVM based anomaly detection algorithm” is used to “capture multiple characteristic factors from different data sources” (i.e. interpreted as the first and second clustering results per gas user accounting reference correlation coefficient(s)) and “predict the detected abnormal users with the trained OCSVM to differentiate suspected users” based on three categories of features: “boiler room attribute features, gas consumption features, and temperature-gas joint features”. Further, in Section 3.3.2, ¶1, p.10, “the daily consumption into 5 intervals” is mapped, this way, “boiler room users can be clustered into two groups: the steady and the vibrating one, shown in Fig. 13a. For the steady group, their gas consumption mainly concentrates in the large-value interval while with minor probability in the small-value intervals. While for the vibrating group, they are turned down more often, so their consumption rises and falls.” Refer to Section 5, ¶1, p.18, wherein a “system for a gas group in northern China” conduct inspections based on “detection results during the 2019-2020 heating season” and utilizes the system “GasShield” and its models to obtain percentage values of abnormal users.)
It would have been obvious to one of ordinary skill in the art before the earliest effective filing date of the claimed invention to modify Siebel to provide the abilities of calculating a reference correlation coefficient based on gas metering data of a gas user and determining associated user(s) of each gas user and if it is a second abnormal user, as taught by Xiuwen in order to “exclude 62 percent boiler rooms as normal users; with the anomaly detection algorithm, we can further detect 21 percent boiler rooms as gas-theft suspects with high recall” (Section 7, ¶1, p.19; Xiuwen), have “potential suspects [that] could be discovered in the early stage with higher accuracy” and to overcome “the label scarcity problem” (Section 1, bullets 3 and 5, p.3; Xiuwen), see also MPEP 2143.I.G.
Regarding claim 10:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 9.
Siebel further teaches:
wherein a clustering parameter corresponding to the second clustering result includes at least one of a complexity degree of a pipeline, and whether the pipeline belongs to a same branch. (In C19; L15 – 26; Fig. 6: under the broadest reasonable interpretation (BRI) of clustering parameters for second clustering results, teaches “a particular energy usage condition in the plurality of energy usage conditions can be represented as a point in N-dimensional space with coordinates based on the signal values.” Thus, such “signal values” can include “at least one of an account attribute signal category, an anomalous load signal category, a calculated status signal category, a consumption on inactive signal category, a current analysis signal category, a missing data signal category, a disconnected signal category, a meter event signal category, a monthly meter anomalous load signal category, a monthly meter consumption on inactive signal category, an outage signal category, a stolen meter signal category, an unusual production signal category, a work order signal category, or a zero reads signal category” (see C18; L51 – 64). Further, “energy providers such as utility companies can create their own signals and provide these signals to the energy management platform 102 to be utilized in addition to or instead of the set of signals determined by the operator of the energy management platform 102”, which the descriptive matter of pipelines as a clustering parameter can be incorporated as a “signal category” along with its features of pipeline belonging to a same branch and complexity degree of the pipeline (see C24; L19 – 28). Thus, the “signal values” can be provided to the “signal data acquisition module 402” that can “determine signal values for the set of signals, such as by applying a set of formulas for the set of signals” that correspond to and are “derived or developed from research, analysis, observation, experimentation, etc.” wherein “each condition of energy usage can be represented by one or more respective signal values. For example, a specific set of signal values can be associated with a current state of a specific utility meter for a specific customer at a specific location and venue” which is another example of a clustering parameter or “signal category” and its value including specific “energy usage conditions” that further include formulated values such as pipeline’s branch and complexity degree of the pipeline (see C20; L1 – 7). For more details per signal category, refer to C15; L47 – 59 through C18; L39 – 50.)
Regarding claim 11:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 9.
Siebel does not explicitly teach the abilities of obtaining an actual correlation coefficient, determine a sub-difference between actual correlation coefficient and a corresponding reference correlation coefficient, obtain a composite difference by weighting gas user’s sub-differences and satisfying a preset difference condition to finally determine a second abnormal user and calculate a second abnormal probability of the second abnormal user. However, Xiuwen further teaches:
wherein the determining, based on the device use data and the gas metering data of the gas user and the associated user of the gas user, whether the gas user is the second abnormal user comprises: for one of the gas users, obtaining an actual correlation coefficient between the gas user and the associated user; determining a sub-difference between the actual correlation coefficient and a corresponding reference correlation coefficient; (In Section 3.3.3, ¶1, p.11; Table 1 Fig. 14: teaches under BRI, that the extraction of “six temperature-gas joint features listed in the third category of Table 1. The Distribution of ShapeVar feature describes the ratio of abnormal days changing with the incremental threshold. As illustrated in Fig. 14a, for normal users, their ShapeVar seldom exceeds thresholds. While for abnormal ones, their ShapeVar deviates from the normal level more severely” wherein the ShapeVar deviation is interpreted as the sub-difference between the actual correlation coefficient and a corresponding reference correlation coefficient.)
obtaining a composite difference by weighting a plurality of sub-differences of the gas user; and in response to the composite difference satisfying a preset difference condition, (In Section 3.3.3, ¶2, p.11; Fig. 14: teaches under BRI, that the construction of the “Dynamic Time Warping (DTW) distance measures the similarity between two time series, where the larger the DTW distance is, the less similar the two time series are. The feature DTW from Gr to Tr for the gas consumption Gr and the opposite outdoor temperature Tr, its distribution can be seen in Fig. 14c. Apart from the sole ΔGr, we also calculate the normalized daily average temperature difference, denoted by ΔTr. Considering the variation dependency between them, the features Mean of Δ Gr/ΔTr and STD of ΔGr/ΔTr are extracted. Their co-distribution and that of ΔGr are alike. It reveals that, for the majority of boiler rooms, the variation of gas consumption obeys consistent laws, which is tightly associated with that of temperature”, in accordance to the composite difference definition given in ¶0135 from Applicant disclosure.)
determining that the gas user is the second abnormal user, and calculating a second abnormal probability of the second abnormal user. (In Section 3.3.3, ¶2, p.11; Fig. 14: teaches that, “with the extracted features, OCSVM adopt the rbf kernel function exp(−γ(∥x−x′∥)2) to learn a decision boundary” which is directed to calculating the second abnormal probability. It first maps the original features into a high dimensional space corresponding to the kernel function, and then separate them from the original one using a decision boundary, which maximizes the distance from this boundary to the origin [8]. For a new sample of abnormal boiler rooms, if it falls on the same side of the decision boundary where most training data fall, it will be classified as a normal sample, otherwise as an anomaly”, in accordance to the second abnormal probability definition given in ¶0142 from Applicant disclosure.)
It would have been obvious to one of ordinary skill in the art before the earliest effective filing date of the claimed invention to modify Siebel to provide the abilities of obtaining an actual correlation coefficient, determine a sub-difference between actual correlation coefficient and a corresponding reference correlation coefficient, obtain a composite difference by weighting gas user’s sub-differences and satisfying a preset difference condition to finally determine a second abnormal user and calculate a second abnormal probability of the second abnormal user, as taught by Xiuwen in order to use “an OCSVM based anomaly detection algorithm to capture multiple characteristic factors from different data sources” wherein these characteristics belong to boiler rooms (i.e. gas users) which are derived from three categories of features, that are further used to train the OCSVM model with “positive samples” and be able to “predict the detected abnormal users with the trained OCSVM to differentiate suspected users” (Section 3.3, ¶1, p.9; Xiuwen). But also, in order to have “potential suspects [that] could be discovered in the early stage with higher accuracy” and to overcome “the label scarcity problem” (Section 1, bullets 3 and 5, p.3; Xiuwen), see also MPEP 2143.I.G.
Regarding claim 12:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 11.
Siebel does not explicitly teach the ability of having a weighing process wherein the sub-difference weight is positively correlated to the reference correlation coefficient. However, Xiuwen further teaches:
wherein in a weighting process, a weight of the sub-difference is positively correlated to the reference correlation coefficient. (In Section 3.3.3, ¶1, p.11; Table 1, Fig. 14: teaches an example wherein the extraction of “six temperature-gas joint features listed in the third category of Table 1. The Distribution of ShapeVar feature describes the ratio of abnormal days changing with the incremental threshold. As illustrated in Fig. 14a, for normal users, their ShapeVar seldom exceeds thresholds. While for abnormal ones, their ShapeVar deviates from the normal level more severely. The higher the threshold is, the less abnormal is detected. The co-distribution of features Mean of ShapeVar and STD of ShapeVar are displayed in Fig. 14b, which is scattered symmetrically. The more gas consumption deviates from the normal level, the larger STD and the absolute value of the mean of ShapeVar are. Therefore, the STD, along with the larger positive Mean, is higher than that with the negative ones”. Examiner notes that the reference correlation coefficient was interpreted as the normal level of the threshold related to temperature (i.e. Temperature – Gas Joint Features; see table 1) and the weight of the sub-difference with a positive correlation is the STD (i.e. deviation) being larger, similar to the more gas consumption deviation which indicates the positive correlations between these two values.)
It would have been obvious to one of ordinary skill in the art before the earliest effective filing date of the claimed invention to modify Siebel to provide the ability of having a weighing process wherein the sub-difference weight is positively correlated to the reference correlation coefficient, as taught by Xiuwen in order to use “an OCSVM based anomaly detection algorithm to capture multiple characteristic factors from different data sources” wherein these characteristics belong to boiler rooms (i.e. gas users) which are derived from three categories of features, that are further used to train the OCSVM model with “positive samples” and be able to “predict the detected abnormal users with the trained OCSVM to differentiate suspected users” (Section 3.3, ¶1, p.9; Xiuwen). But also, in order to have “potential suspects [that] could be discovered in the early stage with higher accuracy” and to overcome “the label scarcity problem” (Section 1, bullets 3 and 5, p.3; Xiuwen), see also MPEP 2143.I.G.
Regarding claim 13:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 11.
Siebel does teach the different clustering results as “N-dimensional representation” per “signal category” or features (see Fig. 6 and C21; L4 – 31; Siebel), such as ““Anomalous Load” signal category which can point to abnormal usage from a “given customer” (see C16; L15 – 25; Siebel). However, Siebel does not explicitly teach the ability of having a weight of the sub-difference correlated to the first clustering result of the gas user and the associated user. Thus, Xiuwen further teaches:
wherein in a weighting process, a weight of the sub-difference is correlated to the first clustering result of the gas user and the associated user. (In Section 3.3.3, ¶1, p.11; Table 1 Fig. 14: teaches the reference correlation coefficient directed to the normal level of the threshold related to temperature (i.e. Temperature – Gas Joint Features; see table 1) and the weight of the sub-difference is the STD (i.e. deviation) that are from the “six temperature-gas joint features” extracted and “listed in the third category of Table 1”, but that are also clustered as shown in Fig. 14b through the function of ShapeVar. See Sections 3.2.3, p.8 and Section 3.3, ¶1, p.9 also.)
It would have been obvious to one of ordinary skill in the art before the earliest effective filing date of the claimed invention to modify Siebel to provide the ability of having a weight of the sub-difference correlated to the first clustering result of the gas user and the associated user, as taught by Xiuwen in order to use “an OCSVM based anomaly detection algorithm to capture multiple characteristic factors from different data sources” wherein these characteristics belong to boiler rooms (i.e. gas users) which are derived from three categories of features, that are further used to train the OCSVM model with “positive samples” and be able to “predict the detected abnormal users with the trained OCSVM to differentiate suspected users” (Section 3.3, ¶1, p.9; Xiuwen). But also, in order to have “potential suspects [that] could be discovered in the early stage with higher accuracy” and to overcome “the label scarcity problem” (Section 1, bullets 3 and 5, p.3; Xiuwen), see also MPEP 2143.I.G.
Regarding claim 14:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 1.
Siebel does teach the different clustering results as “N-dimensional representation” per “signal category” (see Fig. 6 and C21; L4 – 31; Siebel), such as ““Anomalous Load” signal category which can point to abnormal usage from a “given customer” (see C16; L15 – 25; Siebel). However, Siebel does not explicitly teach the abilities of determining a specific target abnormal user based on a first and second abnormal user that are derived from their respective clustering results and determine their respective first and second abnormal probabilities satisfying a preset probability condition. Thus, Xiuwen teaches:
wherein the determining a target abnormal user based on the first abnormal user and the second abnormal user comprises: determining a user belonging to both the first abnormal user and the second abnormal user as a candidate abnormal user; and (In Section 3.3, ¶1, p.9; Fig. 11: teaches the use of an “OCSVM based anomaly detection, which discovers gas-theft suspects among abnormal users considering various gas usage characteristics” (see Section 1, ¶6, p.2). Specifically, it is used, a “deformation-based normality detection algorithm” to exclude “normal boiler rooms from abnormal ones” and for “distinguishing gas-theft suspects from irregular users, as shown in Fig. 11” a “OCSVM based anomaly detection algorithm” is used to “capture multiple characteristic factors from different data sources” (i.e. interpreted as the first and second clustering results per gas user) and “predict the detected abnormal users with the trained OCSVM to differentiate suspected users” based on three categories of features: “boiler room attribute features, gas consumption features, and temperature-gas joint features”.)
determining, based on the candidate abnormal user, the target abnormal user, the first abnormal probability and the second abnormal probability of the target abnormal user satisfying a preset probability condition. (In Section 4.2.3, ¶1 - 2, pp.15 – 16; Figs. 10 – 12: teaches a summary of the “TGSV algorithm can first tell normal users apart from abnormal ones, then RF, GBDT, MLP, and VAE can be trained with normal samples and predict on abnormal ones after that” by comparing the “OCSVM with several typical classifiers. RF, GBDT, and MLP will distinguish the normal and abnormal ones, while abnormal ones contain many users with irregular patterns” wherein the first and second abnormal probabilities are already included (see Table 6 for the first abnormal probability and Section 3.3.3, ¶2, p.11 for the second abnormal probability; see pending claim 11 also) and are further defined by the precision (PR) and recall (RC) metric parameters (see Section 4.1.4, p.14). As for the first and second probabilities satisfying a preset condition, such meeting preset condition was interpreted as the model’s ShapeVar function that measures “deformation correlation between the two time series” which requires that the “ShapeVarr on at least one timestamps surpasses the threshold” for the boiler room user be judged as an anomaly and such threshold is “set by considering both gas-theft labels and expert experience of the upper bound proportion of boiler room users who can be suspicious of stealing gas” as shown in Fig. 10 (see Section 3.2.3, ¶1 - 2, p.8). Refer to Section 5, ¶1, p.18, wherein a “system for a gas group in northern China” conduct inspections based on “detection results during the 2019-2020 heating season” and utilizes the system “GasShield” and its models to obtain percentage values of target abnormal users interpreted as the “confirmed gas-thefts”.)
It would have been obvious to one of ordinary skill in the art before the earliest effective filing date of the claimed invention to modify Siebel to provide the abilities of determining a specific target abnormal user based on a first and second abnormal user that are derived from their respective clustering results and determine their respective first and second abnormal probabilities satisfying a preset probability condition, as taught by Xiuwen in order to “exclude 62 percent boiler rooms as normal users; with the anomaly detection algorithm, we can further detect 21 percent boiler rooms as gas-theft suspects with high recall” (Section 7, ¶1, p.19; Xiuwen), have “potential suspects [that] could be discovered in the early stage with higher accuracy” and to overcome “the label scarcity problem” (Section 1, bullets 3 and 5, p.3; Xiuwen), see also MPEP 2143.I.G.
Regarding claim 15:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 14.
Siebel does not explicitly teach the ability of having a preset probability condition that includes a first preset probability that is related to an outlier threshold and a difference threshold. Thus, Xiuwen teaches:
wherein the preset probability condition includes a first preset probability, the first preset probability being related to at least one of the outlier threshold and a difference threshold. (In Section 3.2.3, ¶1 - 2, p.8; Figs. 10 - 12: teaches the that the model’s ShapeVar function that measures “deformation correlation between the two time series” which requires that the “ShapeVarr on at least one timestamps surpasses the threshold” for the boiler room user be judged as an anomaly and such threshold is “set by considering both gas-theft labels and expert experience of the upper bound proportion of boiler room users who can be suspicious of stealing gas” as shown in Fig. 10 wherein this example is directed to the outlier threshold and the difference threshold claimed. Refer to Section 4.1.2., p.14 for more ShapeVar parameter settings. Examiner notes that such preset probability condition defining non-functional descriptive matter does not hold patentable weight since one of ordinary skilled in the art would determine based on the experimental/observed and industrial data the preset probability conditions that would best fit and accurately define the probability to differentiate irregular users from suspected ones. See MPEP 2111.05 (I) (A and B) for non-functional descriptive material and MPEP 2144.04 (IV) for supporting rationale based on changes in size, shape or sequence of adding ingredients and 2144.05 (I) for obviousness of similar and overlapping ranges amounts and proportions.)
It would have been obvious to one of ordinary skill in the art before the earliest effective filing date of the claimed invention to modify Siebel to provide the ability of having a preset probability condition that includes a first preset probability that is related to an outlier threshold and a difference threshold, as taught by Xiuwen in order to “exclude 62 percent boiler rooms as normal users; with the anomaly detection algorithm, we can further detect 21 percent boiler rooms as gas-theft suspects with high recall” (Section 7, ¶1, p.19; Xiuwen), have “potential suspects [that] could be discovered in the early stage with higher accuracy” and to overcome “the label scarcity problem” (Section 1, bullets 3 and 5, p.3; Xiuwen), see also MPEP 2143.I.G.
Regarding claim 16:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 15.
Siebel does not explicitly teach the ability of having the preset probability condition including a probability summation value that is greater than the first preset probability and the probability summation value being a weighted summation value of the first and second abnormal probabilities. Thus, Xiuwen teaches:
wherein the preset probability condition includes a probability summation value being greater than the first preset probability, the probability summation value being a weighted summation value of the first abnormal probability and the second abnormal probability. (In Section 4.1.2., p.14; Figs. 10 - 12: teaches under BRI, that “for normality detection, the size of sliding window is set to 3. The threshold of TGShapeVar is set to 0.64 uniformly for all the three datasets, based on both gas-theft labels and expert knowledge on the proportion of suspicious users”, in accordance to the probability summation definition given in ¶0153 – 155 from Applicant disclosure. Examiner notes that such preset probability condition further defining more non-functional descriptive matter does not hold patentable weight since one of ordinary skilled in the art would determine the statistical format based on the experimental/observed and industrial data preset probability conditions that would best fit and accurately define the probability to differentiate irregular users from suspected ones. See MPEP 2111.05 (I) (A and B) for non-functional descriptive material and MPEP 2144.04 (IV) for supporting rationale based on changes in size, shape or sequence of adding ingredients and 2144.05 (I) for obviousness of similar and overlapping ranges amounts and proportions.)
It would have been obvious to one of ordinary skill in the art before the earliest effective filing date of the claimed invention to modify Siebel to provide the ability of having the preset probability condition including a probability summation value that is greater than the first preset probability and the probability summation value being a weighted summation value of the first and second abnormal probabilities, as taught by Xiuwen in order to “exclude 62 percent boiler rooms as normal users; with the anomaly detection algorithm, we can further detect 21 percent boiler rooms as gas-theft suspects with high recall” (Section 7, ¶1, p.19; Xiuwen), have “potential suspects [that] could be discovered in the early stage with higher accuracy” and to overcome “the label scarcity problem” (Section 1, bullets 3 and 5, p.3; Xiuwen), see also MPEP 2143.I.G.
Regarding claim 18:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 17.
Siebel further teaches:
further comprising a smart gas user platform, a smart gas service platform, a smart gas sensing network platform, and a smart gas object platform that interact in sequence; (In C5; L5 – 20; Fig. 1 (102, 106 and 104n); Figs 2 – 3: teaches “energy management platform 102 may be implemented as a computer system, such as a server or series of servers and other hardware (e.g., applications servers, analytic computational servers, database servers, data integrator servers, network infrastructure (e.g., firewalls, routers, communication nodes)). The servers may be arranged as a server farm or cluster”. Also, as “another example, embodiments of the present disclosure may be implemented by a combination of servers of the energy management platform 102 and a computer system of the enterprise 106” which is directed to the claimed platforms.)
the smart gas service platform is configured to send the early early warning message to the smart gas user platform; (In C12; L35 – 49; Fig. 1 (102); Fig. 2 (212); Fig. 3 (300): teaches that the 202 system includes “applications servers 212 of FIG. 2 [that] may be implemented with applications server 300 of FIG. 3” (see C8; L12 – 13) which further includes “stream analytic services module 310 also may include stream processing logic, which can be provided by a user of the energy management platform 102” and “may provide an alert based on a calculated result. For example, a utility may want to receive alerts and on-the-fly analysis when there is an unexpected and significant drop or spike in load. This load variation could be caused by a malfunctioning piece of equipment or sudden damage to equipment, and could possibly represent great risk to the distribution system or an end customer. Data about the unexpected load change can be rapidly recognized, analyzed, and used to send the necessary alert.”)
the smart gas object platform is configured to obtain a gas user feature, a gas pipeline network transportation feature, the device use data and the gas metering data, and transmit the gas user feature, the gas pipeline network transportation feature, the device use data and the gas metering data to the smart gas device management platform via the smart gas sensing network platform; (In C8; L45 – 48; Fig. 1 (102); Fig. 3 (302 and 303): teaches that the applications server 300’s “data integrator module 302 is a tool for automatically importing data maintained in software systems or databases of the external data sources 1041-n into the energy management platform 102 of FIG. 1”.)
wherein the smart gas user platform includes a gas user sub-platform, a government user sub-platform, and a supervision user sub-platform; the smart gas service platform includes a smart gas use service sub-platform, a smart operation service sub-platform, and a smart supervision service sub-platform; (In C6; L33 – 44; Fig. 1 (102 and 106): teaches, based on “embodiments of the present disclosure may be implemented by a combination of servers of the energy management platform 102 and a computer system of the enterprise 106” (see C5; L16 – 20), that “enterprise 106 may represent a user (e.g., customer) of the energy management platform 102. The enterprise 106 may include any private or public concern, such as large companies, small and medium businesses, households, individuals, governing bodies, government agencies, non-governmental organizations, nonprofits, etc. The enterprise 106 may include energy providers and suppliers (e.g., utilities), energy service companies (ESCOs), and energy consumers. The enterprise 106 may be associated with one or many facilities distributed over many geographic locations. The enterprise 106 may be associated with any purpose, industry, or other type of profile.”)
the smart gas device management platform includes a smart gas indoor device parameter management sub-platform, a smart gas pipeline network device parameter management sub-platform, and a smart gas data center, wherein the smart gas indoor device parameter management sub-platform includes a device operation parameter monitoring and warning module and a device parameter remote management module, and the smart gas pipeline network device parameter management sub-platform includes a device operation parameter monitoring and warning module and a device parameter remote management module; the smart gas sensing network platform includes a smart gas indoor device sensing network sub-platform and a smart gas pipeline network device sensing network sub-platform; and the smart gas object platform includes a smart gas indoor device object sub-platform and a smart gas pipeline network device object sub-platform. (In C5; L21 – 32; Fig. 1 (102, 104 and 106): teaches that the “energy management platform 102 may communicate with the external data sources 1041-n through APIs and other communication interfaces” (see C4; L61 – 67), but also, “each of the energy management platform 102, the external data sources 1041-n, and the enterprise 106 may be implemented as a computer system” (see C5; L62 – 64). Thus, as part of the system, “external data sources 1041-n may represent a multitude of possible sources of data relevant to energy management analysis. The external data sources 1 041-n may include, for example, grid and utility operational systems, meter data management (MDM) systems, customer information systems (CIS), billing systems, utility customer systems, utility enterprise systems, utility energy conservation measures, and rebate databases. The external data sources 1 041-n also may include, for example, building characteristic systems, weather data sources, third-party property management systems, and industry-standard benchmark databases.”)
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over Siebel (U.S. Patent No. 11886843 B2) in view of Xiuwen in further view of A knowledge-enhanced graph-based temporal-spatial network for natural gas consumption prediction (referred to as Du hereafter by the Examiner).
Regarding claim 8:
The combination of Siebel and Xiuwen, as shown in the rejection above, discloses the limitations of claim 7.
Siebel teaches the first and second clusters and considers the “allowable distance” that a “representation 610” can be within a “cluster 612” to classify the representation as being “associated with NTL (or, alternatively, normal energy usage)” (see C23; L45 – 56 and Fig. 6; Siebel) as well as the gas user environment through the calculation of “signal values” and their “set of formulas” and their association with current meter states based on “specific location and venue.” (see C20; L1 – 7 and C15; L11 – 20; Siebel). Also, Xiuwen teaches the determining number of times the first abnormal user is in the gas user cluster as the histogram distribution example that describe a boiler room user, wherein a “distribution of zero rate appears” and the “high proportion of zero readings indicate that either longtime continuous or frequent irregular shutdown has occurred” as well as other operation patterns of boiler rooms and their gas consumption (see Section 3.1, ¶3, p.4 and ¶ 1 – 3, p. 5; Figs. 3 and 16; Xiuwen). However, neither Siebel or Xiuwen explicitly teach the ability of having prediction model inputs that include an outlier user distribution map with nodes and edges that represent the gas users features and their environment and relationships based on distance. However, Du teaches:
wherein an input of the prediction model includes an outlier user distribution map; a node of the outlier user distribution map corresponds to the gas user determined as the outlier user, and a node feature of the node includes the number of times the gas user being determined as the outlier user, an environment where the gas user is located, and historical maintenance data of a gas metering device of the gas user; and an edge of the outlier user distribution map corresponds to a gas pipeline between the gas users, the edge feature of the edge includes a distance between the gas users. (In C1; ¶1, p.7; Figs. 6 – 7: teaches the construction of a “knowledge-based spatial graph” with “nodes' features and topology information” as shown in Fig. 6 wherein “the pipeline network nodes are considered graph nodes, and the pipeline segments are considered edges. The node features of pipeline network nodes at each time state t can be processed as a feature matrix, where is the number of features for each node. In that way, the spatial dependency patterns that exist among different nodes can be represented as a directed graph, where represents the nodes and is the edges”, in accordance to outlier user distribution map definition and example given in Fig. 4, ¶0107 – 77 and ¶0114 – 115 from Applicant disclosure. Further, in C2; ¶2, p.7, “given the graph constructed according to the natural gas pipeline network and consumption data, the propagation module learns the node's feature and updates the status based on neighbouring pipeline network nodes. Among them, the aggregator acquires the potential representations of each node by extracting feature information from neighbouring pipeline network nodes. Then the status update of each pipeline network node is implemented.”)
It would have been obvious to one of ordinary skill in the art before the earliest effective filing date of the claimed invention to modify Siebel and Xiuwen to provide the ability of having prediction model inputs that include an outlier user distribution map with nodes and edges that represent the gas users features and their environment and relationships based on distance, as taught by Du in order to obtain “spatial dependency patterns among neighbouring pipeline network nodes and “the output spatial features can be applied for corresponding tasks” (C2; ¶2, p.7; Du). But also, because “it is greatly reasonable to incorporate the topological structure of the pipeline network with domain knowledge of pipeline operation for a more accurate prediction of natural gas consumption” (C2; ¶2, p.6; Du), see also MPEP 2143.I.G. Further such ability provided by Du into Siebel and Xiuwen systems would have been obvious because the claimed invention is merely applying a known technique to a known method ready for improvement to yield predictable results. See KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 406 (2007). In other words, all of the claimed elements were known in the prior art and one skilled in the art could have combined the elements as claimed by known methods with no change in their respective functions, and the combination would have yielded nothing more than predictable results to one of ordinary skill in the art at the time of the invention (i.e., predictable results are obtained by applying the known technique of mapping nodes and edges with knowledge-based graphs that relates to abnormal gas users as outlier user distribution for input into a prediction ML model to predict and differentiate abnormal users based on their gas consumption and proximity to the pipeline network and distance to other gas users that might be normal gas users). See also MPEP § 2143(I)(D).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Nottrott (U.S. Patent No. 10962437 B1) is pertinent because it “relates to systems and methods for detecting gas leaks such as methane leaks.”
Geuens (U.S. Pub No. 20220057048 A1) is pertinent because it is an “invention [that] is intended to be able to quantify leaks that occur in a gas network.”
Finkel (U.S. Pub No. 20170244726 A1) is pertinent because it “relates to a system and method for generating data compatible with an external system in an oil and gas asset supply chain, and in particular to an interface and interface method with a secure intermediary platform for generating secure and verifiable data to prevent tampering, or injection of unwanted data resulting from an unauthorized access along a supply chain.”
Ueki (U.S. Pub No. 20100330515 A1) is pertinent because it “relates to a gas shutoff device for controlling so as to limit use of a gas appliance at the CO occurring time using a CO alarm and a gas meter and in particular to a gas shutoff device for determining the CO gas leakage appliance according to an output signal from a CO alarm and securing safety.”
Silva (U.S. Pub No. 20230266195 A1) is pertinent because it “relates generally to gas leaks and, more particularly, to the quantification of gas leaks.”
Tao et. all, Intelligent Urban Sensing for Gas Leakage Risk Assessment (17 April 2023) is pertinent because it aims to “revolutionize the existing leakage detection system via the development of intelligent leakage risk assessment system for gas pipelines.”
Zheng, An online real-time estimation tool of leakage parameters for hazardous liquid pipelines (December 2020) is pertinent because it “establishes a pipeline digital twin model that simulates a pipeline leak to generate leakage data.”
Wu, Graph Neural Networks for Anomaly Detection in Industrial Internet of Things (02 July 2021) is pertinent because it “provides a useful investigation on graph neural networks (GNNs) for anomaly detection in IIoT-enabled smart transportation, smart energy, and smart factory.”
Zhang, Towards deep probabilistic graph neural network for natural gas leak detection and localization without labeled anomaly data (17 June 2023) is pertinent because it “proposes a deep probabilistic graph neural network in which attention-based graph neural network is built to model spatial sensor dependency.”
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Ivonnemary Rivera Gonzalez whose telephone number is (571)272-6158. The examiner can normally be reached Mon - Fri 9:00AM - 5:30PM.
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, Nathan Uber can be reached at (571) 270-3923. 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.
/IVONNEMARY RIVERA GONZALEZ/Examiner, Art Unit 3626
/NATHAN C UBER/Supervisory Patent Examiner, Art Unit 3626
Prosecution Insights