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 .
Detailed Action
The following action is in response to the communication(s) received on 12/17/2025.
As of the claims filed 12/17/2025:
No claims are amended.
Claims 1-20 are pending.
Claims 1, 15, and 19 are independent claims.
Response to Arguments
Applicant’s arguments filed 12/17/2025 have been fully considered, but are not fully persuasive.
With respect to the art rejections under 35 USC § 103, Applicant’s arguments have been fully considered but were unpersuasive.
Applicant asserts the following (p.8 ¶2):
PNG
media_image1.png
306
638
media_image1.png
Greyscale
Examiner respectfully disagrees. The broadest reasonable interpretation of “timeseries of the fault labels” does not require the timeseries of the fault labels to be a continuous stream of timeseries fault labels. Additionally, the Specification regarding the input to the conditional generator does not require the fault labels to be a continuous timeseries stream; the generator merely is configured to receive fault labels as an input to generate the synthetic data ([0013] “Labels for the plurality of fault types are used as inputs to the conditional generator” [0110] “The fault labels indicate moments in the historical data corresponding to equipment faults…at such moments”]; [0112] “receive fault labels as an input”; [0126] “ideal labels”). In other words, the conditional generator disclosed by the Applicant are merely data points that are part of timeseries data and which indicate specific faults. Yan teaches that the conditional generator receives input from the dataset sample containing fault data (Yan [p.3 2nd col last ¶] In the generation phase, CWGAN-GP uses a limited number of real fault samples to generate a large number of fault samples.), while Dan teaches more explicitly that the same dataset that Yan uses (ASHRAE RP-1043) for the input data is timeseries data (Dan [p.44 last ¶]). The generated fault samples are conditioned on the limited number of fault samples, thus corresponding to the conditional generator disclosed by Yan.
Applicant further asserts that Dan does not teach that the fault labels can be timeseries data. Examiner respectfully submits that in view of the reasons above, the combination does teach this limitation.
Applicant further asserts that the dependent claims should be allowed for the reasons given above regarding the fault labels being timeseries data. Examiner respectfully submits that in view of the reasons above, the combination does teach this limitation.
Claim Rejections - 35 USC § 103
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.
Claims 1, 5, 7-11, 15, and 17-20 are rejected under 35 U.S.C. 103 as being unpatentable over Yan, “Chiller Fault Detection and Diagnosis with Anomaly Detective Generative Adversarial Network” (hereinafter Yan) in view of Dan, “Fault Detection and Diagnosis for Chillers and AHUs of Building ACMV Systems” (hereinafter Dan), further in view of Yoon et al., “Time-series Generative Adversarial Networks” (hereinafter Yoon), further in view of Drees et al. US 8731724 B2 (hereinafter Drees)
Regarding Claim 1, Yan teaches:
A method for predicting faults in building equipment and initiating responsive actions, (Yan [Abstract] Data augmentation is one of the necessary steps in the process of automated data-driven fault detection and diagnosis (FDD)… the proposed GAN-based chiller FDD framework with GANomaly achieves the highest FDD accuracy than all compared methods.)
the method comprising: receiving a first amount of historical … data comprising fault labels for a plurality of fault types; (Yan [p.3 2nd col last ¶] In the generation phase, CWGAN-GP uses a limited number of real fault samples to generate a large number of fault samples. [p.8 1st col 2nd ¶] After generating a large number of synthetic fault samples from a small amount of real-world fault data for chillers, two new synthetic data evaluation models are proposed to select the high-quality synthetic samples for better performance of chiller FDD.) (Note: real-world fault data corresponds to the historical data comprising fault labels)
training a conditional generator by operating a generative adversarial network trained to represent … historical… data corresponding to a timeseries of the fault labels, the generative adversarial network comprising the conditional generator; (Yan [p.2 2nd col 2nd ¶] The generative adversarial network (GAN) consists of a generator and a discriminator that are essentially neural networks [37]. The generator converts the input random noise to the synthetic data samples under the supervision of discriminator.
[p.2 2nd col last ¶] In 2014, Mirza et al. proposed CGAN (Conditional Generative Adversarial Network), which added conditional variables to the training process of GAN to guide the convergence process between the generator and the discriminator...
[p.3 2nd col 2nd ¶] In this study, we combine CGAN with WGAN-GP to generate synthetic faulty samples improving the traditional supervised learning chiller FDD framework. The combined neural network is named as conditional Wasserstein generative adversarial network with gradient penalty (CWGAN-GP).
[p.3 1st col 1st ¶]
PNG
media_image2.png
230
656
media_image2.png
Greyscale
[p.4 2nd col 2nd to last ¶]
PNG
media_image3.png
270
648
media_image3.png
Greyscale
)
generating, using the conditional generator, a second amount of synthetic… data… (Yan [p.4 2nd col 3rd ¶] In this paper, CWGAN-GP is utilized to generate a large number of synthetic fault samples. The two evaluation models are used to evaluate and filter out the high-quality generated synthetic samples) (Note: the generated large number of synthetic fault samples corresponds to the second amount of synthetic data)
corresponding to the timeseries of the fault labels, wherein the time series of the labels are used as inputs to the conditional generator, and wherein the second amount is greater than or equal to the first amount;
(Yan [p.3 2nd col 2nd ¶] In this study, we combine CGAN with WGAN-GP to generate synthetic faulty samples improving the traditional supervised learning chiller FDD framework
[p.3 2nd col last ¶] In the generation phase, CWGAN-GP uses a limited number of real fault samples to generate a large number of fault samples. After a series of iterative training, the final generator will generate a large group of synthetic faulty samples, n samples of each fault type in every severity level. Perform the same operation as above for levels 2, 3, and 4.
In the evaluation phase, for the n generated synthetic samples introduced from the CWGAN-GP model at each severity level, every type of fault trains its own evaluation model (a VAE or a GANomaly model) and selects k high-quality synthetic samples.
[p.8 1st col 2nd ¶] After generating a large number of synthetic fault samples from a small amount of real-world fault data for chillers, two new synthetic data evaluation models are proposed to select the high-quality synthetic samples for better performance of chiller FDD.) (Note: training an evaluation model for every type of fault corresponds to using the plurality of fault types as inputs to the conditional generator; the large number of fault samples using the small amount of real-world samples corresponds to the second data being a greater amount than the first amount.)
training a fault prediction model using the synthetic… data; (Yan [p.2 2nd col 2nd ¶] Both synthetic data samples and the real-world data samples are inputted into the discriminator for the discriminator training process.) (Note: the discriminator is part of the fault prediction model, thus the discriminator training process corresponds to training a fault prediction model.)
predicting a fault for building equipment by applying the fault prediction model to real … data relating to the building equipment; (Yan [p.5 1st col 1st ¶] The real-world data samples of the same fault type is used for testing. For each data group DG, if many of the real-world samples are evaluated as anomaly with the reconstruction errors, the DG is considered as low-quality data group. Otherwise, DG is considered high-quality. The number of the anomalies is used as a standard to measure the quality level of each DG.)
Yan does not explicitly teach that the received historical data are timeseries data. However, Dan teaches that the chiller dataset is timeseries data: (Dan [p.44 last ¶] Given that the proposed strategy outputs FDD results for a bunch of monitoring data points, and the raw data of ASHRAE RP1043 are collected every 10 seconds
[p.45 1st ¶] and testing data sets are picked from raw data files of RP-1043 with set size varying from 1 point to 714 successive points (represents data collecting time from 10 seconds to 119 minutes).
[p.75 ¶1] Following the ASHRAE RP-1043, the cooling system studied in this chapter is a typical centrifugal water-cooled chiller system with motor-driven compressor [33]. The experimental chiller fault data has been introduced in Chapter 3, which will not be repeated here. Note that since the sensor measurements are time series…)
Dan and Yan are analogous to the present invention because both are from the same field of endeavor of detecting fault codes. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to implement Dan’s timeseries dataset to Yan’s proposed model. The motivation would be to “detect and diagnose system and component failures that undermine energy efficiency.” (Dan [p.8 last ¶]).
Yan/Dan does not teach, but Yoon further teaches a generative adversarial network trained to represent the temporal dynamics of the historical timeseries data…; generating, using the conditional generator, a second amount of synthetic timeseries data having the temporal dynamics… (Yoon [Abst] A good generative model for time-series data should preserve temporal dynamics, in the sense that new sequences respect the original relationships between variables across time… We propose a novel framework for generating realistic time-series data that combines the flexibility of the unsupervised paradigm with the control afforded by supervised training) (Note: generating realistic time-series data corresponds to representing and generating synthetic data having temporal dynamics using historic timeseries data)
Yoon and Yan/Dan are analogous to the present invention because both are from the same field of endeavor of generative adversarial networks for generating synthetic data. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to implement the generative model method from Yoon to Yan/Dan’s proposed model. The motivation would be to “adhere to the dynamics of the training data during sampling.” (Yoon [Abst]).
Regarding the limitation timeseries of the fault labels, Yan, via Yan/Dan/Yoon, further teaches the method of generating timeseries of the fault labels (Yan [p.3 1st col 1st ¶]
PNG
media_image2.png
230
656
media_image2.png
Greyscale
[p.4 2nd col 2nd to last ¶]
PNG
media_image3.png
270
648
media_image3.png
Greyscale
) (Note: since the generated data is timeseries data, as described above, the labels for the data corresponds to be timeseries of fault labels.)
Yan/Dan/Yoon does not teach, but Drees teaches:
and initiating an automated action in response to predicting the fault for the building equipment. (Drees [5] The method also includes generating an output signal indicative of a chiller fault in response to the identification of the significant deviation from the expected performance.) (Generating an output signal indicative of a fault corresponds to initiating an automated action.)
Yan/Dan/Yoon and Drees are analogous to the present invention because both are from the same field of automated fault detection. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate output signals from Drees into Yan/Dan’s fault detection algorithm. The motivation would be to “allow[] for faults to be presented relative to non-fault equipment or data” (Drees [279]).
Regarding Claim 5, Yan/Dan/Yoon/Drees respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Yan/Dan/Yoon/Drees, via Yan, further teaches:
The method of Claim 1, comprising using, by the generative adversarial network, a reconstructed loss, (Yan [p.4 1st col last ¶] (10)
PNG
media_image4.png
39
177
media_image4.png
Greyscale
x, x′ represent the original samples and synthetic samples respectively) (Note: L-con corresponds to the reconstruction loss)
a weakly supervised loss, (Yan [p.4 1st col last ¶] (9)
PNG
media_image5.png
53
171
media_image5.png
Greyscale
where h, h′ represent the potential variables encoded by encoder1 and encoder2 respectively) (Note: L_enc corresponds to the weakly supervised loss)
and an unsupervised loss. (Yan [p.4 1st col last ¶] (11)
PNG
media_image6.png
55
254
media_image6.png
Greyscale
f(x) and f(x′) represent the output value of a certain layer of the discriminator for the original sample and the generated sample respectively) (Note: L_adv corresponds to the unsupervised loss.)
Regarding Claim 7, Yan/Dan/Yoon/Drees respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Yan/Dan/Yoon/Drees, via Yan, further teaches:
The method of Claim 1, wherein training the fault prediction model using the synthetic timeseries data comprises ranking a plurality of sets of the synthetic timeseries data and selecting a highest ranked of the plurality of sets of the synthetic timeseries data for use in training the fault prediction model. (Yan [p.5 1st col 1st ¶] For each data group DG, if many of the real-world samples are evaluated as anomaly with the reconstruction errors, the DG is considered as low-quality data group. Otherwise, DG is considered high-quality. The number of the anomalies is used as a standard to measure the quality level of each DG. Among the m groups of DG, the group with the least number of anomalies is selected as the high-quality synthetic samples group.) (Note: considering a DG as low or high quality corresponds to ranking a plurality of sets of the synthetic timeseries data. Selecting the group with the least number of anomalies corresponds to selecting a highest ranked of the plurality of sets.)
Regarding Claim 8, Yan/Dan/Yoon/Drees respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Yan/Dan/Yoon/Drees, via Yan, further teaches:
The method of Claim 7, wherein ranking the plurality of sets of the synthetic timeseries data comprises comparing the plurality of sets of the synthetic timeseries data to historical training data. (Yan [p.5 1st col 1st ¶] For each data group DG, if many of the real-world samples are evaluated as anomaly with the reconstruction errors, the DG is considered as low-quality data group. Otherwise, DG is considered high-quality.)
Regarding Claim 9, Yan/Dan/Yoon/Drees respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Yan/Dan does not teach, but Drees further teaches:
The method of Claim 1, wherein the automated action comprises altering an internal operation of the building equipment to correct, mitigate, or prevent the fault. (Drees [54] The responses to detected or diagnosed faults may include providing an alert message to a user, a maintenance scheduling system, or a control algorithm configured to attempt to repair the fault or to work-around the fault.) (Note: the control algorithm configured to repair the fault corresponds to an automated method of correcting the fault)
Regarding Claim 10, Yan/Dan/Yoon/Drees respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Yan/Dan/Yoon/Drees, via Drees, further teaches:
The method of Claim 1, wherein the automated action comprises altering a load on the building equipment to mitigate or prevent the fault. (Drees [43] For example, the integrated control layer 116 may be configured to assure that a demand response-driven upward adjustment to the setpoint for chilled water temperature … does not result in an increase in fan energy… that would result in greater total building energy use than was saved at the chiller. The integrated control layer 116 may also be configured to provide feedback to the demand response layer 112 so that the demand response layer 112 checks that constraints… are properly maintained... The constraints may also include setpoint or sensed boundaries relating to safety, equipment operating limits and performance, comfort, fire codes, electrical codes, energy codes, and the like.) (Note: adjusting the setpoints for chilled water of the chiller based on safety corresponds to altering a load on the building equipment to prevent the fault.)
Regarding Claim 11, Yan/Dan/Yoon/Drees respectively teaches and incorporates the claimed limitations and rejections of Claim 1. Yan/Dan/Yoon/Drees, via Drees, further teaches:
The method of Claim 1, wherein the automated action comprises performing maintenance on the building equipment to mitigate or prevent the fault. (Drees [54] The responses to detected or diagnosed faults may include providing an alert message to a user, a maintenance scheduling system, or a control algorithm configured to attempt to repair the fault or to work-around the fault.)
Independent Claim 12 recites a One or more non-transitory computer-readable media storing program instructions that, when executed by one or more processors, cause the one or more processors to execute operations comprising (Yan [p.4 1st col last ¶] And the generator and discriminator are trained alternately until the end of iteration.) to perform precisely the methods of Claim 1. Thus, Claim 12 is rejected for reasons set forth in Claim 1. (Note: training a generator requires a computing system comprising one or more non-transitory computer-readable media with instructions that are to be executed.)
Claims 16 and 17, dependent on 12, also recite the One or more non-transitory computer-readable media configured to perform precisely the methods of Claims 7 and 8, respectively. Thus, Claims 16 and 17 are rejected for reasons set forth in Claims 7 and 8, respectively.
Regarding Claim 18, Yan/Dan/Yoon/Drees respectively teaches and incorporates the claimed limitations and rejections of Claim 12. Yan/Dan/Yoon/Drees, via Drees, further teaches:
The one or more non-transitory computer-readable media of Claim 12, wherein the operations further comprise mitigating or preventing the fault by altering an internal operation of the building equipment, altering a load on the building equipment, or causing maintenance to be performed on the building equipment. (Drees [43] For example, the integrated control layer 116 may be configured to assure that a demand response-driven upward adjustment to the setpoint for chilled water temperature … does not result in an increase in fan energy… that would result in greater total building energy use than was saved at the chiller. The integrated control layer 116 may also be configured to provide feedback to the demand response layer 112 so that the demand response layer 112 checks that constraints… are properly maintained... The constraints may also include setpoint or sensed boundaries relating to safety, equipment operating limits and performance, comfort, fire codes, electrical codes, energy codes, and the like.) (Note: adjusting the setpoints for chilled water of the chiller based on safety corresponds to altering a load on the building equipment to prevent the fault.)
Independent Claim 19 recites train a conditional generator by operating a generative adversarial network comprising the conditional generator; generate, using the conditional generator, synthetic timeseries data corresponding to a plurality of fault types, wherein labels for the plurality of fault types are used as inputs to the conditional generator; train a fault prediction model using the synthetic timeseries data; and predict a fault for building equipment by applying the fault prediction model to real timeseries data relating to the building equipment. which are precisely the methods of Claim 1. Thus, Claim 19 is rejected for reasons set forth in Claim 1 for these limitations under Yan in view of Dan, further in view of Drees. Regarding the rest of the limitations, the Yan/Dan does not teach, but Drees further teaches:
A system, comprising: a unit of building equipment (Drees [31] Each of building subsystems 128 includes any number of devices, controllers, and connections for completing its individual functions and control activities. For example, HVAC subsystem 140 may include a chiller… for controlling the temperature within a building.)
and computing hardware communicable with the unit of building equipment (Drees (32) In an exemplary embodiment, the smart building manager 106 is configured to include a communications interface… (33) FIG. 1B illustrates a more detailed view of smart building manager 106, according to an exemplary embodiment. In particular, FIG. 1B illustrates smart building manager 106 as having a processing circuit 152. Processing circuit 152 is shown to include a processor 154 and memory device 156. [34] Communications interfaces 107, 109 can be or include wired or wireless interfaces (e.g., jacks, antennas, transmitters, receivers, transceivers, wire terminals, etc.) for conducting data communications with, e.g., smart grid 104, energy providers and purchasers 102, building subsystems 128, or other external sources via a direct connection or a network connection (e.g., an Internet connection, a LAN, WAN, or WLAN connection, etc.).)
Yan/Dan and Drees are analogous to the present invention because both are from the same field of endeavor of automated fault detection. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate the smart building manager and building subsystem from Drees into Yan/Dan’s fault detection algorithm. The motivation would be to “reduce energy waste, extend equipment life, or assure proper control response.” (Drees [54]).
Regarding Claim 20, Yan/Dan/Yoon/Drees respectively teaches and incorporates the claimed limitations and rejections of Claim 19. Yan/Dan/Yoon/Drees, via Drees, further teaches:
The system of Claim 19, wherein the computer hardware is further programmed to alter an operation of the unit of building equipment in response to a prediction of the fault. (Drees [199] Embodiments disclosed herein provide systems and methods for continuously and automatically evaluating chiller performance. A chiller performance model is used to detect potential faults of a chiller (e.g., a water-cooled centrifugal chiller). Significant deviation of the actual chiller performance from the modeled performance is used as a general indicator of a fault. A building operator or chiller mechanic may then perform a follow-up investigation to validate the fault, identify root causes for the fault, and to initiate repairs. The model-based approach described herein may be implemented in a chiller controller, in an FDD layer as described above, or in another system or device that is configured to retrieve data from the building automation system or chiller.) (Note: initiating repairs of the chiller indicating a fault corresponds to altering the operation of a unit of building equipment in response to a prediction of the fault.)
Claims 2, 3, 4, 13, and 14 are rejected under 35 U.S.C. 103 as being unpatentable over Yan/Dan/Yoon/Drees further in view of Kruus et al., US 20200250304 A1 (hereinafter Kruus).
Regarding Claim 2, Yan/Dan/Yoon/Drees respectively teaches and incorporates the claimed limitations and rejections of Claim 1.
The combination does not teach, but Kruus further teaches:
The method of Claim 1, wherein operating the generative adversarial network further comprises creating, by an embedder, a representation of preprocessed training data having reduced dimensionality relative to the preprocessed training data, (Kruus [Abstract] The method includes generating encoder direct output by projecting, via an encoder, input data items to a low-dimensional embedding vector of reduced dimensionality with respect to the one or more input data items to form a low-dimensional embedding space) (Note: the input data items correspond to the preprocessed training data.)
Yan/Dan/Yoon/Drees and Kruus are analogous to the present invention because both are from the same field of endeavor of generative adversarial networks. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to implement the low-dimensional embedding from Kruus to Yan/Dan/Yoon/Drees’s fault detection method. The motivation would be to “to remove the adversarial perturbations from the resultant manifold by minimizing the optimal transport cost between the feature distribution, possibly at different levels of abstraction” (Kruus [0015]).
Yan/Dan/Yoon/Drees/Kruus, via Yan, further teaches:
attempting, by a recovery, to reconstruct the preprocessed training data from the representation, (Yan [p.4 2nd col 2nd ¶] The reconstructed sample x′ is obtained through the decoder.
PNG
media_image7.png
233
624
media_image7.png
Greyscale
) (Note: x’ corresponds to the reconstructed training data from z, which corresponds to the representation. The decoder corresponds to a recovery.)
and attempting, by a discriminator, to discriminate to determine whether the synthetic timeseries data is synthetic. (Yan [p.2 2nd col 2nd ¶] Both synthetic data samples and the real-world data samples are inputted into the discriminator for the discriminator training process. When the discriminator cannot determine the authenticity of the input sample, the entire training process terminates.)
Regarding Claim 3, Yan/Dan/Yoon/Drees/Kruus respectively teaches and incorporates the claimed limitations and rejections of Claim 2. Yan/Dan/Yoon/Drees/Kruus, via Yan, further teaches:
The method of Claim 2, comprising: receiving, by the embedder and the conditional generator, the preprocessed training data; (Yan [fig. 3]
PNG
media_image7.png
233
624
media_image7.png
Greyscale
) (Note: encoder1 corresponds to the embedder; and decoder corresponds to the generator.)
providing, by the conditional generator, generated data to the discriminator; (Yan [fig. 3]
PNG
media_image7.png
233
624
media_image7.png
Greyscale
) (Note: x’ corresponds to the generated data)
providing, by the embedder, a first output to the discriminator; (Yan [fig. 3]
PNG
media_image7.png
233
624
media_image7.png
Greyscale
) (Note: z corresponds to the first output; the discriminator receives the representation of the first output, thus corresponding to the embedder providing a first output to the discriminator.)
and providing, by the embedder, a second output to the recovery. (Yan [p.4 2nd col 2nd ¶] The reconstruction loss l1 between the n-dimensional vector z' and the n-dimensional vector z output by the encoder in the first subnetwork is considered as the inferential exception.) (Note: z’ corresponds to the second output.)
Regarding Claim 4, Yan/Dan/Yoon/Drees/Kruus respectively teaches and incorporates the claimed limitations and rejections of Claim 2. Yan/Dan/Yoon/Drees/Kruus, via Yan, further teaches:
The method of Claim 2, comprising enabling, by the embedder, learning of temporal dynamics from a latent space. (Yan [p.2 2nd col 2nd ¶] The synthetic data sample x˜ is encoded into a n-dimensional vector z through encoder1.) (Note: encoder1 corresponds to the embedder; since the dataset is time series data, the encoding into a n-dimensional vector z corresponds to enabling learning of temporal dynamics from a latent space.
Claims 13 and 14, dependent on 12, also recite the One or more non-transitory computer-readable media configured to perform precisely the methods of Claims 2 and 3, respectively. Thus, Claims 13 and 14 are rejected for reasons set forth in Claims 2 and 3, respectively.
Claims 6 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Yan/Dan/Yoon/Drees further in view of Liu et al., “A statistical-based online cross-system fault detection method for building chillers” (hereinafter Liu).
Regarding Claim 6, Yan/Dan/Yoon/Drees respectively teaches and incorporates the claimed limitations and rejections of Claim 1. The combination does not teach, but Liu further teaches:
The method of Claim 1, wherein training the conditional generator is based on first actual timeseries data for a first unit of building equipment,
the method further comprising updating the conditional generator for a second unit of building equipment by operating the generative adversarial network based on actual timeseries data for the second unit of building equipment. (Liu [Abstract] The proposed FD model can be cross-utilized between building chillers with various specifications while it only needs to update the original fault detection model by the normal operation data of the new chiller system, thus saving huge fault experimental costs for the fault detection of new chiller.) (Note: the normal operation data of the new chiller system corresponds to the actual timeseries data for the second unit of building equipment.)
Yan/Dan/Yoon/Drees and Liu are analogous to the present invention because both are from the same field of endeavor of fault detection methods for building equipment. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the cross-utilization from Liu with Yan’s fault detection method. The motivation would be to “save[] huge fault experimental costs for the fault detection of new chiller” (Liu [Abstract]).
Regarding Claim 15, Yan/Dan/Yoon/Drees respectively teaches and incorporates the claimed limitations and rejections of Claim 12. The combination does not teach, but Liu further teaches:
The one or more non-transitory computer-readable media of Claim 12, wherein the operations further comprise updating the conditional generator for a new unit of building equipment by operating the generative adversarial network based on actual timeseries data for the new unit of building equipment. (Liu [Abstract] The proposed FD model can be cross-utilized between building chillers with various specifications while it only needs to update the original fault detection model by the normal operation data of the new chiller system, thus saving huge fault experimental costs for the fault detection of new chiller.) (Note: the normal operation data of the new chiller system corresponds to the actual timeseries data for the second unit of building equipment.)
Yan/Dan/Yoon/Drees and Liu are analogous to the present invention because both are from the same field of endeavor of fault detection methods for building equipment. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply the cross-utilization from Liu with Yan’s fault detection method. The motivation would be to “save[] huge fault experimental costs for the fault detection of new chiller” (Liu [Abstract]).
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 JOSEP HAN whose telephone number is (703)756-1346. The examiner can normally be reached Mon-Fri 9am-5pm.
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, Kakali Chaki can be reached on (571) 272-3719. 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.
/J.H./Examiner, Art Unit 2122
/KAKALI CHAKI/Supervisory Patent Examiner, Art Unit 2122