DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claims 1-20 have been presented for examination based on the application filed on 6/18/2026.
Claims 13-20 are new.
Claims 1-20 are rejected under 35 U.S.C. 101 .
Claims 1-8, 11-12, 16-20 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by NPL by Zhao, et al. "A learning-to-infer method for real-time power grid topology identification." arXiv preprint arXiv:1710.07818 (2017).
Claim(s) 9, 10, 14-15 are rejected under 35 U.S.C. 103 as being unpatentable over NPL by Zhao, et al. "A learning-to-infer method for real-time power grid topology identification." arXiv preprint arXiv:1710.07818 (2017), in view of De Jongh, Steven, et al. "Physics-informed geometric deep learning for inference tasks in power systems." Electric Power Systems Research 211 (2022): 108362 (Jongh hereafter).
This action is made Final.
---- This page is left blank after this line ----
Response to Arguments
(Argument 1) Applicant has argued in Remarks Pg.9-13:
PNG
media_image1.png
254
652
media_image1.png
Greyscale
(Response 1) Examiner points to Zhao where the machine learning used to predict the state of line (sl), and Zhao further states such can be used to determine the Admittance Matrix which comprises conductance (Gmn) and suceptance (Bmn), which are electrical values associated with components (such as a bus).
PNG
media_image2.png
506
1119
media_image2.png
Greyscale
(Argument 2) Applicant has argued in Remarks Pg.13:
PNG
media_image3.png
166
654
media_image3.png
Greyscale
(Response 2) Zhao teaches adjusting parameters of the machine learning model responsive to comparisons between the output inferences with the simulation results indicating the ground truth electrical values of components (Zhao : §II.B ground truth values as measurements, and §III.A. adjusting p(s|y) and parameter .beta. based on the ground truth and monte carlo simulation as seen in Table I) of corresponding electrical system topologies (Zhao : Section V.B "... In the output layer we employ hinge loss as the loss function. In training the classifiers, we use stochastic gradient descent (SGD) with momentum update and Nesterov's acceleration..."). Examiner does not find applicant’s arguments persuasive and respectfully maintains the position.
---- This page is left blank after this line ----
(Argument 3) Applicant has argued in Remarks Pg.13-14:
PNG
media_image4.png
365
672
media_image4.png
Greyscale
(Response 3) Jongh teaches real value output and Zhao teaches status of line as output of neural network. However it is also clear from Zhao that both are related to each other using Admittance matrix. The rationale here is to supplement the teachings of Zhao with Jongh such that line status/unmeasured node which is unknown to be computed based on the known input which are same in both Zhao (See Eqn(3) and Jongh (Fig.3 at least). Examiner does not find applicant’s arguments persuasive and respectfully maintains the rejection.
PNG
media_image5.png
852
1119
media_image5.png
Greyscale
---- This page is left blank after this line ----
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 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to mental process without any additional elements that provide a practical application or amount to significantly more than the abstract idea.
Claims 1, 11 & 12:
Step 1: the respective claims are drawn to a method, system and article of manufacture respectively, falling under one of the four statutory categories of invention.
Step 2A, Prong 1: This part of the eligibility analysis evaluates whether the claim recites a judicial exception. As explained in MPEP 2106.04, subsection II, a claim “recites” a judicial exception when the judicial exception is “set forth” or “described” in the claim. The limitations are bolded for abstract idea/judicial exception identification.
Claim 1
Mapping Under Step 2A Prong 1
1. A method comprising: training a machine learning model to process a graph that represents an electrical system model for an electrical system to infer, from the graph, one or more unknown electrical values associated with at least one component of the electrical system that is specified by the electrical system model
obtaining data defining a plurality of graphs, each graph representing a respective electrical system topology;
obtaining, for each electrical system topology and from an electrical simulation system, simulation results indicating ground truth electrical values of component of the respective electrical system topology; and
training the machine learning model to predict electrical behaviors values of components for each electrical system topology of electrical systems including by;(i) applying the data defining each graph of the plurality of graphs as input to the machine learning model to cause the machine learning model to generate obtain respective output inferences representing electrical values of the components of the respective electrical topology, and
ii adjusting parameters of the machine learning model responsive to comparisons between the output inferences with the simulation results indicating ground-truth electrical values of components of corresponding electrical system topologies,
thereby, enabling the machine leaning model to predict, using the trained machine learning model and for the graph that represents the electrical system model for the electrical system, the one or more unknown electrical values associated with the at least one component of the electrical system that is specified by the electrical system model.
Abstract Idea/Mathematical Concept: The training the machine learning model given the level of generality it is claimed is considered to be mathematical calculations (as in MPEP 2106.04(a)(2)(I)(C)). The process of infer[ring] the electrical value can be considered as mathematical concept or mental step to infer the unknown electrical values (evaluation /judgement /opinion) based on mathematical construct (like graph (nodes, edges) as observation.). See MPEP 2106.04(a)(2)(III)(A)). Association of electrical model to electrical system & generic recitation of component does not overcome that the graph model remains an abstract idea.
See Step 2A Prong 2.
See Step 2A Prong 2.
Abstract Idea/Mathematical Concept: The training the machine learning model given the level of generality it is claimed is considered to be mathematical calculations (as in MPEP 2106.04(a)(2)(I)(C)). The process to predict electrical behaviors of electrical systems can be considered as mathematical concept or mental step to infer the unknown electrical values (evaluation /judgement /opinion) based on mathematical construct (like graph (nodes, edges) as observation.). See MPEP 2106.04(a)(2)(III)(A)).
Please note this mapping in done in view of latest Memo dated December 5, 2025 related to
Ex Parte Desjardins:
https://www.uspto.gov/patents/laws/examination-policy/subject-matter-eligibility?MURL=PatentEligibility
Abstract Idea/Mental step/Mathematical concept: Adjusting parameter is mental step as the basis of adjustment and how much adjustment is opinion based on the ground truth data. The comparison aspect can also be mental step and mathematical concept (to compare two numbers).
Abstract Idea/Mathematical concept: predicting … one or more unknown electrical values associated with the at least one component of the electrical system that is specified by the electrical system model is based on the mathematical model and therefore considered as abstract idea.
Also See Step 2A Prong 2.
Step 2A, Prong 2: This part of the eligibility analysis evaluates whether the claim as a whole integrates the recited judicial exception into a practical application of the exception. This evaluation is performed by (1) identifying whether there are any additional elements recited in the claim beyond the judicial exception, and (2) evaluating those additional elements individually and in combination to determine whether the claim as a whole integrates the exception into a practical application. See MPEP 2106.04(d). As per (1) the additional elements are identified as bolded parts of the limitations in column 1 of the table below, and as per (2) the evaluation is shown in the mapping section of the table.
In accordance with this step, the judicial exception is not integrated into a practical application.
Claim 1
Mapping Under Step 2A Prong 2
1. A method comprising: training a machine learning model to process a graph that represents an electrical system model for an electrical system to infer, from the graph, one or more unknown electrical values associated with at least one component of the electrical system that is specified by the electrical system model
obtaining data defining a plurality of graphs, each graph representing a respective electrical system topology;
obtaining, for each electrical system topology and from an electrical simulation system, simulation results indicating ground truth electrical values of component of the respective electrical system topology; and
training the machine learning model to predict electrical behaviors values of components for each electrical system topology of electrical systems including by;(i) applying the data defining each graph of the plurality of graphs as input to the machine learning model to cause the machine learning model to generate obtain respective output inferences representing electrical values of the components of the respective electrical topology, and
ii adjusting parameters of the machine learning model responsive to comparisons between the output inferences with the simulation results indicating ground-truth electrical values of components of corresponding electrical system topologies,
thereby, enabling the machine leaning model to predict, using the trained machine learning model and for the graph that represents the electrical system model for the electrical system, the one or more unknown electrical values associated with the at least one component of the electrical system that is specified by the electrical system model.
See Step 2A Prong 1.
Under MPEP 2106.05(g) determining whether a claim integrates the judicial exception into a practical application in Step 2A Prong Two or recites significantly more in Step 2B is whether the additional elements add more than insignificant extra-solution activity to the judicial exception. In this case the this is mere data gathering related to topology/graphs provided as input.
Under MPEP 2106.05(g) determining whether a claim integrates the judicial exception into a practical application in Step 2A Prong Two or recites significantly more in Step 2B is whether the additional elements add more than insignificant extra-solution activity to the judicial exception. In this case the this is mere data gathering related simulation results indicating an electrical behavior of the respective electrical system topology, as input.
Besides rejection under Step 2A Prong 1 above, this step may also be rejected under Step2A Prong 2: Under MPEP 2106.05(f)(1) the claim recites only the idea of a solution or outcome i.e., the claim fails to recite details of how a solution to a problem is accomplished. The recitation of claim limitations that attempt to cover any solution (adjusting parameters of the machine learning model) to an identified problem with no restriction on how the result is accomplished and no description of the mechanism for accomplishing the result (responsive to comparisons between the output inferences with simulation results of corresponding electrical system topologies), does not integrate a judicial exception into a practical application or provide significantly more because this type of recitation is equivalent to the words "apply it".
Further Under MPEP 2106.05(h) the use of machine learning to infer unknown electrical values is merely field of use at best because the claim intends to infer the unknown values in a abstract construct of graph. Whether that graph represents an electrical grid or lymphatic system or a river delta is mere field of use as the results do not improve the functioning of the claimed application.
Further Under MPEP 2106.05(a) the claim must include the components or steps of the invention that provide the improvement described in the specification. The specification1 merely alleges improvement due to use of machine learning but does not detail how the machine learning is itself improved or how the application of machine learning improves the electrical grid itself. Hence the claim is not directed to the improvement in the computer as no specific data structure is claimed (in contrast with Enfish) or is an improvement in the technical field (in contrast with McRO).
Further in view of Memo dated December 5, 2025 related to Ex Parte Desjardins: While the instant claim/application parallels Ex Parte Desjardins (i.e. the claimed inventions in both are a method of training a machine learning model) the instant limitation does not improve the machine learning (in contrast with Desjardins which “identified improvements as to how the machine learning model itself operates, including training a machine learning model to learn new tasks while protecting knowledge about previous tasks to overcome the problem of “catastrophic forgetting” encountered in continual learning systems.” – as added to MPEP § 2106.04(d), subsection III per the Memo).
Most importantly Importantly, examiner evaluates the claims as a whole in discerning at least the limitation “adjusting parameters ..” does not reflect as improvement in the machine learning by itself and specification (need not explicitly set forth the improvement), but it must describe the invention such that the improvement would be apparent. The improvement is not apparent because the use of machine learning is generic, which is applied to the field of use (electrical system as graph). No details are disclosed how any problems with machine learning are overcome such that it would be apparent to one skilled in the art that an improvement is made to machine learning itself. See MPEP § 2106.04(d)(1) as revised per the Memo above and contrasted with Ex Parte Desjardins2. Hence, when the claim is considered as a whole is at best limited to mathematical calculations.
Additionally this is simply field of use (use of graph models and machine learning in the field of electrical systems) under MPEP 2106.05(h).
Claim 1 does not recite any additional elements like a computer or a processor and appears to be an academic exercise in using a machine learning.
Claim 11 recites a system with additional elements of one or more computers, one or more storage devices…coupled with one or more computers performing the operations similar to claim 1. The recitation of generic computer components3 does not improve on the computer technology (e.g. does not show how as implied in specification [0023] which states "... machine learning models such as graph neural networks are significantly more efficient and easier to parallelize, this can not only improve the accuracy of overall power flow simulations, but also significantly reduce the consumption of computational resources..." , do improve on parallelism or reduce the consumption of computational resources. Such assertions without details in the specification how such an improvement is brought in instant claimed machine learning application is mere recitation of generic computer components.
Claim 12 recites One or more non-transitory computer storage media storing instructions that when executed by one or more computers cause the one or more computers to perform operations for training a machine learning model. Generic recitation of non-transitory computer storage media as additional element does not improve on functioning of the computer. See MPEP 2106.05(a) and (g).
Step 2B: This part of the eligibility analysis evaluates whether the claim as a whole amounts to significantly more than the recited exception i.e., whether any additional element, or combination of additional elements, adds an inventive concept to the claim. See MPEP 2106.05.
As discussed above with respect to integration of the abstract idea into a practical application, the additional element of using a one or more storage devices…coupled with one or more computers to perform the claimed steps amounts to no more than mere instructions to apply the exception using a generic computer/processing component. Mere instructions to apply an exception using a generic computer component cannot provide an inventive concept (see MPEP 2106.05(f)). (in context of claim 1, 11 & 12). The claims do not improve on the machine learning as a whole and generically apply it to electrical system. The claims 1, 11 & 12 are therefore considered to be patent ineligible.
Claims 2 & 3 recite generally extra solution activity and generally an attempt to link the field of use. This type of limitation merely confines the use of the abstract idea to a particular technological environment (adjusting performance/production of well based on simulation) and thus fails to add an inventive concept to the claims. MPEP 2106.05(g) & (h).
Claims 4-10, 13-20 recite further add merely to abstract idea as claimed in claim 1 & 11, and 12 respectively. The claims do not disclose any additional limitations that integrate the judicial exception into practical application (Step 2A Prong 2) or contribute significantly more (Step 2B).
----- This page is left blank after this line -----
Claim Rejections - 35 USC § 102
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 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 the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1-8, 11-12, 16-20 are rejected under 35 U.S.C. 102(a)(1) as being anticipated by NPL by Zhao, et al. "A learning-to-infer method for real-time power grid topology identification." arXiv preprint arXiv:1710.07818 (2017).
Regarding Claims 1, 11 & 12 (Updated 8/28/26)
Zhao teaches (Claim 1) A method (Zhao: Abstract "... a discriminative learning problem based on Monte Carlo samples generated with power flow simulations. A major advantage of the developed Learning-to-Infer method is that the labeled data used for training can be generated in an arbitrarily large amount fast and at very little cost"...") ,
(Claim 11) A system (Zhao : Section V.C "... On a laptop with an Intel Core i7 3.1-GHz CPU and 8 GB of RAM, with the 200K training samples, it takes about
14.7 hours...") comprising: one or more computers; and one or more storage devices communicatively coupled to the one or more computers, wherein the one or more storage devices store instructions that, when executed by the one or more computers, cause the one or more computers to perform operations for training a machine learning model/
(Claim 12) One or more non-transitory computer storage media storing instructions that when executed by one or more computers cause the one or more computers to perform operations (Zhao : Section V.C "... On a laptop with an Intel Core i7 3.1-GHz CPU and 8 GB of RAM, with the 200K training samples, it takes about 14.7 hours...")
the method/operations comprising: training a machine learning model (Zhao : Section V.B "...We employ two-layer (i.e., one hidden layer) fully connected neural networks for both the separate training architecture and the feature sharing architecture. Rectified Linear Units (ReLUs) are
employed as the activation functions in the hidden layer...")) to process a graph that represents an electrical system model for an electrical system (Zhao : Section II.A "... a power system with N-buses, and its baseline topology (i.e., the network topology when there is no line outage) with L lines...") to infer, from the graph, one or more unknown electrical values associated with at least one component (Zhao : §II.A. which shows that machine learning is us used to predict the status of line (sl), which can in turn be used to predict the electrical values – bus admittance matrix Y, conductance and susceptance at each of the places. Fig.11 shows such a graph for which Sl is predicted using machine learning:
PNG
media_image6.png
822
1088
media_image6.png
Greyscale
)
of the electrical system (Zhao: Section V.A"... We would like our predictor to be able to identify the topology for arbitrary values of power injections...") that is specified by the electrical system model (Zhao: See components of admittance matrix used in further computations pf real and reactive power injection in Eqn (1) on pg. 2 ¶II.A
PNG
media_image7.png
506
462
media_image7.png
Greyscale
) , comprising: obtaining data defining a plurality of graphs, each graph representing a respective electrical system topology (Zhao: Section V.C "...the generated 300K 30-bus topologies are distinct from each other, so are that of the generated SOOK
118 bus topologies and that of the 2.2M 300 bus topologies .... classifiers
trained with the generated data sets...") ; obtaining, for each electrical system topology and from an electrical simulation system, simulation results indicating ground truth electrical values of components (Zhao : §II.B.
PNG
media_image8.png
214
464
media_image8.png
Greyscale
) of the respective electrical system topology (Zhao: Section V.A "...we employ the DC
power flow model (2) to generate the data sets .... To generate a data set {st,P
t,yt, t = 1, ... , T}, we assume the prior distribution p(s,P) factors as p(s)p(P). As
such, we generate the network topologies sand the power injections P
PNG
media_image6.png
822
1088
media_image6.png
Greyscale
independently...") ; and training the machine learning model to predict electrical values of components for each electrical system topology (Zhao : Machine learning the value of status of line (sl) is predicted, as shown in Table I. It is also shown that given computed sl, admittance matrix can be determined which represent electrical values such as conductance and susceptance) by:
applying the data (Zhao : §II.A training includes inputting y as described in Eqn.(3) in §II.B) defining each graph of the plurality of graphs (Zhao : Fig.15 showing topology graph where each sl is between buses m and n as described in §II.A) as input to the machine learning model cause the machine learning model to generate respective output inferences (Zhao : Section V.A "We would like our predictor to be able to identify the topology") representing electrical values of the components of the respective electrical topology (Zhao: Table I see offline computation as training to generate output inferences in step 2-3 and §III.C. “Offline Learning for online Inference”), and
adjusting parameters of the machine learning model responsive to comparisons between the output inferences with the simulation results indicating the ground truth electrical values of components (Zhao : §II.B ground truth values as measurements, and §III.A. adjusting p(s|y) and parameter .beta. based on the ground truth and monte carlo simulation as seen in Table I) of corresponding electrical system topologies (Zhao : Section V.B "... In the output layer we employ hinge loss as the loss function. In training the classifiers, we use stochastic gradient descent (SGD) with momentum update and Nesterov's acceleration..."); Also see Fig.4 training and validation losses in reference to Section V.C) thereby, enabling the machine leaning model to predict, using the trained machine learning model and for the graph that represents the electrical system model for the electrical system (Zhao: §V and Fig.11) , the one or more unknown electrical values associated with the at least one component of the electrical system that is specified by the electrical system model. (Zhao: Once the sl is predicted by the machine learning as discussed in §V, reverting back to §II, computation of
PNG
media_image6.png
822
1088
media_image6.png
Greyscale
admittance matrix can be done based on the Sl .
).
Regarding Claim 2 (Updated 8/28/26)
Zhao teaches the method of claim 1, wherein the simulation results indicating the ground truth electrical values of components of the respective electrical system topology are generated using a ground-truth electrical simulation system (Zhao: Section I "... the labeled data set for training the variational model can be generated in an arbitrarily large amount…”; and Section V.A and e,g, as stated "... we employ the DC power flow model (2) to generate the data sets ..."; Here the P, Qin are related to computation based on the electrical values of components, namely conductance Gmn and susceptance Bmn, as shown in Eqn(1)
PNG
media_image9.png
506
1119
media_image9.png
Greyscale
) .
Regarding Claim 3 (Updated 8/28/26)
Zhao teaches the method of claim 1, wherein data defining the plurality of graphs, each graph representing a respective electrical system topology (Zhao: Section V.C "...the generated 300K 30-bus topologies are distinct from each other, so are that of the generated SOOK 118 bus topologies and that of the 2.2M 300 bus topologies .... classifiers trained with the generated data sets..."; selecting one such graph is shown in process performed by Zhao and would be applicable to others (In re Harza)), includes one or more unknown electrical values within the electrical system (Zhao: Section I "... real-time prediction of the network topology based on newly observed instant measurements from the real world...") .
Regarding Claim 4
Zhao teaches the method of claim 1, wherein the electrical system represented by the graph is a real- world electrical power grid, and wherein the electrical system topology is a topology of the real- world electrical power grid (Zhao: Section I "... real-time prediction of the network topology based on newly observed instant measurements from the real world..."; Section V.C "...the testing procedure, i.e., real time topology identification, is performed extremely fast: In all of our numerical experiments, the testing time per data sample is under a millisecond. The extremely fast testing speed demonstrates that the proposed approach applies very well to real-time tasks, such as failure identification during cascading failures...." ).
Regarding Claim 5 (Updated 8/28/26), 17 (New), 19 (New)
Zhao teaches the method/system/media of claim 1/11/12, wherein the graph that represents the electrical system model for the electrical system (Zhao : e.g. Fig.11) comprises a plurality of nodes and a plurality of edges, wherein: (i) each node represents a bus in the electrical system and is associated with respective node features (Zhao: Section II.A "... we also employ smn E {1, O} to denote whether two buses m and n are connected by a line or not ... We denote the real and reactive power injections at all the buses by P ,Q E RN , and the voltage magnitudes and phase angles by V ,8 E RN...") , and (ii) edges in the graph are defined by a nodal admittance matrix that corresponds to a number of buses in the electrical system (Zhao: Section II.A "... Given the bus admittance matrix Y , the nodal power injections and the nodal voltage...") – here the bus admittance matrix defines the edge of the nodes connecting the two buses m and n connected by a line (edge) , each edge in the graph connects a pair of nodes in the graph, is associated with respective edge features, and represents a conductor in the electrical system that connects a pair of buses represented by the pair of nodes (Zhao: Section II.A "... "employ smn E {1,
O} to denote whether two buses m and n are connected by a line or not. Given a
network topology s, the system's bus admittance matrix Y can be determined
accordingly with the physical parameters of the system [21 ]: Ymn = smn (Gmn +
jBmn), where Gmn and Bmn denote conductance and susceptance
respectively. Note that, when two buses m and n are not connected, Ymn = smn
= O...") .
Regarding Claim 6 (Updated 8/28/26), 18 (New), 20 (New)
Zhao teaches the method/system/media of claim 5/17/12, wherein obtaining the data defining the plurality of graphs (Zhao: Section V.C "...the generated 300K 30-bus topologies are distinct from each other, so are that of the generated SOOK 118 bus topologies and that of the 2.2M 300 bus topologies .... classifiers trained with the generated data sets..."), each graph representing a respective electrical system topology comprises, for each graph of the plurality of graphs: obtaining data defining an admittance matrix (Zhao: Section II.A "... Given a network topology s, the system's bus admittance matrix Y can be determined accordingly with the physical
parameters of the system [21]: Y,,rn = s,,,n (Gmn + jB,,,,,), where Cmn and Bmn denote conductance and susceptance respectlvely. Note that, when two buses m and n are not connected, Yrnn = Smn = 0...") ; and assigning an edge between a pair of nodes in the graph based on values specified by the admittance matrix (Zhao: Section II.A "... Given a network topology s, the system's bus admittance matrix Y can be determined accordingly with the physical parameters of the system [21]: Y,,rn = s,,,n (Gmn + jBmn), where Gmn and Bmn denote conductance and susceptance respectively. Note that, when two buses m and n are not connected, Ymn = Smn = 0...")- assignment of edge is based on value of Ymn (admittance matrix), when Ymn = 0 , the busses are not connected) .
Regarding Claim 7
Zhao teaches the method of claim 6, wherein one or more of the nodes in the graph represent different bus types in the electrical system, and wherein the bus types include: a swing bus4 (Zhao: Section II.A teaching swing bus as slack bus - "... Typically, apart from a slack bus, most buses are "PC) buses" at which
the real and reactive power injections are controlled inputs, and the remaining buses are "PV buses" at which the real power injection and voltage magnitude are controlled inputs [21]. We refer the readers to [21] for more details of solving
AC power flow equations....") , a generator (Zhao: Section V.A "... With each pair of generated st and pt, we consider two types of measurements that constitute y: nodal voltage ph;ise ;ingle measurements and nodal power injection measurements. For these, a) we generate IID Gaussian voltage phase angle measurement noises with a standard deviation of 0.01 degree, the state-of-the-art PMU accuracy 1271, and b) we assume power injections are measured accurately. Here, we consider that measurements of voltage phase angles and power injections are collected at all the buses...." – power injections are mapped to busses with generator/load type5; Also see Section V.C “power injection”), and a load (Zhao: Section V.A - power injections are mapped to busses with generator/load type; Also see Section V.C “load”) .
Regarding Claim 8
Zhao teaches the method of claim 6, wherein the node features associated with each node that represents a respective bus in the electrical system include one or more of: a voltage magnitude, a voltage angle, an active power, and a real power (Zhao: Section II.A "... voltage magnitudes and phase angles ... Given the bus admittance matrix Y , the nodal power injections and the nodal voltages...") , and wherein the edge features associated with each edge include a current associated with the conductor in the electrical system represented by the edge (Zhao: Section II.A “conductance”6) .
Regarding Claim 16 (New)
Zhao teaches The method of claim 1, wherein the at least one component of the electrical system that is specified by the electrical system model comprises a bus (Zhao teaches: §II.A.¶1 "... The actual topology of the network can then be represented by s = [ s 1 , ... , s L]T. Generalizing this notation, we also employ smn E { 1, O} to denote whether two buses m and n are connected by a line or not....") .
---- This page is left blank after this line ----
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 for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claim(s) 9, 10, 14-15 are rejected under 35 U.S.C. 103 as being unpatentable over NPL by Zhao, et al. "A learning-to-infer method for real-time power grid topology identification." arXiv preprint arXiv:1710.07818 (2017), in view of De Jongh, Steven, et al. "Physics-informed geometric deep learning for inference tasks in power systems." Electric Power Systems Research 211 (2022): 108362 (Jongh hereafter).
Regarding Claim 9
Zhao teaches the method of claim 1, wherein the machine learning model is a graph neural network (Zhao: Section V.B "... two-layer (i.e., one hidden layer) fully connected neural networks for both the separate training architecture and the feature sharing architecture. Rectified Linear Units (ReLUs) are employed as the activation functions in the hidden layer...") , and wherein training the machine learning model to predict the electrical values of components for each electrical topology (Zhao : §II.A. showing how admittance matrix is determined by status of line sl, which is computed based on machine learning output as done in at least Table I) by applying the data defining each graph of the plurality of graphs (Zhao: Section V.C "...the generated 300K 30-bus topologies are distinct from each other, so are that of the generated SOOK 118 bus topologies and that of the 2.2M 300 bus topologies .... classifiers trained with the generated data sets...") as the input to the machine learning model to cause the machine learning to generate the respective output inferences (Zhao : Table I online portion step 2 generating inferences qB(sl,y) and sl) representing electrical values of the components of the respective electrical topology (Zhao : §II.A. showing how admittance matrix is determined by status of line sl, which is computed based on machine learning output as done in at least Table I) comprises, for each graph representing the respective electrical system topology: updating the graph at each of one or more update iterations (Zhao: Section V.B "... In the output layer we employ hinge loss as the loss function. In training the classifiers, we use stochastic gradient descent (SGD) with momentum update and Nesterov
'~ acceleration [28]. While this optimization algorithm works sufficiently well for our experiments, we note that other algorithms may further accelerate the training procedure [29]....") , comprising, at each update iteration:
Zhao does not explicitly teach comprising, at each update iteration: processing data defining the graph using the graph neural network in accordance with a set of graph neural network parameters to update a current node representation of each node in the graph and a current edge representation of each edge in the graph.
Jongh teaches comprising, at each update iteration: processing data defining the graph using the graph neural network in accordance with a set of graph neural network parameters to update a current node representation of each node in the graph (Jongh : Section 2.3 "... Furthermore,𝑈 is theupdate function which updates the node features of node𝑗 [nodes represent the grid or bus], given the messages received and the previous node features...."; ); Section 3.2 and a current edge representation of each edge in the graph (Jongh: Section 3.2 "... As input to the respective GNN architectures the nodal matrices 𝐆 and 𝐁 can be supplied as edge weights. ..." "... The randomly initiated weights of the GNN get updated using back propagation for each batch in the training data....").
It would have been obvious to one (e.g. a designer) of ordinary skill in the art before the effective filing date of the claimed invention to apply the teachings of Jongh to Zhao to further detail how the geometric deep learning techniques (like in Zhao Section I "... The proposed approach is also not restricted to specific models and learning methods, but can exploit any powerful models such as deep neural networks...") are applied to learn/infer from approximate models for power system estimation and calculation tasks (Jongh: Abstract). The motivation to combine would have been that Jongh and Zhao are analogous art to the instant claim in the field of modeling electrical/power grid/network as a neural network to infer power characteristics of the electrical/power grid/network (Jongh: Abstract Section 2.3, Section 3;Zhao: Abstract & Section III) .
Regarding Claim 10 (Updated 8/28/26)
Jongh teaches The method of claim 9, further comprising: after the updating (Jongh: Section 3.2 as above) , processing the respective current node representation for each node in the graph to generate a respective final feature corresponding to each node in the graph (Jongh: Section 3.2 as output nodal features
PNG
media_image10.png
228
364
media_image10.png
Greyscale
) , and processing the current edge representation for each edge in the graph to generate a respective final feature corresponding to each edge in the graph (Jongh: Section 3.2 "... As input to the respective GNN architectures the nodal matrices 𝐆 and 𝐁 can be supplied as edge weights. ..." "... The randomly initiated weights of the GNN get updated using back propagation for each batch in the training data....") ; and based on the respective final feature corresponding to each node in the graph and the respective final feature corresponding to each edge in the graph, generating the respective output inferences representing electrical values of components of the respective electrical topology (Jongh & Zhao: Jongh: Section 4
PNG
media_image11.png
170
524
media_image11.png
Greyscale
here the pseudo measurement are electrical values of components, e.g. active (p) and reactive (q) are electrical values which can lead to values of conductance and susceptance as disclosed in Zhao §II.A.
PNG
media_image12.png
506
1119
media_image12.png
Greyscale
Jongh further states:
PNG
media_image13.png
184
362
media_image13.png
Greyscale
).
Regarding Claim 14 (New)
Jongh teaches compiling the predictions of the one or more unknown electrical values generated by the trained machine learning model for a particular update iteration with simulation results generated by the electrical simulation system for the electrical system model for a corresponding simulation time step (Jongh: §5 Pg.6 Col.2 "... To consider additional failures
of PMUs, a randomly chosen fixed amount of measurements is set to
the average value of the respective physical quantity (e.g. setting all
voltage angles at failing PMUs to the average value of all voltage angles
at the given time step). ..." §3.1 "... The ground truth data is generated using non-linear power flow calculations for each time step in the yearly load profiles. It is assumed that the connection point to a high voltage grid acts as a slack node while all other nodes are modeled as 𝑃𝑄-nodes....") .
Regarding Claim 15 (New)
Jongh teaches The method of claim 1, wherein the electrical system model is missing one or more electrical values associated with at least one component of the electrical system that is specified by the electrical system model (Jongh teaches: Pg.6 Col.1 "... For this purpose, each GNN is trained on the power flow prediction task where the goal is to predict nodal voltage magnitudes and angles given active- and reactive nodal power injections...."; §4 & Fig.3).
---- This page is left blank after this line ----
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.
---- This page is left blank after this line ----
Communication
Any inquiry concerning this communication or earlier communications from the examiner should be directed to AKASH SAXENA whose telephone number is (571)272-8351. The examiner can normally be reached Mon-Fri, 7AM-3:30PM.
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, RYAN PITARO can be reached on (571) 272-4071. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
AKASH SAXENA
Primary Examiner
Art Unit 2188
/AKASH SAXENA/Primary Examiner, Art Unit 2188 Monday, March 16, 2026
1 See Specification [0023] "... Because machine learning models such as graph neural networks are significantly more efficient and easier to parallelize, this can not only improve the accuracy of overall power flow simulations, but also significantly reduce the consumption of computational resources (e.g., memory and computing power)...."
2 MPEP § 2106.04(d)(1) are revised to read: See, e.g., Ex Parte Desjardins, Appeal No. 2024-000567 (PTAB September 26, 2025, Appeals Review Panel Decision) (precedential), in which the specification identified the improvement to machine learning technology by explaining how the machine learning model is trained to learn new tasks while protecting knowledge about previous tasks to overcome the problem of “catastrophic forgetting,” and that the claims reflected the improvement identified in the specification. Indeed, enumerated improvements identified in the Desjardins specification included disclosures of the effective learning of new tasks in succession in connection with specifically protecting knowledge concerning previously accomplished tasks; allowing the system to reduce use of storage capacity; and the enablement of reduced complexity in the system. Such improvements were tantamount to how the machine learning model itself would function in operation and therefore not subsumed in the identified mathematical calculation.
3 Specification [0077]-[0084]
4 E.g. US 20130035885 A1, US 20130204556 A1 discussing swing bus in as slack bus in the network topology – can be used as prior art in future.
5 E.g. US 11063472 B1 by Ashrafi; Ashkan et al.(2021) Col.13 Line 35- Col.14 Line 9 showing injection (load or generator) /non-injection busses.
6 Conductance (e.g. Gmn in Zhao)) and current (I) have a direct linear relationship, defined by I=GV (where V is voltage), meaning higher conductance allows more current to flow at a given voltage.
Prosecution Insights