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 .
Specification
The abstract of the disclosure is objected to because line 2 is missing the definition of EPG. A corrected abstract of the disclosure is required and must be presented on a separate sheet, apart from any other text. See MPEP § 608.01(b).
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 1-3, 6-7, 9-10, 13-15, and 18-19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Modugula et al. (US20230259798A1 -hereinafter Modugula) in view of Umunnakwe et al. (US20240399181A1 -hereinafter Umunnakwe) in view of Sun et al. (US20240088676A1 -hereinafter Sun).
Regarding Claim 1, Modugula teaches a computer-implemented method for mitigating risks …comprising:
fusing collected risk data and …data to obtain fused data; (see [0128]; Modugula: “At step 406, the event server may be configured to generate combined data using data fusion. For example, the combined data may be generated based on the localized data from the distributed multi-sensor network and at least one of the property data, community infrastructure data or the environmental data. In some embodiments, combined data may include, for each building associated with each property of the community, a number of stories, a building foundation, a year of construction, first floor elevation (above ground), distance to nearest stream, a catchment area upstream of the property, and/or the like. In some embodiments, the combined data may include a hazard map.”)
predicting a vulnerability metric and a fragility metric …based on risk profiles generated from the fused data with a …neural network (PINN) trained with the fused data; (see [0049]; Modugula: “A vulnerability model may need to be interpolated using machine learning (e.g., recurrent neural network or convolutional neural network models) and/or fragility functions.” See [0053]: “using an artificial intelligence algorithm to determine a natural hazard risk score associated with the location, the natural hazard risk score being based on the natural hazard data and the vulnerability data;”)
developing …threat metrics by integrating the vulnerability metric, fragility metric, and the risk profiles into an integrated score that determines the probability of failure …caused by natural destructive forces; and (see [0069]; Modugula: “the model may be a machine learning model and using the model to determine the risk score associated with the location may include providing the initial hazard data and the vulnerability data as inputs to the machine learning model. Embodiments may further comprise training the model to generate a risk score using historical”. See [0033]: “A risk score as described herein may be based on different future time horizons and may be keyed to hydrological predictions, such as risks associated with a seven-year flood, a one-hundred year flood, and so on, while recognizing that the flood depth associated with these floods is a moving target. It may estimate the likelihood of floods of a certain size and associated damage. As an example, a model may predict the likelihood of a six-inch or 3-foot flood at a property and estimate property damages associated with the flood based on vulnerability, allowing insurance companies to estimate the probability of a claim being filed of a given magnitude.”)
performing, with an automated helper, a corrective action to mitigate the risks …caused by the natural destructive forces determined from the …threat metrics. (see [0034]; Modugula: “Embodiments of the present disclosure may include monitoring and data reporting for periodic or continuous score assessment, enabling the system to respond to triggering events. The system may include triggering a utility pump, for example, based on a determined flood risk.”)
However, Modugula does not explicitly teach:
… mitigating risks for electrical power grids,
electrical power grid (EPG) data,
predict …metric of the EPG …with a physics-informed neural network (PINN)…
developing EPG threat metrics…
determines the probability of failure of the EPG…
…mitigate the risks to the EPG caused by the natural destructive forces determined from the EPG threat metrics.
Umunnakwe from the same or similar field of endeavor teaches:
… mitigating risks for electrical power grids, (see [0047]; Umunnakwe: “It provides a spatiotemporal data-driven technique in modeling wildfire threats, better suited for risk reduction for the bulk power grid.”)
electrical power grid (EPG) data, (see [0031]; Umunnakwe: “collecting data from the bulk power grid”)
developing EPG threat metrics… (see [0075]; Umunnakwe: “Part of the risk assessment model also includes proposed power system-wildfire metrics that enable optimal operational strategies such as choosing mitigation vs. restoration (“let-burn”).”)
determines the probability of failure of the EPG… (see [0052]; Umunnakwe: “The first stage of the framework estimates the spatiotemporal probability of potential wildfire ignition and propagation using a deep neural network (DNN) in combination with the wildfire physical spread model.”)
…mitigate the risks to the EPG caused by the natural destructive forces determined from the EPG threat metrics. (see [0101]; Umunnakwe: “Metrics for power grid wildfire risk assessment are developed to aid utility decision making process and operational strategies in the wake of a wildfire threat.” See [0179]: “Optimization of system UAV resources can be used to mitigate both endogenous and exogenous wildfires before they occur, or manage these fires in real-time, if they occur. The SL-PWR utilizes potential ignitions pre-wildfire in order to prepare for critical scenarios and proceed with optimal strategies to better respond to and mitigate risks arising from extreme wildfire events including the propensity of outages caused by exogenous wildfires and power shutoff to customers as a result of wildfire threat (de-energization to prevent endogenous/power equipment-caused wildfire).”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify the teaching of Modugula to include Umunnakwe’s features of mitigating risks for electrical power grids; electrical power grid (EPG) data; developing EPG threat metrics; determining the probability of failure of the EPG; and mitigate the risks to the EPG caused by the natural destructive forces determined from the EPG threat metrics. Doing so would improve resilience in power utility wildfire response. (Umunnakwe, Abstract)
However, it does not explicitly teach: predict …metric of the EPG …with a physics-informed neural network (PINN)…
Sun from the same or similar field of endeavor teaches
predict …metric of the EPG …with a physics-informed neural network (PINN)… (see [0024]; Sun: “a machine learning solution with a physics-informed neural network (PINN) that solves the optimization problem by training a neural network to minimize a lost function including the cost function.” See [0127]: “It is assumed that the predictions of load demands, solar and wind generations are available and can be treated as fixed values”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify the teaching of Modugula and Umunnakwe to include Sun’s features of predicting metric of the EPG with a physics-informed neural network (PINN). Doing so would provide a new optimization space suitable for power flow analysis allowing to reduce the computational burden and maintain system stability. (Sun, [0005] and [0009])
Regarding Claim 2, the combination of Modugula, Umunnakwe, and Sun teaches all the limitations of claim 1 above, Modugula further teaches wherein fusing the collected risk data and the …data, further comprises assigning weights based on a significance of each feature for natural destructive force predictions to obtain assigned feature weights. (see [0138]; Modugula: “each feature fed into the machine learning model may be weighted or scaled according to importance. For example, a higher weight for a feature may mean that it has a greater impact on the community risk score while a lower weight for a feature may mean that it has less of an impact.”)
Umunnakwe further teaches …EPG data, (see [0031]; Umunnakwe: “collecting data from the bulk power grid”)
The same motivation to combine Modugula and Umunnakwe a set forth for Claim 1 equally applies to Claim 2.
Regarding Claim 3, the combination of Modugula, Umunnakwe, and Sun teaches all the limitations of claim 2 above, Modugula further teaches wherein fusing the collected risk data and the …data, further comprises combining features in grid cells using the assigned feature weights. (see [0138]; Modugula: “As an exemplary embodiment, combined data may include pre-processed environmental data, such as a flood hazard map, and property data, such as a number of stories, building foundation, year of construction, and the like associated with each building for each property of the community. The flood hazard map may give flood depths for each grid cell (e.g., 30×30 meter grid cell) and the machine learning model may be configured to determine a risk score for each grid cell by comparing flood depths between grid cells.”)
Umunnakwe further teaches …EPG data, (see [0031]; Umunnakwe: “collecting data from the bulk power grid”)
The same motivation to combine Modugula and Umunnakwe a set forth for Claim 1 equally applies to Claim 3.
Regarding Claim 6, the combination of Modugula, Umunnakwe, and Sun teaches all the limitations of claim 1 above, Modugula further teaches wherein the predicting vulnerability metric and the fragility metric further comprises continuously learning the vulnerability metric and the fragility metric using feedback obtained from a feedback system and newly collected data while retaining previously learned knowledge. (see [0040]; Modugula: “L/AI models 104 may be trained over time using one or more feedback loops. For example, flood depth data from one or more location devices 102 may be used to train L/AI models 104. In some embodiments, L/AI models 104 may be located on a cloud platform. Additionally or alternatively, L/AI models 104 may be configured to receive data from location device 102 and external data sources 108. Based on the received data, L/AI models 104 may be configured to compute one or more risk scores and send the computed one or more risk scores to customer platform 106. In some embodiments, L/AI models 104 may be configured to communicate computed risk scores to customer platform 106 in real time.”)
Regarding Claim 7, the combination of Modugula, Umunnakwe, and Sun teaches all the limitations of claim 1 above, Modugula further teaches wherein developing the …threat metrics further comprises generating a visualization object that presents the …threat metric in a map that includes the risk profiles, the fragility and vulnerability metrics.... (see [0150]; Modugula: “method 400 may further comprise generating a visualization. For example, the visualization may comprise a map, chart, graph, dashboard, 3D model, infographic, or any other visualization capable of communicating risk of hazard. Maps may include static maps, interactive maps, animations, or any other map to show the extent and severity of a hazard, as well as the areas that are most vulnerable to its impacts. Charts and graphs may be generated to visualize trends in the hazard over time, such as changes in water levels, wind speeds, and the like. Dashboards may provide real-time information on the status of a hazard and its potential impacts using data from sensors and other sources. 3D models may provide a more immersive visualization of the hazard and its potential impacts, allowing stakeholders to see the risks and vulnerabilities from different perspectives. Infographics may provide clear and concise information on the hazard and its potential impacts by using a combination of visual elements and text. The visualization may be generated according to any method known in the art. In some embodiments, generating a visualization may be performed by the event server. In some embodiments, generating a visualization may be performed by a device associated with a management entity.”)
Umunnakwe further teaches developing EPG threat metrics… (see [0075]; Umunnakwe: “Part of the risk assessment model also includes proposed power system-wildfire metrics that enable optimal operational strategies such as choosing mitigation vs. restoration (“let-burn”).”)
The same motivation to combine Modugula and Umunnakwe a set forth for Claim 1 equally applies to Claim 7.
Regarding Claim 9, the limitations in this claim is taught by the combination of Modugula, Umunnakwe, and Sun as discussed connection with claim 1.
Regarding Claim 10, the combination of Modugula, Umunnakwe, and Sun teaches all the limitations of claim 9 above, Umunnakwe further teaches further comprising a network of sensors to collect natural destructive force data and EPG data from the EPG and its surroundings. (see Abstract; Umunnakwe: “monitoring one or more sensors for an indication of a wildfire event.” See [0031]: “collecting data from the bulk power grid.”)
Regarding Claim 13, the limitations in this claim is taught by the combination of Modugula, Umunnakwe, and Sun as discussed connection with claim 1.
Regarding Claim 14, the limitations in this claim is taught by the combination of Modugula, Umunnakwe, and Sun as discussed connection with claim 2.
Regarding Claim 15, the limitations in this claim is taught by the combination of Modugula, Umunnakwe, and Sun as discussed connection with claim 3.
Regarding Claim 18, the limitations in this claim is taught by the combination of Modugula, Umunnakwe, and Sun as discussed connection with claim 6.
Regarding Claim 19, the limitations in this claim is taught by the combination of Modugula, Umunnakwe, and Sun as discussed connection with claim 7.
Claim(s) 4-5 and 16-17 is/are rejected under 35 U.S.C. 103 as being unpatentable over Modugula in view of Umunnakwe in view of Sun in view of Birru et al. (US20230351159A1 -hereinafter Birru).
Regarding Claim 4, the combination of Modugula, Umunnakwe, and Sun teaches all the limitations of claim 1 above, Modugula further teaches wherein the predicting vulnerability metric and the fragility metric… (see [0049]; Modugula: “A vulnerability model may need to be interpolated using machine learning (e.g., recurrent neural network or convolutional neural network models) and/or fragility functions.” See [0053]: “using an artificial intelligence algorithm to determine a natural hazard risk score associated with the location, the natural hazard risk score being based on the natural hazard data and the vulnerability data;”)
However, it does not explicitly teach: …further comprises training the PINN by minimizing a joint-loss function that balances data-driven predictions and physics-based constraints that is controlled by a hyperparameter.
Birru from the same or similar field of endeavor teaches…further comprises training the PINN by minimizing a joint-loss function that balances data-driven predictions and physics-based constraints that is controlled by a hyperparameter. (see [0132]; Birru: “The PINN is trained using optimization algorithms such as stochastic gradient descent (SGD) or Adam. The training process iteratively adjusts the network parameters to minimize the combined loss function.” See [0130]: “The loss function used in training the PINN is defined to incorporate both data-driven and physics-driven terms. The data-driven term quantifies the discrepancy between the predicted outputs of the network and the labeled data. The physics-driven term ensures that the network satisfies the fluid dynamics equations. It is typically formulated as the residual of the governing equations, representing the difference between the predicted and true derivatives or gradients.”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify the teaching of Modugula, Umunnakwe, and Sun to include Birru’s features of training the PINN by minimizing a joint-loss function that balances data-driven predictions and physics-based constraints that is controlled by a hyperparameter. Doing so would achieve a powerful framework for solving complex scientific problems while ensuring physical consistency and interpretability. (Birru, [0009])
Regarding Claim 5, the combination of Modugula, Umunnakwe, and Sun teaches all the limitations of claim 1 above, Modugula further teaches wherein the predicting vulnerability metric and the fragility metric … (see [0049]; Modugula: “A vulnerability model may need to be interpolated using machine learning (e.g., recurrent neural network or convolutional neural network models) and/or fragility functions.” See [0053]: “using an artificial intelligence algorithm to determine a natural hazard risk score associated with the location, the natural hazard risk score being based on the natural hazard data and the vulnerability data;”)
However, it does not explicitly teach: … further comprises filtering out predictions based on a confidence threshold computed using a softmax layer.
Birru from the same or similar field of endeavor teaches …further comprises filtering out predictions based on a confidence threshold computed using a softmax layer. (see [0254]-[0255]; Birru: “In the context of a transformer, SoftMax 1314 refers to a mathematical function called the SoftMax function, which is commonly used to normalize a vector of values into a probability distribution. The SoftMax function is utilized in certain parts of the transformer model to convert scores or logits into probabilities. In the self-attention mechanism of the transformer, attention weights are computed by applying the SoftMax function to the attention scores. These attention scores represent the relevance or importance of different positions in the input sequence. By using SoftMax, the scores are transformed into a probability distribution, ensuring that the weights sum up to 1 and can be interpreted as attention probabilities.”)
The same motivation to combine Modugula, Umunnakwe, Sun, and Birru a set forth for Claim 4 equally applies to Claim 5.
Regarding Claim 16, the limitations in this claim is taught by the combination of Modugula, Umunnakwe, Sun, and Birru as discussed connection with claim 4.
Regarding Claim 17, the limitations in this claim is taught by the combination of Modugula, Umunnakwe, Sun, and Birru as discussed connection with claim 5.
Claim(s) 8 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Modugula in view of Umunnakwe in view of Sun in view of Sun (US20220268827A1 -hereinafter Sun27).
Regarding Claim 8, the combination of Modugula, Umunnakwe, and Sun teaches all the limitations of claim 1 above; however, it does not explicitly teach: wherein performing the corrective action further comprises redirecting power generated from one node determined to be affected by the natural destructive forces to another node.
Sun27 from the same or similar field of endeavor teaches wherein performing the corrective action further comprises redirecting power generated from one node determined to be affected by the natural destructive forces to another node. (see [0018]; Sun: “rerouting power and restoring service to disconnected power or loads of the feeder with the fault.” See [0136]: “Event: is considered some action that caused damage to at least a portion of the power grid, resulting in a potential of, a destabilization of or loss of, power in the power distribution network, which causes an interruption of suppling continuous power either immediately or sometime in a near future. Some examples of events may be considered as natural disaster event (weather, earthquake, etc.), an intentional damaging event (terrorist attack, etc.) or an unintentional damaging event (fallen trees, plane crash, train wreck, etc.).”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify the teaching of Modugula, Umunnakwe, and Sun to include Sun27’s features of redirecting power generated from one node determined to be affected by the natural destructive forces to another node. Doing so would improve the accuracy of fault detection and location and achieve quick restoration. (Sun27, [0002] and [0133])
Regarding Claim 20, the limitations in this claim is taught by the combination of Modugula, Umunnakwe, and Sun as discussed connection with claim 8.
Claim(s) 11 is/are rejected under 35 U.S.C. 103 as being unpatentable over Modugula in view of Umunnakwe in view of Sun in view of Eusebi et al. (NPL: “Physics-Informed Neural Networks for Hurricane Reconstruction and Data Assimilation” -hereinafter Eusebi).
Regarding Claim 11, the combination of Modugula, Umunnakwe, and Sun teaches all the limitations of claim 9 above; however, it does not explicitly teach: wherein the PINN further comprises a physics-informed layer that ensures predictions adhere to real-world natural destructive force dynamics.
Eusebi from the same or similar field of endeavor teaches wherein the PINN further comprises a physics-informed layer that ensures predictions adhere to real-world natural destructive force dynamics. (see page 1, last paragraph; Eusebi: “Here, we show that a Physics-Informed Neural Network 11,12 can provide a promising alternative for data assimilation for TC initial conditions. Using synthetic training data sparsely sampled from hurricanes simulated in a forecast model, a PINN is able to accurately reconstruct full 2- and 3-dimensional wind and pressure fields of a TC.
We also demonstrate how a set of sparse, real-time observations, can be used to accurately reconstruct Hurricane Ida.”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify the teaching of Modugula, Umunnakwe, and Sun to include Eusebi’s features of a physics-informed layer that ensures predictions adhere to real-world natural destructive force dynamics. Doing so would achieve effective preparation, evacuation, and mitigation. (Eusebi, page 2, first paragraph)
Claim(s) 12 is/are rejected under 35 U.S.C. 103 as being unpatentable over Modugula in view of Umunnakwe in view of Sun in view of Imrie te al. (NPL: “River flow prediction using artificial neural networks: generalisation beyond the calibration range” -hereinafter Imrie)
Regarding Claim 12, the combination of Modugula, Umunnakwe, and Sun teaches all the limitations of claim 9 above, Sun further teaches wherein the PINN… (see [0024]; Sun: “a machine learning solution with a physics-informed neural network (PINN) that solves the optimization problem by training a neural network to minimize a lost function including the cost function.”
The same motivation to combine Modugula, Umunnakwe, and Sun a set forth for Claim 1 equally applies to Claim 12.
However, it does not explicitly teach further comprises an activation layer that employs a custom activation function formulated for natural destructive force dynamics using a cubic function that employs parameters learned during training and input data.
Imrie from the same or similar field of endeavor teaches further comprises an activation layer that employs a custom activation function formulated for natural destructive force dynamics using a cubic function that employs parameters learned during training and input data. (see Abstract; Imrie: “The ability of ANNs trained with different output activation functions to extrapolate beyond the calibration data is assessed.” See page 151: “Fig. 9. Portion of the measured River Dove ¯ow at Marston on Dove and corresponding 4-h predictions made by original cascade-correlation ANN and modified cascade-correlation ANN with cubic polynomial output activation function”.)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify the teaching of Modugula, Umunnakwe, and Sun to include Imrie’s features of an activation layer that employs a custom activation function formulated for natural destructive force dynamics using a cubic function that employs parameters learned during training and input data. Doing so would ensure model generality in terms of overall performance and the ability to estimate extreme values. (Imrie, page 138, left paragraph)
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Cook (US20240168198A1) discloses activation functions (for neural networks, including [but not limited to] rectified linear unit, softmax, and sigmoid)
Misyris et al. (NPL: “Physics-Informed Neural Networks for Power Systems” (2020)) discloses a framework for physics-informed neural networks in power system applications.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to VI N TRAN whose telephone number is (571)272-1108. The examiner can normally be reached Mon-Fri 9:00-5:00.
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, ROBERT FENNEMA can be reached at (571) 272-2748. 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.
/V.N.T./Examiner, Art Unit 2117
/Christopher E. Everett/Primary Examiner, Art Unit 2117
Prosecution Insights