Power grid line power flow information inference threat assessment method and system

Abstract

The invention belongs to the field of power system information security and data inference, and particularly provides a power grid line power flow information inference threat assessment method and system, and the method comprises the following steps: obtaining node features and edge features from pre-public data, and taking the node features and the edge features as the input of a pre-constructed power grid line power flow information inference model, obtaining a power grid line power flow information inference value; an error value between the power grid line power flow information inference value and a preset label is calculated, the power grid line power flow information leakage threat degree is evaluated according to the error value, and the power grid line power flow information inference model comprises a feature projection module, a graph representation learning module, a depth feature extraction module and an inference module; the method improves the recognition precision of the power grid line power flow information leakage threat, and quantifies the possibility of inferring the key operation state of the power grid by using limited public information.

Description

Technical Field

[0001] The present invention belongs to the field of power system information security and data inference, and in particular relates to a power grid line flow information inference threat assessment method and system. Background Art

[0002] As digital transformation deepens, power data has become a key production factor, and its strategic value is increasingly prominent. In order to maximize the value of power data, power market information disclosure has become an important measure to ensure transparent market operation. However, due to the significant physical coupling and business association between power data, attackers may infer key sensitive information such as power grid line flow through public data, which poses a potential threat to the security of the power system.

[0003] However, there are many deficiencies in the current research on power system security, such as: First, existing research mainly limits the problem of inferring non-public sensitive data based on public data to the scope of parameter identification, focusing on inferring system sensitive parameters, while in-depth research on system sensitive operation data is relatively scarce.

[0004] Second, there is a complex nonlinear correlation between public data and power grid line flow information. It is difficult to accurately infer the flow state under different combinations of public data by relying solely on physical models, and the physical mechanisms of some scenarios have not been fully revealed. Relying solely on machine learning methods cannot effectively integrate the physical constraints of the power grid and has limited generalization capabilities.

[0005] Therefore, modern power systems need a method to evaluate the possibility of inferring power grid line flow information based on public data, and provide a reference for improving the information disclosure mechanism of the power market. Summary of the invention

[0006] The object of the present invention is to provide a method and system for assessing the threat of power line flow information inference, so as to solve the problem that the current power line flow information inference leakage threat assessment is difficult.

[0007] In order to achieve the above object, the technical solution adopted by the present invention is: In a first aspect, the present invention provides a method for assessing the threat of power line flow information inference, comprising the following steps: Obtain node features, edge features, and grid topology features from pre-public data of the power market, input the node features, edge features, and grid topology features into a pre-built grid line power flow information inference model, and obtain grid line power flow information inference values; The error value between the power line flow information inference value and the preset label is calculated, and the threat level of power line flow information leakage is evaluated according to the error value. The pre-built power line flow information inference model includes: Feature projection module, used to perform linear transformation on input node features and edge features; A graph representation learning module is used to update the output of the feature projection module in combination with the grid topology features; A deep feature extraction module is used to sequentially concatenate, transform and merge the outputs of the graph representation learning module; The inference module is used to perform dimension mapping on the output of the deep feature extraction module and output the inferred value of the power grid line flow information corresponding to the pre-public data.

[0008] Preferably, the loss function of the pre-constructed power grid line flow information inference model is constructed based on prior knowledge combined with a penalty term constrained by the physical characteristics of the power grid.

[0009] Preferably, the loss function of the pre-built power grid line flow information inference model is for:

[0010] in, All are non-negative weight coefficients; is the mean square error between the inferred value and the actual value; Penalty term for violating line capacity constraints; The penalty term for violating the shadow price constraint.

[0011] Preferably, the graph represents an expression of a learning module:

[0012]

[0013]

[0014]

[0015]

[0016]

[0017]

[0018] in, Represents the concatenation operation of features; They are multi-layer perceptrons with activation functions PReLU that match the pre-activation value, node feature, and edge feature dimensions respectively; and They are two linear layers with activation function Sigmoid and a single scalar output; It is the BatchNorm layer; For the Layer Node Features; For the Layer Node Features; For the Layer Edge Features; For the node The set of connected nodes; are the pre-activation values ​​of the features; For the Layer Node Features; No. Layer Edge Features; is a collection of nodes; is the set of edges; For the Characteristics of layer nodes; For the Features of layer edges.

[0019] Preferably, the deep feature extraction module performs splicing by the following formula:

[0020]

[0021] in, It is the concatenation of node features and edge features output by the graph representation learning module; Nodes output by the graph representation learning module Features; Nodes output by the graph representation learning module Features; is the edge output by the graph representation learning module Features; Represents the concatenation operation of features; is the set of edges; Nodes output by the graph representation learning module Features and nodes Features and edges The features after splicing.

[0022] Preferably, after evaluating the threat level of power grid line flow information leakage according to the error value, the method further includes identifying data having a key impact on the power grid line flow information inference in the public data of the power market by using the error value between the power grid line flow information inference value and the preset label.

[0023] In a second aspect, the present invention provides a power grid line flow information inference threat assessment system, comprising: A power flow inference value acquisition unit is used to acquire node characteristics, edge characteristics and power grid topology characteristics from pre-public data of the power market, input the node characteristics, edge characteristics and power grid topology characteristics into a pre-built power grid line power flow information inference model, and obtain the power grid line power flow information inference value; The threat level assessment unit is used to calculate the error value between the power line flow information inference value and the preset label, and assess the power line flow information leakage threat level according to the error value, wherein the pre-built power line flow information inference model includes: Feature projection module, used to perform linear transformation on input node features and edge features; A graph representation learning module is used to update the output of the feature projection module in combination with the grid topology features; A deep feature extraction module is used to sequentially concatenate, transform and merge the outputs of the graph representation learning module; The inference module is used to perform dimension mapping on the output of the deep feature extraction module and output the inferred value of the power grid line flow information corresponding to the pre-public data.

[0024] In a third aspect, the present invention provides a computer device, comprising: a processor suitable for executing a computer program; A computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the method described is performed.

[0025] In a fourth aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the described method when executed by a processor.

[0026] In a fifth aspect, the present invention provides a computer program product, characterized in that the computer program product comprises a computer program, and the computer program implements the described method when executed by a processor.

[0027] Compared with the prior art, the present invention has the following beneficial effects: The present invention provides a power line flow information inference threat assessment method, which pre-constructs a power line flow information inference model including a feature projection module, a graph representation learning module, a deep feature extraction module and an inference module, obtains node features and edge features from pre-public data, and uses the feature projection module to linearly transform the input node features and edge features to obtain feature representations of unified dimensions, and then uses the graph representation learning module to effectively learn and update the node features and edge features of unified dimensions at the same time, while the traditional network only focuses on the update of node features, resulting in insufficient utilization of edge features; then uses the deep feature extraction module to integrate the output features of the graph representation learning module according to the system topology structure, and then uses the residual connection mechanism to perform deep feature learning to effectively alleviate the gradient vanishing problem, and finally uses the inference module to map the extracted high-dimensional features to the target output dimension, and integrates the node features (such as power generation, load demand, node marginal electricity price) and edge features (such as capacity, shadow price) through the power line flow information inference model, reveals the implicit association between public data and power line flow information, and quantifies the possibility of inferring the key operating status of the power grid using limited public information.

[0028] Furthermore, the present invention proposes to integrate the penalty term of the physical characteristics constraints of the power grid into the loss function in the power grid line flow information inference model, so as to improve the accuracy of power flow inference under data sparse conditions and ensure the physical feasibility and operational consistency of the inference results. Compared with traditional physical models and pure machine learning methods, this method has stronger generalization ability and robustness.

[0029] Furthermore, the present invention can also identify and quantify low-sensitivity information in public data that has a key impact on the inference of power grid line flows. Through sensitivity analysis, it reveals the contribution of different types of data to the inference accuracy, clarifies the categories of sensitive data that need to be protected, and helps to optimize the information disclosure strategy of the power market. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is a flow chart of an embodiment of the present invention; Figure 2 A schematic diagram of a specific process of an embodiment of the present invention; Figure 3 This is a diagram of a neural network architecture for power grid line flow according to an embodiment of the present invention; Figure 4 This is a topology diagram of the IEEE 14-node test system in an embodiment of the present invention. DETAILED DESCRIPTION

[0031] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0032] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or combinations thereof.

[0033] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0034] As used in the specification and appended claims of this application, the term "if" can be interpreted as "when" or "uponce" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "uponce it is determined" or "in response to determining" or "uponce [described condition or event] is detected" or "in response to detecting [described condition or event]", depending on the context.

[0035] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0036] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0037] Example 1 like Figure 1As shown, a method for assessing power grid line flow information inference threat provided by this embodiment includes the following steps: Step 1: Obtain historical public data and undisclosed power grid line flow information corresponding to the power market, wherein node features, edge features, and power grid topology features are obtained from the historical public data, and the obtained node features, edge features, and power grid topology features are used to construct training sets and verification sets; and the undisclosed power grid line flow information is used as a preset label ; Step 2, constructing a neural network architecture for power line flow inference, embedding the penalty term of the power grid physical characteristic constraint into the loss function of the neural network architecture for power line flow inference based on prior knowledge, and generating a power line flow information inference model; Step 3: Optimize the model parameters using the training set, supervise the training process through the validation set, and obtain a trained power line flow information inference model; Step 4: Obtain node features and edge features from pre-public data, and input them into the power grid line flow information inference model to obtain the power grid line flow information inference value, calculate the error value between the power grid line flow information inference value and the preset label, and evaluate the threat level of power grid line flow information leakage based on the error value.

[0038] Example 2 like Figure 2 As shown, a method for assessing power grid line flow information inference threat provided by this embodiment includes the following steps: Step 1: Obtain the historical public data and unpublished power grid line flow information corresponding to the power market, wherein node features, edge features and power grid topology features are obtained from the historical public data, and the obtained node features, edge features and power grid topology features are used to construct training sets and verification sets; the unpublished power grid line flow information is used as a preset label ;Specifically: S101: Inferring a threat assessment target based on power grid line flow, obtaining historical public data corresponding to the power market and undisclosed power grid line flow information, wherein the historical public data includes power grid construction parameters, market transaction data, and system operation data, wherein: Grid construction parameters include: topology, node distribution (generation / substation / load), voltage level, line capacity, etc.; Market transaction data include: node marginal electricity price, shadow price, marginal energy price and marginal loss price; System operation data include: unit status, actual load, system backup information, power transmission status of important channels, transmission section constraints, inter-provincial interconnection line flow, key equipment flow data, etc.

[0039] S102: Structural processing is performed on the obtained historical public data to obtain node features, edge features and power grid topology features, where: The topological structure diagram of the power grid is set as ,in, Represents a collection of nodes, represents the set of edges; the number of nodes in the power grid is , the number of edges is ; Extract the data corresponding to each node in the topological structure diagram from the historical public data and construct the node features ; Extract the data corresponding to each edge in the topological structure graph from historical public data and construct edge features ; The topological characteristics of the power grid are obtained based on the connection relationship between nodes in the topological structure diagram; S103: Dataset Division and Storage The obtained node features, edge features and power grid topology features are used to construct training sets and validation sets; the processed data sets are saved in an efficient storage format to facilitate model training and validation.

[0040] Step 2: Construct a neural network architecture for power line flow inference, embed the penalty term of the power grid physical characteristic constraints into the loss function based on prior knowledge, and generate a power line flow information inference model. The specific method is: S201: Construct a neural network architecture for power line flow inference, the input of which is node features , edge features and grid topology characteristics, and the output is the inferred value of grid line flow information , the power grid line flow inference neural network architecture includes: Feature projection module: The input node features are projected through the fully connected layer and edge features Perform linear transformation and output hidden features; Graph representation learning module: A two-layer cascaded XENet network is used to update the hidden features and power grid topology features output by the feature projection module. Each layer of the XENet network updates the input hidden features using the following formula:

[0041]

[0042]

[0043]

[0044]

[0045]

[0046]

[0047] in, Represents the concatenation operation of features; They are multi-layer perceptrons with activation functions PReLU that match the pre-activation value, node feature, and edge feature dimensions respectively; and They are two linear layers with activation function Sigmoid and a single scalar output; It is the BatchNorm layer; For the Layer Node Features; For the Layer Node Features; For the Layer Edge Features; For the node connected nodes; are the pre-activation values ​​of the features; For the Layer Node Features; No. Layer Edge Features; For the Characteristics of layer nodes; For the Features of layer edges.

[0048] Deep feature extraction module: The following formula is used to concatenate the updated node features and edge features output by the graph representation learning module; then, the concatenated features are transformed through two series-connected fully connected layers. Each fully connected layer is followed by a ReLU activation function, and a residual connection is introduced to add the input features to the transformed features, and the merged features are output.

[0049]

[0050]

[0051] in, It is the concatenation of node features and edge features output by the graph representation learning module; Nodes output by the graph representation learning module Features; Nodes output by the graph representation learning module Features; is the edge output by the graph representation learning module Features; Represents the concatenation operation of features; is the set of edges; Nodes output by the graph representation learning module Features and nodes Features and edges The features after splicing.

[0052] Inference module: Use a fully connected layer to map the dimension of the merged features output by the deep feature extraction module to the target output dimension.

[0053] Step S202, constructing a loss function of a power grid line flow inference neural network architecture, wherein the loss function embeds a penalty term constrained by the physical characteristics of the power grid based on prior knowledge to obtain a power grid line flow information inference model.

[0054] In addition to the traditional data fitting term, i.e., the mean squared error (MSE) between the inferred value and the actual value, the following two penalty terms based on the constraints of the physical characteristics of the power grid are introduced to characterize the known mechanism of inferring line power flows from public data:

[0055]

[0056]

[0057] Penalty for violating line capacity constraints :According to the safe operation constraints of the power market model, the power line flow should not exceed its capacity limit, so the maximum capacity of the power line flow is set to ,have:

[0058] Penalty for violating the shadow price constraint :It is used to describe the relationship between power line flow, line capacity and shadow price. Assume that the line shadow price is ,have:

[0059] In summary, the overall loss function can be expressed as:

[0060] in, They are all non-negative weight coefficients, which are used to balance the magnitude of each loss component and regulate its relative contribution to model training. The weights are adjusted so that each loss component is at the same order of magnitude.

[0061] Optionally, other known physical constraints on power grid line flow in historical public data may be analyzed to construct penalty terms to be added to the loss function.

[0062] Step 3, using the training set to optimize the parameters of the power line flow information inference model, supervising the training process through the verification set, and obtaining a trained power line flow information inference model; S301, determining a set of model parameters to be optimized in a power grid line flow inference model; S302, in each training iteration, a batch of training data is obtained by random sampling, and input into the power grid line flow information inference model to obtain corresponding estimated values; S303: Optimize and update the model parameters based on the loss function until the loss value converges to the error threshold. In the following, the model training adopts the training set data, and the model hyperparameter optimization is adjusted by the validation set data.

[0063] Step 4: Process the pre-public data obtained from the power market according to step 1 to obtain the node features, edge features and grid topology features corresponding to the pre-public data; use the node features, edge features and grid topology features as the input of the grid line flow information inference model to obtain the grid line flow information inference value, and calculate the grid line flow information inference value and the preset label The error value between them is used to evaluate the threat level of leakage of power grid line flow information in the pre-public data.

[0064] Example 3 On the basis of Embodiment 1 or 2, this embodiment provides a method for assessing the threat of power line flow information inference, which, after assessing the threat level of power line flow information leakage according to the error value, also includes using the error value between the power line flow information inference value and the preset label to identify data in the public data of the power market that has a key impact on the power line flow information inference, specifically: Adjust input data, including selectively adding or removing specific categories of data, retraining the model based on the adjusted input data and performing power line flow inference; Calculates inferred values ​​and preset labels of power line flow information obtained using adjusted input data The error value between By comparing the changes in inference errors before and after input data adjustment, the data that has a key impact on the inference of power grid line flow information is identified.

[0065] Example 4 This embodiment is simulated and verified based on the IEEE 14-node test system, and its topology diagram is as follows: Figure 4 shown.

[0066] According to the current grid data disclosure rules, the benchmark data combination is set as shown in Table 1, where, in this example, the node characteristics are composed of the blocking components of the node type, voltage level and node marginal electricity price, and the edge characteristics include important line flow data, transformer flow data, line capacity and shadow price.

[0067] Table 1 Description of known benchmark data

[0068] In order to simulate the load fluctuation over time, the load data published by the 2012 Global Energy Forecasting Competition is used to normalize the real-time load to obtain the load fluctuation trend, and the fluctuation is multiplied by the active power value of the system node to obtain the real-time active data of the node. With a sampling interval of one hour, a total of 10,000 samples are generated, which are divided into training set, validation set, and test set in a ratio of 8:1:1.

[0069] To evaluate the inference effect, this embodiment uses the following two indicators: (a) Weighted Mean Absolute Percentage Error (wMAPE): used to evaluate relative error. In the actual operation of the power grid, due to the dynamic changes in load levels and power generation output, there may be situations where the flow values ​​of individual power grid lines are very small at certain times. In view of this feature, this embodiment selects wMAPE rather than the traditional MAPE as the relative error evaluation indicator. This is because when the actual flow value is close to zero, MAPE may produce a relative error that tends to infinity, affecting the reliability of the overall evaluation index; the calculation formula of wMAPE is:

[0070] in, and Grid lines Actual and estimated values ​​of power flow.

[0071] (b) Mean Absolute Error (MAE): It is used to evaluate the absolute error between the inferred value and the actual value. The calculation formula is:

[0072] The baseline method is set as Graph Attention Network (GAT) to verify the effectiveness of the proposed method.

[0073] First Experiment: Inferring Threat Assessment The model was trained on the IEEE 14-node test system under the conditions of the benchmark data combination to infer the power flow value of the entire network line. The experimental results on the test set show that this embodiment achieves a wMAPE of 5.16% and a MAE of 0.2694, and exhibits extremely high inference accuracy at multiple times, which reflects that the line power flow data has a significant leakage risk caused by inference from public data. In contrast, if the baseline method is selected, its wMAPE is 13.51% and MAE is 0.7056, which shows that this application has a significant performance advantage over the baseline method in line power flow inference tasks.

[0074] Second experiment: Identification of important low-sensitivity / public data during inference In order to evaluate the threat of different types of low-sensitivity data to line power flow inference and to identify key low-sensitivity data in the inference process, this embodiment designs five groups of different data combinations for comparative experiments, as shown in Table 2.

[0075] Table 2 Description of known data

[0076] Table 3 Line power flow inference error under different known data combinations for 14-bus system

[0077] For the IEEE 14-bus test system, the known data is adjusted to train the model multiple times, and the inference accuracy of the line flow is obtained as shown in Table 3. The analysis is as follows: Experimental results show that our method consistently outperforms the baseline method in terms of inference accuracy. Specifically, our method achieves lower wMAPE values, ranging from 5.16% to 11.67%, for all data combinations, while the baseline method achieves significantly higher wMAPE values, ranging from 13.51% to 65.83%. This performance advantage stems from our method's better ability to learn edge features.

[0078] The analysis found that the introduction of various types of low-sensitivity data would increase the risk of leakage of highly sensitive data. By comparing the results of combination 1 with combinations 2 to 4, it can be seen that different types of low-sensitivity data can improve the inference accuracy of line flow to varying degrees. Specifically, when combination 1 only knows part of the line flow, the wMAPE inferred by this method is 11.67% and the MAE is 0.6097. When combination 2 is compared with combination 1, when the node type and voltage level are additionally known, the wMAPE of this method is reduced to 6.97% and the MAE is reduced to 0.3639, which shows that the power grid construction parameters play a significant role in the task of power grid line flow inference, which is highly consistent with the physical characteristics of the line flow flowing from the power generation node to the load node, indicating that the system operator should avoid disclosing too many power grid construction parameters such as the distribution of power generation / load nodes and voltage levels. The results of combination 5 further confirmed this, in which the wMAPE of this method was further reduced to 5.16% and the MAE was reduced to 0.2695, indicating that the coordinated use of multiple low-sensitivity data can produce a cumulative effect, resulting in a significant increase in the threat of leakage of highly sensitive data.

[0079] In summary, this embodiment verifies the superiority of this method in power grid power flow information inference threat assessment.

[0080] Example 5 This embodiment provides a power grid line flow information inference threat assessment system, including: A power line flow information inference value acquisition unit is used to acquire node characteristics, edge characteristics and power line topology characteristics from pre-public data of the power market, input the node characteristics, edge characteristics and power line topology characteristics into a pre-built power line flow information inference model, and obtain the power line flow information inference value; The threat level assessment unit is used to calculate the error between the inferred value of the power line flow information and the preset label, and assess the threat level of power line flow information leakage according to the error, wherein: Pre-built power line flow information inference models include: Feature projection module, used to perform linear transformation on input node features and edge features; A graph representation learning module is used to update the output of the feature projection module in combination with the grid topology features; A deep feature extraction module is used to sequentially concatenate, transform and merge the outputs of the graph representation learning module; The inference module is used to perform dimension mapping on the output of the deep feature extraction module and output the inferred value of the power grid line flow information corresponding to the pre-public data.

[0081] Example 6 This embodiment provides a computer device, including: a memory, used to store a computer program; and a processor, used to implement the steps of a computer method when executing the computer program.

[0082] When the processor executes the computer program, the steps of the above computer method are implemented.

[0083] Alternatively, when the processor executes the computer program, the functions of each module in the above system are realized. The computer device may be a computing device such as a desktop computer, a notebook, a PDA, and a cloud server. The computer device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that the above are examples of computer devices and do not constitute a limitation on computer devices. The computer device may include more components than the above, or a combination of certain components, or different components. For example, the computer device may also include input and output devices, network access devices, buses, etc.

[0084] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the computer device, and uses various interfaces and lines to connect various parts of the entire computer device.

[0085] The memory may be used to store the computer programs and / or modules, and the processor implements various functions of the computer device by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory.

[0086] The memory may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SmartMediaCard, SMC), a secure digital (SecureDigital, SD) card, a flash card (FlashCard), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0087] Example 7 This embodiment further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method described above are implemented.

[0088] If the module / unit integrated in the computer system is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0089] Based on such understanding, the present invention implements all or part of the processes in the above method, and can also be completed by instructing related hardware through a computer program, and the computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of the above computer method can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or preset intermediate form, etc.

[0090] The computer-readable storage medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0091] It should be noted that the content contained in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable storage media do not include electrical carrier signals and telecommunication signals.

[0092] Example 8 This embodiment provides a computer product, which includes a computer program, and the computer program is stored in a computer-readable storage medium; a processor of a computer device reads the computer program from the computer-readable storage medium, and the processor executes the computer program, so that the computer device can execute the method in Embodiment 1, which will not be repeated here.

[0093] It should be noted that a person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods.

[0094] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for assessing power grid line flow information inference threat, characterized in that: The following steps are involved: Obtain node features, edge features, and grid topology features from pre-public data of the power market, input the node features, edge features, and grid topology features into a pre-built grid line power flow information inference model, and obtain grid line power flow information inference values; The error value between the power line flow information inference value and the preset label is calculated, and the threat level of power line flow information leakage is evaluated according to the error value. The pre-built power line flow information inference model includes: Feature projection module, used to perform linear transformation on input node features and edge features; A graph representation learning module is used to update the output of the feature projection module in combination with the grid topology features; A deep feature extraction module is used to sequentially concatenate, transform and merge the outputs of the graph representation learning module; The inference module is used to perform dimension mapping on the output of the deep feature extraction module and output the inferred value of the power grid line flow information corresponding to the pre-public data.

2. A method for assessing power grid line flow information inference threats according to claim 1, characterized in that: The loss function of the pre-constructed power grid line flow information inference model is constructed based on prior knowledge combined with penalty terms constrained by the physical characteristics of the power grid.

3. A method for assessing power grid line flow information inference threats according to claim 1 or 2, characterized in that: The loss function of the pre-built power line flow information inference model for: in, All are non-negative weight coefficients; is the mean square error between the inferred value and the actual value; Penalty term for violating line capacity constraints; The penalty term for violating the shadow price constraint.

4. A method for assessing power grid line flow information inference threat according to claim 1, characterized in that: The graph represents the expression of the learning module: in, Represents the concatenation operation of features; They are multi-layer perceptrons with activation functions PReLU that match the pre-activation value, node feature, and edge feature dimensions respectively; and They are two linear layers with activation function Sigmoid and a single scalar output; It is the BatchNorm layer; For the Layer Node Features; For the Layer Node Features; For the Layer Edge Features; For the node The set of connected nodes; are the pre-activation values ​​of the features; For the Layer Node Features; No. Layer Edge Features; is a collection of nodes; is the set of edges; For the Characteristics of layer nodes; For the Features of layer edges.

5. A method for assessing power grid line flow information inference threats according to claim 1, characterized in that: The deep feature extraction module is spliced ​​by the following formula: in, It is the concatenation of node features and edge features output by the graph representation learning module; Nodes output by the graph representation learning module Features; Nodes output by the graph representation learning module Features; is the edge output by the graph representation learning module Features; Represents the concatenation operation of features; is the set of edges; Nodes output by the graph representation learning module Features and nodes Features and edges The features after splicing.

6. A method for assessing power grid line flow information inference threats according to claim 1, characterized in that: After evaluating the threat level of power grid line flow information leakage according to the error value, the method further includes: The error value between the inferred value of power grid line flow information and the preset label is used to identify the data in the public data of the power market that has a key impact on the inference of power grid line flow information.

7. A power grid line flow information inference threat assessment system, characterized in that: include: A power flow inference value acquisition unit is used to acquire node characteristics, edge characteristics and power grid topology characteristics from pre-public data of the power market, input the node characteristics, edge characteristics and power grid topology characteristics into a pre-built power grid line power flow information inference model, and obtain the power grid line power flow information inference value; The threat level assessment unit is used to calculate the error value between the power line flow information inference value and the preset label, and assess the power line flow information leakage threat level according to the error value, wherein the pre-built power line flow information inference model includes: Feature projection module, used to perform linear transformation on input node features and edge features; A graph representation learning module is used to update the output of the feature projection module in combination with the grid topology features; A deep feature extraction module is used to sequentially concatenate, transform and merge the outputs of the graph representation learning module; The inference module is used to perform dimension mapping on the output of the deep feature extraction module and output the inferred value of the power grid line flow information corresponding to the pre-public data.

8. A computer device, characterized in that: include: a processor suitable for executing a computer program; A computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by the processor, the method according to any one of claims 1 to 6 is performed.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images