Power grid fault risk determination method and device, program product and electronic equipment

Abstract

The invention discloses a power grid fault risk determination method and device, a program product and electronic equipment, and relates to the technical field of electric power, and the method comprises the steps: firstly, determining M fault vectors according to the operation data of a power grid, enabling the M fault vectors to be used for representing M fault types corresponding to the power grid, and then updating an initial model graph according to the M fault vectors, the target model diagram is used for representing the connection state of the L devices after the power grid breaks down, then determining a feature vector corresponding to each device in the L devices according to the target model diagram and a first preset model, the feature vectors are used for representing the fault occurrence probability of the corresponding equipment and the importance degree of the corresponding equipment in the power grid, and finally determining the fault risk of the power grid according to the L feature vectors. The technical problem of low determination accuracy of the power grid fault risk caused by determining the power grid fault risk based on expert experience in the prior art is solved.

Description

Technical Field

[0001] The present application relates to the field of electric power technology, and in particular to a method, device, program product and electronic equipment for determining power grid failure risk. Background Art

[0002] In the process of building new power systems, the safety and efficiency of grid dispatching and operation have become key factors affecting the quality and efficiency of operation of various business chains. Traditional power system operation and risk assessment rely on expert experience for manual decision-making and lack intelligent support. When there are a large number of devices in the power grid and the connection relationship between devices is complex, the results of the assessment of grid failure risks based on expert experience will be inaccurate, resulting in the technical problem of low efficiency in determining grid failure risks in existing technologies.

[0003] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0004] The present application provides a method, device, program product and electronic device for determining power grid failure risk, so as to at least solve the technical problem of low accuracy of determining power grid failure risk caused by determining power grid failure risk based on expert experience in the prior art.

[0005] According to one aspect of the present application, a method for determining a power grid failure risk is provided, comprising: determining M fault vectors based on operating data of the power grid, wherein the operating data is used to characterize the operating status of L devices included in the power grid, and the M fault vectors are used to characterize the M types of faults corresponding to the power grid, and L and M are both positive integers; updating an initial model graph based on the M fault vectors to obtain a target model graph, wherein the initial model graph is used to characterize the connection status of the L devices at the current moment, and the target model graph is used to characterize the connection status of the L devices after a power grid failure occurs; determining a feature vector corresponding to each of the L devices based on the target model graph and a first preset model, wherein the feature vector is used to characterize the probability of failure of the corresponding device and the importance of the corresponding device in the power grid, and the first preset model is used to simulate the propagation path of the fault in the power grid; determining the failure risk of the power grid based on the L feature vectors.

[0006] Optionally, in the process of determining M fault vectors based on the operating data of the power grid, the method for determining the power grid fault risk also includes: preprocessing the operating data to obtain first operating data, wherein the preprocessing is at least used to remove noise data and abnormal data in the operating data; performing data cleaning on the first operating data to obtain target operating data, wherein the data cleaning is at least used to unify the format of the first operating data; inputting the target operating data into a second preset model to obtain M fault vectors, wherein the second preset model is a neural network model obtained by training N historical fault records of the power grid, and N is a positive integer.

[0007] Optionally, in the process of updating the initial model diagram according to M fault vectors to obtain the target model diagram, the method for determining the power grid failure risk also includes: obtaining a first matrix and a second matrix corresponding to the initial model diagram, wherein the first matrix is ​​used to characterize the topological connection structure between L devices at the current moment, and the second matrix is ​​used to characterize the connection strength between the L devices at the current moment, and the connection strength is at least used to characterize the physical distance and line impedance between interconnected devices; updating the first matrix and the second matrix according to the M fault vectors, and determining the target model diagram based on the updated first matrix and the second matrix.

[0008] Optionally, in the process of determining the characteristic vector corresponding to each of the L devices based on the target model diagram and the second preset model, the method for determining the power grid failure risk also includes: obtaining a third matrix and a fourth matrix corresponding to the target model diagram, wherein the third matrix is ​​used to characterize the topological connection structure between the L devices after a power grid failure occurs, and the fourth matrix is ​​used to characterize the connection strength between the L devices after a power grid failure occurs; inputting the third matrix and the fourth matrix into the first preset model to obtain the characteristic vector corresponding to each device.

[0009] Optionally, before inputting the target operating data into the second preset model, the method for determining the power grid fault risk also includes: preprocessing and data cleaning N historical fault records to obtain N target fault records; obtaining a conditional label corresponding to each of the N target fault records, wherein the conditional label is used to characterize the number of devices and fault types that have failed in the target fault record; classifying the N target fault records according to the conditional labels to obtain P record sets, wherein the conditional labels corresponding to any two target fault records in the record sets are the same; training an initial generator and an initial discriminator based on random noise and the P record sets to obtain P sub-models, wherein each of the P sub-models is used to generate predicted fault records belonging to different conditional labels, and the predicted fault records are used to characterize the type of faults that may occur in the equipment in the power grid at a future moment; and using the P sub-models as the second preset model.

[0010] Optionally, the conditional label is any one of a first conditional label, a second conditional label and a third conditional label, wherein the first conditional label is used to characterize that a single device in the power grid has failed; the second conditional label is used to characterize that at least two devices in the power grid have failed at the same time; the third conditional label is used to characterize that the fault type of the device in the power grid is any one of a short circuit fault type, an overload fault type, a ground fault type or a disconnection fault type.

[0011] Optionally, in the process of determining the fault risk of the power grid based on L eigenvectors, the method for determining the fault risk of the power grid also includes: determining the fault impact value and weight value corresponding to each device based on the L eigenvectors, wherein the fault impact value is used to characterize the probability of a failure of the device at a future moment, and the weight value is used to characterize the importance of the device in the power grid; performing weighted summation of the fault impact value and weight value corresponding to each device in the L devices to obtain a risk value of the power grid, wherein the risk value of the power grid is used to characterize the magnitude of the operating risk of the power grid.

[0012] According to another aspect of the present application, a computer program product is further provided, in which a computer program is stored, wherein when the computer program is run, the computer program product is controlled to execute any one of the above-mentioned methods for determining power grid failure risks.

[0013] According to another aspect of the present application, an electronic device is also provided, wherein the electronic device includes one or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by one or more processors, the one or more processors implement any of the above-mentioned methods for determining the risk of power grid failure.

[0014] According to another aspect of the present application, a device for determining a power grid fault risk is also provided, comprising: a first determination unit, configured to determine M fault vectors based on operating data of the power grid, wherein the operating data is used to characterize the operating status of L devices included in the power grid, and the M fault vectors are used to characterize the M fault types corresponding to the power grid, and L and M are both positive integers; an updating unit, configured to update an initial model diagram based on the M fault vectors to obtain a target model diagram, wherein the initial model diagram is used to characterize the connection status of the L devices at the current moment, and the target model diagram is used to characterize the connection status of the L devices after a fault occurs in the power grid; a second determination unit, configured to determine a feature vector corresponding to each of the L devices based on the target model diagram and a first preset model, wherein the feature vector is used to characterize the probability of a corresponding device failing and the importance of the corresponding device in the power grid, and the first preset model is used to simulate the propagation path of the fault in the power grid; a third determination unit, configured to determine the fault risk of the power grid based on the L feature vectors.

[0015] In the present application, first, M fault vectors are determined based on the operating data of the power grid, wherein the operating data is used to characterize the operating status of the L devices included in the power grid, and the M fault vectors are used to characterize the M fault types corresponding to the power grid, and L and M are both positive integers. Afterwards, the present application updates the initial model diagram based on the M fault vectors to obtain a target model diagram, wherein the initial model diagram is used to characterize the connection status of the L devices at the current moment, and the target model diagram is used to characterize the connection status of the L devices after a fault occurs in the power grid. Then, the present application determines the feature vector corresponding to each device in the L devices based on the target model diagram and the first preset model, wherein the feature vector is used to characterize the probability of failure of the corresponding device and the importance of the corresponding device in the power grid, and the first preset model is used to simulate the propagation path of the fault in the power grid. Finally, the present application determines the fault risk of the power grid based on the L feature vectors.

[0016] It can be seen from the above content that when facing a power grid with a complex connection structure, the present application first collects the operating data of the power grid. After that, the present application predicts M types of faults that may occur in the power grid in the future (i.e., M fault vectors) based on the operating data of the power grid. After that, the present application determines the initial model diagram based on the actual connection status between multiple devices in the power grid, and updates the initial model diagram based on the M fault vectors, thereby achieving the purpose of predicting the connection status between L devices after the power grid fails (i.e., the target model diagram). At the same time, the present application can feedback the operating status of the power grid in the future through the target model diagram, thereby improving the accuracy of the subsequent assessment of the power grid failure risk based on the target model diagram.

[0017] Then, the present application simulates the real-time propagation path of power grid faults in the power grid through the first preset model, and analyzes the target model graph through the first prediction model to obtain the probability of failure of each device and the importance of each device in the power grid (i.e., L eigenvectors). Finally, the present application evaluates the power grid risk by using the L eigenvectors determined by the first preset model, thereby achieving the purpose of intelligently evaluating the fault risk of the power grid through the first preset model, thereby improving the accuracy of determining the fault risk of the power grid.

[0018] It can be seen that the present application achieves the technical effect of improving the accuracy of determining power grid failure risks by intelligently evaluating power grid risks based on L characteristic vectors determined by the first preset model, thereby solving the technical problem of low accuracy in determining power grid failure risks caused by determining power grid failure risks based on expert experience in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0020] Figure 1 is a flow chart of an optional method for determining a power grid failure risk according to an embodiment of the present application;

[0021] Figure 2 is a flowchart of another optional method for determining a power grid failure risk according to an embodiment of the present application;

[0022] Figure 3 is a schematic diagram of an optional device for determining a power grid failure risk according to an embodiment of the present application;

[0023] Figure 4 is a schematic diagram of an optional electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] It should also be noted that the relevant information (including but not limited to information for display and analysis) and data (including but not limited to power grid operation data) involved in this application are all information and data authorized by the user or fully authorized by all parties. For example, an interface is set between this system and the relevant user or organization. Before obtaining relevant information, it is necessary to send an acquisition request to the aforementioned user or organization through the interface, and obtain relevant information after receiving the consent information fed back by the aforementioned user or organization.

[0027] In addition, the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant information and data involved in this application comply with the relevant laws, regulations and standards of the relevant regions, and necessary confidentiality measures have been taken and do not violate public order and good morals. In addition, this application provides corresponding operation entrances for users to choose to agree to authorization or refuse authorization. If the user chooses to refuse authorization, he / she will enter the corresponding expert decision-making process.

[0028] In an optional embodiment, a method for handling power grid faults based on safety analysis is provided, which scans the equipment in the power grid to obtain fault events, then simulates faults and adjusts the power grid topology, performs static safety calculations, generates load transfer strategies, and then selects the optimal strategy for fault handling based on the optimization model. This embodiment relies on a predefined strategy library and static safety calculations, and has a limited degree of intelligence. It mainly performs safety calculations and load transfers based on static data, and it is difficult to reflect the latest status of the power grid in real time. In addition, the simulated faults in this embodiment are mainly based on the actions of the associated fault points, and cannot fully cover all fault scenarios.

[0029] In an optional embodiment, a green power tracing method based on regional power grid flow topology analysis is provided, which simplifies the power grid model, establishes an electrical flow calculation model, enters relevant data, calculates the node power flow direction and shunt coefficient, calculates the line path using a depth-first search algorithm, and verifies and outputs a green power distribution matrix. This embodiment generates a static power distribution matrix based on pre-entered power output, substation, and line flow data, lacks real-time and dynamic update capabilities, and this embodiment involves the generation and calculation of multiple matrices, such as unit injection matrix, node flow matrix, and node load matrix, etc., and requires frequent adjustment of the calculation model to adapt to new data.

[0030] According to an embodiment of the present application, an embodiment of a method for determining a power grid failure risk is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0031] The present application provides a system for determining a power grid failure risk (hereinafter referred to as a determination system) for executing a method for determining a power grid failure risk in the present application, wherein the determination system can be implemented in software or a combination of software and hardware.

[0032] In an optional embodiment, Figure 1 is a flow chart of an optional method for determining a power grid failure risk according to an embodiment of the present application, such as Figure 1 As shown, the method comprises the following steps:

[0033] Step S101, determining M fault vectors according to operation data of the power grid.

[0034] In step S101, the operation data is used to characterize the operation status of L devices included in the power grid, and the M fault vectors are used to characterize the M fault types corresponding to the power grid, where L and M are both positive integers.

[0035] Optionally, the operating status of the L devices includes at least actual operating parameters and device status of all devices in the power grid, wherein the actual operating parameters of the devices include at least the current, voltage and power of the devices, and the device status includes at least normal operating status, shutdown status and maintenance status, to determine that the system can monitor the health status and performance information of all devices in the power grid through operating data.

[0036] Step S102, updating the initial model graph according to the M fault vectors to obtain a target model graph.

[0037] In step S102, the initial model graph is used to represent the connection status of L devices at the current moment, and the target model graph is used to represent the connection status of the L devices after a power grid failure occurs.

[0038] Optionally, the determination system first obtains the topological connection structure and connection strength between all devices (i.e., L devices) included in the power grid, wherein the topological connection structure is used to characterize the connection relationship between the devices, and the connection strength is used to characterize the physical distance and line impedance between the interconnected devices. Afterwards, the determination system generates L nodes based on the L devices, wherein the L nodes correspond one-to-one to the L devices. Then, the determination system determines the initial adjacency matrix (corresponding to the first matrix) and the initial weight matrix (corresponding to the second matrix) corresponding to the L nodes based on the topological connection structure and the connection strength between the L devices. Finally, the determination system generates an initial model graph based on the initial adjacency matrix and the initial weight matrix corresponding to the L nodes.

[0039] Optionally, for a power grid containing L devices, the size of the corresponding initial adjacency matrix A is L*L, and the element A in the initial adjacency matrix A is ijRepresents the connection relationship between the i-th device and the j-th device in L nodes, where i and j are both positive integers less than or equal to L.

[0040] Alternatively, if there is a direct physical connection between the i-th device and the j-th device (for example, a transmission line connecting two substations), then A ij =1; otherwise, A ij =0.

[0041] Optionally, the connection relationship in the power grid is usually undirected, so the adjacency matrix is ​​a symmetric matrix, that is, A ji = ij A.

[0042] Optionally, the initial weight matrix is ​​an extended matrix of the initial adjacency matrix, used to further refine the connection strength or connection importance between L devices, wherein the weight matrix can not only characterize whether there is a connection relationship between the devices, but also quantify the connection characteristics (such as physical distance and line impedance).

[0043] Optionally, the connection characteristics also include the transmission capacity of the transmission line between the devices, wherein the element W in the initial weight matrix ij There is a positive correlation between the value and the transmission capacity between the i-th device and the j-th device.

[0044] Optionally, the determination system can also determine the element W in the initial weight matrix according to the inverse of the physical distance between the i-th device and the j-th device. ij Value, where the element W in the initial weight matrix ij There is a negative correlation between the value and the physical distance between the i-th device and the j-th device.

[0045] Optionally, the determination system can also determine the element W in the initial weight matrix according to the line impedance or other electrical characteristics between the i-th device and the j-th device. ij value.

[0046] Optionally, when there is no direct connection between the i-th device and the j-th device, W ij =0.

[0047] For example, the construction method of the initial weight matrix W is as follows:

[0048] Optionally, the determination system first selects an indicator representing the connection strength according to the needs of power grid analysis, such as transmission capacity, physical distance, impedance, etc. or historical failure rate.

[0049] Alternatively, assume that the transmission capacity C of the transmission line between the devices is used ijTo define the connection strength between devices, the element W of the initial weight matrix ij =C ij For example, if the transmission capacity of the transmission line between the i-th device and the j-th device is 100MW, then the corresponding element W in the weight matrix is ij =100.

[0050] Optionally, in order to unify the connection strength values ​​corresponding to different dimensions of the transmission capacity of the transmission line, the determination system also needs to normalize the connection strength. For example, if the transmission capacity of the transmission line between the i-th device and the j-th device is between 0 and 1000 MW, the determination system will W ij Normalized to between 0 and 1, the calculation formula of the normalized connection strength is shown in the following formula (1):

[0051]

[0052] In the above formula (1), C max It is the maximum capacity value among the transmission capacities of the transmission lines of all devices.

[0053] Optionally, the determination system updates the initial adjacency matrix and the initial weight matrix corresponding to the initial model graph according to the M fault vectors, thereby achieving the purpose of obtaining a target model graph that can characterize the connection status of L devices after a power grid fault occurs.

[0054] Step S103, determining a feature vector corresponding to each of the L devices according to the target model graph and the first preset model.

[0055] In step S103, the characteristic vector is used to characterize the probability of failure of the corresponding device and the importance of the corresponding device in the power grid, and the first preset model is used to simulate the propagation path of the fault in the power grid.

[0056] Optionally, the determination system first pre-creates a first preset model based on the target model graph and M fault vectors, wherein the first preset model is preferably a GNN (Graph Neural Network) model. Thereafter, the determination system inputs the target adjacency matrix and weight matrix corresponding to the target model graph into the first preset model. The first preset model can utilize the data relationship corresponding to the graph structure of the target model graph, thereby achieving the purpose of capturing the interactions and dependencies between the L nodes in the target model graph.

[0057] Step S104: determining the fault risk of the power grid according to the L characteristic vectors.

[0058] Optionally, after determining the fault risk of the power grid according to the L characteristic vectors, the determination system can further classify the levels of the fault risk, so as to further evaluate the power grid fault according to the fault risk level of the power grid.

[0059] It can be seen from the above content that when facing a power grid with a complex connection structure, the present application first collects the operating data of the power grid. After that, the present application predicts M types of faults that may occur in the power grid in the future (i.e., M fault vectors) based on the operating data of the power grid. After that, the present application determines the initial model diagram based on the actual connection status between multiple devices in the power grid, and updates the initial model diagram based on the M fault vectors, thereby achieving the purpose of predicting the connection status between L devices after the power grid fails (i.e., the target model diagram). At the same time, the present application can feedback the operating status of the power grid in the future through the target model diagram, thereby improving the accuracy of the subsequent assessment of the power grid failure risk based on the target model diagram.

[0060] Then, the present application simulates the real-time propagation path of power grid faults in the power grid through the first preset model, and analyzes the target model graph through the first prediction model to obtain the probability of failure of each device and the importance of each device in the power grid (i.e., L eigenvectors). Finally, the present application evaluates the power grid risk by using the L eigenvectors determined by the first preset model, thereby achieving the purpose of intelligently evaluating the fault risk of the power grid through the first preset model, thereby improving the accuracy of determining the fault risk of the power grid.

[0061] It can be seen that the present application achieves the technical effect of improving the accuracy of determining power grid failure risks by intelligently evaluating power grid risks based on L characteristic vectors determined by the first preset model, thereby solving the technical problem of low accuracy in determining power grid failure risks caused by determining power grid failure risks based on expert experience in the prior art.

[0062] In an optional embodiment, the determination system first preprocesses the operating data to obtain first operating data, wherein the preprocessing is at least used to remove noise data and abnormal data in the operating data. Thereafter, the determination system performs data cleaning on the first operating data to obtain target operating data, wherein the data cleaning is at least used to unify the format of the first operating data. Then, the determination system inputs the target operating data into a second preset model to obtain M fault vectors, wherein the second preset model is a neural network model obtained by training N historical fault records of the power grid, and N is a positive integer.

[0063] Optionally, before inputting the operation into the second preset model, the determination system obtains target operation data by preprocessing and cleaning the operation data, and then inputs the target operation data into the second preset model, thereby achieving the purpose of improving the data quality input into the second preset model, and the preprocessing operation can reduce the amount of data that needs to be processed by the second preset model, thereby improving the subsequent operation efficiency of the second preset model.

[0064] Optionally, the second preset model is a GAN (Generative Adversarial Networks) model, which includes at least multiple sub-models (corresponding to P sub-models), each of which is composed of two neural networks, a generator and a discriminator. The GAN model controls the generator and the discriminator to compete and learn against each other, so that the generator can gradually generate more realistic data samples, and the discriminator will continue to improve its discrimination ability. Ultimately, the data samples generated by the generator will become closer and closer to the real data, thereby achieving the purpose of more accurately predicting the types of faults that may occur in the power grid in the future.

[0065] In an optional embodiment, the determination system first obtains a first matrix and a second matrix corresponding to an initial model diagram, wherein the first matrix is ​​used to characterize the topological connection structure between L devices at the current moment, and the second matrix is ​​used to characterize the connection strength between the L devices at the current moment, and the connection strength is at least used to characterize the physical distance and line impedance between interconnected devices. Then, the determination system updates the first matrix and the second matrix based on M fault vectors, and determines the target model diagram based on the updated first matrix and the second matrix.

[0066] For example, assuming that L is 5, and the value of a fault vector among the M fault vectors is [0, 1, 0, 1, 0], then the fault vector indicates that the second device and the fourth device in the power grid have failed.

[0067] Optionally, the determination system can modify the element values ​​in the first matrix and the second matrix according to the M fault vectors, so that the fault state of the power grid is reflected through the modified first matrix and the second matrix.

[0068] Optionally, if the ith device in the power grid fails, the connections between the ith device and other devices fail. In the initial adjacency matrix (corresponding to the first matrix), it is determined that the system can set all element values ​​in the row (i.e., the ith row) and column (i.e., the ith column) corresponding to the ith node in the initial adjacency matrix to 0, which means that the ith node has no valid connection with any other node.

[0069] Optionally, if the connection line between the i-th device and the j-th device in the power grid fails, then in the initial weight matrix (corresponding to the second matrix), it is determined that the system will A ij and A ji Both are set to 0.

[0070] For example, suppose L is 4, A ij =1 means there is a connection between node i and node j, A ij =0 means no connection, the initial adjacency matrix is ​​as shown below:

[0071]

[0072] When the second device in the power grid fails, the initial adjacency matrix is ​​updated according to the fault vector [0,1,0,0]. All elements in the second row and second column of the initial adjacency matrix are updated to 0. The updated adjacency matrix is ​​shown in the following matrix:

[0073]

[0074] Optionally, assuming that the connection line between the first device and the third device in the power grid fails, the W in the initial weight matrix is ​​replaced by 13 and W 31 Updated to 0.

[0075] In an optional embodiment, the determination system first obtains a third matrix and a fourth matrix corresponding to the target model diagram, wherein the third matrix is ​​used to characterize the topological connection structure between L devices after a power grid failure, and the fourth matrix is ​​used to characterize the connection strength between the L devices after a power grid failure. Then, the determination system inputs the third matrix and the fourth matrix into the first preset model to obtain a feature vector corresponding to each device.

[0076] Optionally, the first preset model is preferably a GNN (Graph Neural Network) model, wherein the training process of the first preset model includes the following steps:

[0077] (1) Training data preparation:

[0078] The training data of the first model is the operation data of the power grid at different times. Based on the operation data of the power grid at different times, multiple sets of node feature matrices and node adjacency matrices corresponding to the power grid are determined, wherein each row of the node feature matrix represents a device feature vector of a device, and the device feature vector is used to characterize at least the real-time operating parameters of the device (such as voltage, current, temperature), device type, historical fault data, topological location and device health status information. The node adjacency matrix can reflect the connection relationship between devices in the power grid.

[0079] (2) Constructing the initial graph neural network model:

[0080] The core of the GNN model is GCN (Graph Convolutional Layer). GCN can aggregate the information of each node and its neighboring nodes through the adjacency matrix and feature matrix to perform fault propagation analysis.

[0081] (3) Define the loss function:

[0082] The GNN model aims to predict the failure probability of equipment and defines the cross entropy loss function as shown in the following formula (2):

[0083]

[0084] In the above formula (2), y i is the true label of node i, is the predicted value of the model.

[0085] (4) Model training:

[0086] Forward propagation: Input multiple sets of node feature matrices and node adjacency matrices, calculate the embedding representation of each node through the graph convolution layer, and obtain the predicted value of the node (such as the failure probability of the equipment) through the output layer.

[0087] Backpropagation and parameter update: The gradient of the loss function with respect to the model parameters is calculated using the backpropagation algorithm.

[0088] Use an optimization algorithm (such as stochastic gradient descent) to update the model parameters to minimize the loss function value.

[0089] Iterative training: Repeat the process of forward propagation and back propagation to iteratively update the model parameters until the loss function converges or the preset number of training times is reached.

[0090] (5) Model evaluation:

[0091] Use the validation set to evaluate the performance of the model, adjust the hyperparameters (e.g. learning rate, size of hidden layers), and finally, use the test set to evaluate the generalization ability of the model.

[0092] Optionally, after determining that the system inputs the third matrix and the fourth matrix into the first preset model, the multi-layer GCN obtained by training the GNN model performs fault propagation analysis and feature extraction on the equipment in the power grid through the following formula (3).

[0093]

[0094] In the above formula (3), H (l)is the node feature matrix of the i-th layer, is the normalized form of the node adjacency matrix, W (l) is the weight matrix of the lth layer, and σ is the activation function.

[0095] Optionally, in each layer of GCN, the GNN model achieves the purpose of mapping device features into a new feature space by controlling the multiplication of the node feature matrix and the weight matrix, wherein the weight matrix enables the GNN to extract high-level features of the device layer by layer through linear transformation in each layer, and then understand the relationship between devices in the power grid. The high-level features of the device can determine the propagation mode and impact range of the fault in the network.

[0096] Afterwards, the GNN model uses the trained GNN model to simulate the propagation path of the fault in the power grid and predicts the fault impact value (corresponding to the feature vector) of each node, where the fault impact value h i The calculation formula is shown in the following formula (4):

[0097]

[0098] In the above formula (4), It is the feature vector of the i-th node output by the final layer of the GNN model.

[0099] Optionally, the fault impact value can not only represent the impact of a node failure on the power grid system, but also represent the probability that the node cannot operate normally.

[0100] Optionally, the fault impact value can also represent the degree of impact of the node fault on other nodes in the power grid or the entire power grid.

[0101] Optionally, in the output layer of GNN, the fault impact value of each node is calculated based on the input features and graph structure information. The GNN model finally calculates the fault impact value of each node through multi-layer propagation and feature extraction. The fault impact value can be used to predict the failure probability of the node, that is, the possibility of the node failing in the future under the current conditions. The fault impact value can also measure the impact of the node on the entire power grid when a fault occurs.

[0102] Optionally, if the fault impact value represents the fault probability, then a higher fault impact value means a greater possibility of failure at the node; if the fault impact value represents the degree of impact, then a higher fault impact value means a greater potential threat of the node's failure to the power grid.

[0103] In an optional embodiment, the determination system first preprocesses and cleans N historical fault records to obtain N target fault records, and then the determination system obtains a conditional label corresponding to each target fault record in the N target fault records, wherein the conditional label is used to characterize the number of faulty devices and the type of fault in the target fault record, and then the determination system classifies the N target fault records according to the conditional labels to obtain P record sets, wherein the conditional labels corresponding to any two target fault records in the record sets are the same, and the determination system trains an initial generator and an initial discriminator based on random noise and the P record sets to obtain P sub-models, wherein each of the P sub-models is used to generate predicted fault records belonging to different conditional labels, and the predicted fault records are used to characterize the type of faults that may occur to equipment in the power grid at a future moment, and finally, the determination system uses the P sub-models as the second preset models.

[0104] Optionally, the second preset model adopts a multi-GAN architecture design, and the second preset model includes multiple GAN sub-models (ie, P sub-models), and each GAN sub-model focuses on generating a certain specific type of fault samples (ie, fault vectors).

[0105] For example, assuming that P is equal to 3, the GAN1 sub-model is used to generate fault samples of a single-node fault type, the GAN2 sub-model is used to generate fault samples of a double-node fault type, and the GAN3 sub-model is used to generate fault samples of a triple-node fault type. Each GAN sub-model contains its own generator and discriminator.

[0106] Optionally, the input random noise vector for GAN sub-model training is z i and single node failure label y i , where y1 represents a single node failure, y2 represents a double node failure, and y3 represents a multi-node failure. The output is the node failure sample G i (z,y).

[0107] Optionally, the objective function of the generator of the GAN sub-model is as shown in the following formula (5):

[0108]

[0109] Optionally, the objective function of the discriminator of the GAN sub-model is as shown in the following formula (6):

[0110]

[0111] Optionally, during the training process of the GAN sub-model, the determination system first initializes the parameters of the GAN sub-model, and then randomly selects a batch of fault samples and corresponding conditional labels to train the generator and discriminator in the GAN sub-model. During training, the determination system uses the generator to generate fault samples, and combines the generated fault samples with real fault samples, and updates the parameters of the discriminator in the GAN sub-model based on the combined samples, thereby achieving the purpose of distinguishing real fault samples from generated virtual fault samples, and then updates the parameters of the generator to improve the authenticity of the generated fault samples, and repeats the above training steps until the GAN sub-model converges.

[0112] Optionally, the conditional label is any one of a first conditional label, a second conditional label and a third conditional label, wherein the first conditional label is used to characterize that a single device in the power grid has failed; the second conditional label is used to characterize that at least two devices in the power grid have failed at the same time; the third conditional label is used to characterize that the fault type of the device in the power grid is any one of a short circuit fault type, an overload fault type, a ground fault type or a disconnection fault type.

[0113] For example, if two switch devices in the power grid fail at the same time, the conditional tags corresponding to the two switch devices are the second conditional tags.

[0114] Optionally, the determination system generates fault samples belonging to different conditional labels according to different GAN sub-models, thereby reducing the complexity of the fault sample types generated by each GAN sub-model and improving the training efficiency of the GAN sub-model.

[0115] In an optional embodiment, the determination system first determines the fault impact value and weight value corresponding to each device based on L eigenvectors, wherein the fault impact value is used to characterize the probability of a device failure in the future, and the weight value is used to characterize the importance of the device in the power grid. Thereafter, the determination system performs weighted summation of the fault impact value and weight value corresponding to each of the L devices to obtain a risk value of the power grid, wherein the risk value of the power grid is used to characterize the magnitude of the operating risk of the power grid.

[0116] Optionally, in the process of determining the risk value of the power grid by the system, the calculation formula used for weighted summing of the fault impact value and the weight value corresponding to each device in the L devices is as shown in the following formula (7):

[0117]

[0118] In the above formula (7), R represents the risk value of the power grid, α i is the weight of the ith node.

[0119] Optionally, after a new device is introduced into the power grid, the system is determined to be able to fine-tune the GAN model using the operating data of the latest device, thereby ensuring that the generator of the GAN model can generate fault samples that reflect the current true state of the power grid. After that, the GNN model parameters are updated in real time, and incremental training is performed using the new training data, thereby achieving dynamic updating of the power grid risk value.

[0120] In an optional embodiment, Figure 2 is a flow chart of another optional method for determining the risk of power grid failure according to an embodiment of the present application, such as Figure 2 As shown, the method includes:

[0121] Optionally, the determination system first performs data collection and preprocessing operations, wherein the data collected by the determination system includes at least topological structure data of L devices in the power grid, the connection strength between the L devices, and N historical fault records of the power grid. Thereafter, the determination system performs preprocessing and data cleaning operations on the collected data, thereby achieving the purpose of removing noise and abnormal data in the received data, thereby ensuring the accuracy and consistency of the processed data.

[0122] Then, the determination system generates an initial model graph based on the topological structure data of L devices and the connection strength between the L devices, and obtains a GAN model by training N historical fault records of the power grid. After that, the determination system generates fault samples (i.e., M fault vectors in this application) through the trained GAN model, and the determination system updates the nodes and edges in the initial model graph based on the fault samples to obtain a target model graph. After that, the determination system performs fault propagation analysis based on the target model graph through a pre-created GNN model to obtain a feature vector corresponding to each device. Finally, the determination system determines the power grid risk based on the feature vector corresponding to each device in the L devices.

[0123] Alternatively, existing grid risk assessment methods are mostly based on static data, which cannot reflect the dynamic state of the grid in real time as a whole, and it is difficult to effectively deal with complex topological changes and fault conditions, resulting in insufficient flexibility and low adaptability. When the topological structure of the grid changes or new fault data appears, the models in the existing technology usually require a large number of update calculations, which are computationally intensive and have low robustness, making it difficult to quickly and accurately provide risk assessment and decision support.

[0124] In this application, a multi-architecture GAN model is used to generate different types of fault samples, thereby simulating all possible fault conditions in the power grid, and combining the GNN model to perform fault propagation analysis and risk assessment, which can accurately capture the propagation path and impact range of the fault in the power grid, thereby improving the accuracy and comprehensiveness of risk assessment. Among them, this application uses a multi-architecture GAN model, so that risk assessment of dynamic data can be achieved by only fine-tuning the GNN network parameters, without the need for frequent adjustment of the GAN model, thereby reducing the workload of model maintenance and updating, and improving the flexibility and maintainability of the determination system.

[0125] It can be seen from the above content that when facing a power grid with a complex connection structure, the present application first collects the operating data of the power grid. After that, the present application predicts M types of faults that may occur in the power grid in the future (i.e., M fault vectors) based on the operating data of the power grid. After that, the present application determines the initial model diagram based on the actual connection status between multiple devices in the power grid, and updates the initial model diagram based on the M fault vectors, thereby achieving the purpose of predicting the connection status between L devices after the power grid fails (i.e., the target model diagram). At the same time, the present application can feedback the operating status of the power grid in the future through the target model diagram, thereby improving the accuracy of the subsequent assessment of the power grid failure risk based on the target model diagram.

[0126] Then, the present application simulates the real-time propagation path of power grid faults in the power grid through the first preset model, and analyzes the target model graph through the first prediction model to obtain the probability of failure of each device and the importance of each device in the power grid (i.e., L eigenvectors). Finally, the present application evaluates the power grid risk by using the L eigenvectors determined by the first preset model, thereby achieving the purpose of intelligently evaluating the fault risk of the power grid through the first preset model, thereby improving the accuracy of determining the fault risk of the power grid.

[0127] It can be seen that the present application achieves the technical effect of improving the accuracy of determining power grid failure risks by intelligently evaluating power grid risks based on L characteristic vectors determined by the first preset model, thereby solving the technical problem of low accuracy in determining power grid failure risks caused by determining power grid failure risks based on expert experience in the prior art.

[0128] According to another aspect of the embodiment of the present application, a device for determining a power grid failure risk is also provided. Figure 3 is a schematic diagram of an optional device for determining a power grid failure risk according to an embodiment of the present application, such as Figure 3 As shown, the apparatus for determining a power grid failure risk includes: a first determining unit 301 , an updating unit 302 , a second determining unit 303 and a third determining unit 304 .

[0129] Optionally, a first determination unit is used to determine M fault vectors based on operating data of the power grid, wherein the operating data is used to characterize the operating status of L devices included in the power grid, and the M fault vectors are used to characterize the M fault types corresponding to the power grid, and L and M are both positive integers; an updating unit is used to update the initial model diagram based on the M fault vectors to obtain a target model diagram, wherein the initial model diagram is used to characterize the connection status of the L devices at the current moment, and the target model diagram is used to characterize the connection status of the L devices after a fault occurs in the power grid; a second determination unit is used to determine a feature vector corresponding to each of the L devices based on the target model diagram and a first preset model, wherein the feature vector is used to characterize the probability of failure of the corresponding device and the importance of the corresponding device in the power grid, and the first preset model is used to simulate the propagation path of the fault in the power grid; a third determination unit is used to determine the fault risk of the power grid based on the L feature vectors.

[0130] In an optional embodiment, the first determination unit includes: a preprocessing subunit, a data cleaning subunit, and a first input subunit.

[0131] Optionally, a preprocessing subunit is used to preprocess the operating data to obtain first operating data, wherein the preprocessing is at least used to remove noise data and abnormal data in the operating data; a data cleaning subunit is used to clean the first operating data to obtain target operating data, wherein the data cleaning is at least used to unify the format of the first operating data; an input subunit is used to input the target operating data into a second preset model to obtain M fault vectors, wherein the second preset model is a neural network model obtained by training N historical fault records of the power grid, and N is a positive integer.

[0132] In an optional embodiment, the updating unit includes: a first acquiring subunit and an updating subunit.

[0133] Optionally, a first acquisition subunit is used to acquire a first matrix and a second matrix corresponding to an initial model diagram, wherein the first matrix is ​​used to characterize the topological connection structure between L devices at the current moment, and the second matrix is ​​used to characterize the connection strength between the L devices at the current moment, and the connection strength is at least used to characterize the physical distance and line impedance between interconnected devices; an update subunit is used to update the first matrix and the second matrix based on M fault vectors, and determine the target model diagram based on the updated first matrix and the second matrix.

[0134] In an optional embodiment, the second determining unit includes: a second acquiring subunit and a second input subunit.

[0135] Optionally, the second acquisition subunit is used to acquire a third matrix and a fourth matrix corresponding to the target model diagram, wherein the third matrix is ​​used to characterize the topological connection structure between L devices after a power grid failure, and the fourth matrix is ​​used to characterize the connection strength between the L devices after a power grid failure; the second input subunit is used to input the third matrix and the fourth matrix into the first preset model to obtain a feature vector corresponding to each device.

[0136] In an optional embodiment, the apparatus for determining a power grid failure risk further includes: a processing unit, an acquisition unit, a classification unit, a training unit, and a fourth determination unit.

[0137] Optionally, a processing unit is used to preprocess and clean N historical fault records to obtain N target fault records; an acquisition unit is used to obtain a conditional label corresponding to each target fault record in the N target fault records, wherein the conditional label is used to characterize the number of faulty devices and the type of fault in the target fault record; a classification unit is used to classify the N target fault records according to the conditional labels to obtain P record sets, wherein the conditional labels corresponding to any two target fault records in the record sets are the same; a training unit is used to train an initial generator and an initial discriminator based on random noise and P record sets to obtain P sub-models, wherein each of the P sub-models is used to generate predicted fault records belonging to different conditional labels, and the predicted fault records are used to characterize the type of faults that may occur to equipment in the power grid at a future moment; a fourth determination unit is used to use the P sub-models as second preset models.

[0138] In an optional embodiment, the conditional label is any one of a first conditional label, a second conditional label and a third conditional label, wherein the first conditional label is used to characterize that a single device in the power grid has failed; the second conditional label is used to characterize that at least two devices in the power grid have failed at the same time; and the third conditional label is used to characterize that the fault type of the device in the power grid is any one of a short circuit fault type, an overload fault type, a ground fault type or a disconnection fault type.

[0139] In an optional embodiment, the third determination unit includes: a determination subunit and a weighted summation subunit.

[0140] Optionally, a determination subunit is used to determine a fault impact value and a weight value corresponding to each device based on L eigenvectors, wherein the fault impact value is used to characterize the probability of a device failing in the future, and the weight value is used to characterize the importance of the device in the power grid; a weighted summation subunit is used to perform a weighted summation of the fault impact value and the weight value corresponding to each of the L devices to obtain a risk value of the power grid, wherein the risk value of the power grid is used to characterize the magnitude of the operating risk of the power grid.

[0141] It can be seen from the above content that when facing a power grid with a complex connection structure, the present application first collects the operating data of the power grid. After that, the present application predicts M types of faults that may occur in the power grid in the future (i.e., M fault vectors) based on the operating data of the power grid. After that, the present application determines the initial model diagram based on the actual connection status between multiple devices in the power grid, and updates the initial model diagram based on the M fault vectors, thereby achieving the purpose of predicting the connection status between L devices after the power grid fails (i.e., the target model diagram). At the same time, the present application can feedback the operating status of the power grid in the future through the target model diagram, thereby improving the accuracy of the subsequent assessment of the power grid failure risk based on the target model diagram.

[0142] Then, the present application simulates the real-time propagation path of power grid faults in the power grid through the first preset model, and analyzes the target model graph through the first prediction model to obtain the probability of failure of each device and the importance of each device in the power grid (i.e., L eigenvectors). Finally, the present application evaluates the power grid risk by using the L eigenvectors determined by the first preset model, thereby achieving the purpose of intelligently evaluating the fault risk of the power grid through the first preset model, thereby improving the accuracy of determining the fault risk of the power grid.

[0143] It can be seen that the present application achieves the technical effect of improving the accuracy of determining power grid failure risks by intelligently evaluating power grid risks based on L characteristic vectors determined by the first preset model, thereby solving the technical problem of low accuracy in determining power grid failure risks caused by determining power grid failure risks based on expert experience in the prior art.

[0144] According to another aspect of an embodiment of the present application, a computer program product is further provided, the computer program product comprising a stored computer program, wherein when the computer program is running, the computer program product is controlled to execute any one of the above-mentioned methods for determining power grid failure risks.

[0145] According to another aspect of an embodiment of the present application, there is also provided an electronic device, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute any one of the above-mentioned methods for determining the risk of power grid failure by executing the executable instructions.

[0146] Optionally, Figure 4 is a schematic diagram of an optional electronic device according to an embodiment of the present application, such as Figure 4 As shown, an embodiment of the present application provides an electronic device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, any one of the above-mentioned methods for determining the risk of power grid failure is implemented.

[0147] Optionally, when the processor executes the program, it can implement the following steps: determine M fault vectors based on the operating data of the power grid, wherein the operating data is used to characterize the operating status of L devices included in the power grid, and the M fault vectors are used to characterize the M fault types corresponding to the power grid, and L and M are both positive integers; update the initial model graph based on the M fault vectors to obtain a target model graph, wherein the initial model graph is used to characterize the connection status of the L devices at the current moment, and the target model graph is used to characterize the connection status of the L devices after a fault occurs in the power grid; determine the feature vector corresponding to each of the L devices based on the target model graph and a first preset model, wherein the feature vector is used to characterize the probability of failure of the corresponding device and the importance of the corresponding device in the power grid, and the first preset model is used to simulate the propagation path of the fault in the power grid; determine the fault risk of the power grid based on the L feature vectors.

[0148] Optionally, when the processor executes the program, it can implement the following steps: preprocessing the operating data to obtain first operating data, wherein the preprocessing is at least used to remove noise data and abnormal data in the operating data; performing data cleaning on the first operating data to obtain target operating data, wherein the data cleaning is at least used to unify the format of the first operating data; inputting the target operating data into a second preset model to obtain M fault vectors, wherein the second preset model is a neural network model obtained by training N historical fault records of the power grid, and N is a positive integer.

[0149] Optionally, the processor can implement the following steps when executing the program: obtain a first matrix and a second matrix corresponding to the initial model diagram, wherein the first matrix is ​​used to characterize the topological connection structure between L devices at the current moment, and the second matrix is ​​used to characterize the connection strength between the L devices at the current moment, and the connection strength is at least used to characterize the physical distance and line impedance between interconnected devices; update the first matrix and the second matrix based on M fault vectors, and determine the target model diagram based on the updated first matrix and the second matrix.

[0150] Optionally, when the processor executes the program, it can implement the following steps: obtain a third matrix and a fourth matrix corresponding to the target model diagram, wherein the third matrix is ​​used to characterize the topological connection structure between L devices after a power grid failure, and the fourth matrix is ​​used to characterize the connection strength between the L devices after a power grid failure; input the third matrix and the fourth matrix into the first preset model to obtain a feature vector corresponding to each device.

[0151] Optionally, when the processor executes the program, it can implement the following steps: preprocess and clean N historical fault records to obtain N target fault records; obtain the conditional label corresponding to each target fault record in the N target fault records, wherein the conditional label is used to characterize the number of faulty devices and the type of fault in the target fault record; classify the N target fault records according to the conditional labels to obtain P record sets, wherein the conditional labels corresponding to any two target fault records in the record sets are the same; train the initial generator and the initial discriminator based on random noise and the P record sets to obtain P sub-models, wherein each of the P sub-models is used to generate predicted fault records belonging to different conditional labels, and the predicted fault records are used to characterize the type of faults that may occur to the equipment in the power grid at a future moment; and use the P sub-models as the second preset models.

[0152] Optionally, the conditional label in the process of the processor executing the program is any one of a first conditional label, a second conditional label and a third conditional label, wherein the first conditional label is used to characterize that a single device in the power grid has failed; the second conditional label is used to characterize that at least two devices in the power grid have failed at the same time; the third conditional label is used to characterize that the fault type of the device in the power grid is any one of a short circuit fault type, an overload fault type, a ground fault type or a disconnection fault type.

[0153] Optionally, when the processor executes the program, it can implement the following steps: determine the fault impact value and weight value corresponding to each device based on L feature vectors, wherein the fault impact value is used to characterize the probability of a device failure in the future, and the weight value is used to characterize the importance of the device in the power grid; perform weighted summation of the fault impact value and weight value corresponding to each device in the L devices to obtain the risk value of the power grid, wherein the risk value of the power grid is used to characterize the magnitude of the operating risk of the power grid.

[0154] It can be seen from the above content that when facing a power grid with a complex connection structure, the present application first collects the operating data of the power grid. After that, the present application predicts M types of faults that may occur in the power grid in the future (i.e., M fault vectors) based on the operating data of the power grid. After that, the present application determines the initial model diagram based on the actual connection status between multiple devices in the power grid, and updates the initial model diagram based on the M fault vectors, thereby achieving the purpose of predicting the connection status between L devices after the power grid fails (i.e., the target model diagram). At the same time, the present application can feedback the operating status of the power grid in the future through the target model diagram, thereby improving the accuracy of the subsequent assessment of the power grid failure risk based on the target model diagram.

[0155] Then, the present application simulates the real-time propagation path of power grid faults in the power grid through the first preset model, and analyzes the target model graph through the first prediction model to obtain the probability of failure of each device and the importance of each device in the power grid (i.e., L eigenvectors). Finally, the present application evaluates the power grid risk by using the L eigenvectors determined by the first preset model, thereby achieving the purpose of intelligently evaluating the fault risk of the power grid through the first preset model, thereby improving the accuracy of determining the fault risk of the power grid.

[0156] It can be seen that the present application achieves the technical effect of improving the accuracy of determining power grid failure risks by intelligently evaluating power grid risks based on L characteristic vectors determined by the first preset model, thereby solving the technical problem of low accuracy in determining power grid failure risks caused by determining power grid failure risks based on expert experience in the prior art.

[0157] The above-mentioned embodiments or examples disclosed in the present application are not exhaustive, but are only illustrative of some embodiments or examples, and are not intended to be specific limitations on the scope of protection disclosed in the present application. In the absence of contradiction, each step in a certain embodiment or example in the present application can be implemented as an independent example, and the steps can be combined arbitrarily. For example, the scheme after removing some steps in a certain embodiment or example can also be implemented as an independent example, and the order of the steps in a certain embodiment or example can be arbitrarily exchanged. In addition, the optional methods or optional examples in a certain embodiment or example can be combined arbitrarily; in addition, the various embodiments or examples can be combined arbitrarily, for example, some or all steps of different embodiments or examples can be combined arbitrarily, and a certain embodiment or example can be combined arbitrarily with the optional methods or optional examples of other embodiments or examples.

[0158] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0159] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1A device that provides the functions specified in a block or multiple blocks.

[0160] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0161] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0162] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory. The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0163] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0164] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0165] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0166] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for determining power grid failure risk, characterized in that: include: Determine M fault vectors according to operation data of the power grid, wherein the operation data is used to characterize the operation status of L devices included in the power grid, and the M fault vectors are used to characterize M types of faults corresponding to the power grid, and L and M are both positive integers; An initial model graph is updated according to the M fault vectors to obtain a target model graph, wherein the initial model graph is used to represent the connection status of the L devices at the current moment, and the target model graph is used to represent the connection status of the L devices after a fault occurs in the power grid; Determine a feature vector corresponding to each device in the L devices according to the target model graph and a first preset model, wherein the feature vector is used to characterize the probability of failure of the corresponding device and the importance of the corresponding device in the power grid, and the first preset model is used to simulate the propagation path of the fault in the power grid; The failure risk of the power grid is determined based on the L eigenvectors.

2. The method for determining the risk of power grid failure according to claim 1, characterized in that: Determining M fault vectors according to the operation data of the power grid includes: Preprocessing the operation data to obtain first operation data, wherein the preprocessing is at least used to remove noise data and abnormal data in the operation data; Performing data cleaning on the first operation data to obtain target operation data, wherein the data cleaning is at least used to unify the format of the first operation data; The target operation data is input into a second preset model to obtain M fault vectors, wherein the second preset model is a neural network model obtained by training N historical fault records of the power grid, and N is a positive integer.

3. The method for determining the risk of power grid failure according to claim 1, characterized in that: The initial model graph is updated according to the M fault vectors to obtain a target model graph, including: Obtaining a first matrix and a second matrix corresponding to the initial model graph, wherein the first matrix is ​​used to characterize the topological connection structure between the L devices at the current moment, and the second matrix is ​​used to characterize the connection strength between the L devices at the current moment, and the connection strength is at least used to characterize the physical distance and line impedance between the interconnected devices; The first matrix and the second matrix are updated according to the M fault vectors, and the target model graph is determined according to the updated first matrix and the second matrix.

4. The method for determining the risk of power grid failure according to claim 1, characterized in that: Determining a feature vector corresponding to each device in the L devices according to the target model graph and the second preset model includes: Obtaining a third matrix and a fourth matrix corresponding to the target model graph, wherein the third matrix is ​​used to characterize the topological connection structure between the L devices after the power grid fails, and the fourth matrix is ​​used to characterize the connection strength between the L devices after the power grid fails; The third matrix and the fourth matrix are input into the first preset model to obtain the feature vector corresponding to each device.

5. The method for determining the risk of power grid failure according to claim 2, characterized in that: Before inputting the target operation data into the second preset model, the method for determining the power grid failure risk further includes: Preprocessing and data cleaning the N historical fault records to obtain N target fault records; Obtain a condition tag corresponding to each target fault record in the N target fault records, wherein the condition tag is used to characterize the number of devices and fault types that have failed in the target fault record; Classifying the N target fault records according to the conditional labels to obtain P record sets, wherein the conditional labels corresponding to any two target fault records in the record sets are the same; Training an initial generator and an initial discriminator according to random noise and the P record sets to obtain P sub-models, wherein each of the P sub-models is used to generate a predicted fault record belonging to a different conditional label, and the predicted fault record is used to characterize the type of fault that may occur to the equipment in the power grid at a future moment; The P sub-models are used as the second preset models.

6. The method for determining the risk of power grid failure according to claim 5, characterized in that: The condition tag is any one of a first condition tag, a second condition tag and a third condition tag, wherein the first condition tag is used to indicate that a single device in the power grid has failed; The second condition tag is used to indicate that at least two devices in the power grid fail at the same time; The third condition tag is used to indicate that the fault type of the equipment in the power grid is any one of a short circuit fault type, an overload fault type, a ground fault type or a disconnection fault type.

7. The method for determining the risk of power grid failure according to claim 2, characterized in that: Determining the fault risk of the power grid according to the L characteristic vectors includes: Determine a fault impact value and a weight value corresponding to each device according to the L eigenvectors, wherein the fault impact value is used to characterize the probability of a device failing at a future time, and the weight value is used to characterize the importance of the device in the power grid; A weighted sum is performed on the fault impact value and the weight value corresponding to each device in the L devices to obtain a risk value of the power grid, wherein the risk value of the power grid is used to characterize the magnitude of the operation risk of the power grid.

8. A device for determining a power grid failure risk, characterized in that: include: A first determining unit is configured to determine M fault vectors according to operation data of a power grid, wherein the operation data is used to characterize the operation states of L devices included in the power grid, and the M fault vectors are used to characterize M types of faults corresponding to the power grid, and L and M are both positive integers; an updating unit, configured to update an initial model graph according to the M fault vectors to obtain a target model graph, wherein the initial model graph is used to represent the connection status of the L devices at the current moment, and the target model graph is used to represent the connection status of the L devices after a fault occurs in the power grid; a second determination unit, configured to determine a feature vector corresponding to each device in the L devices according to the target model graph and a first preset model, wherein the feature vector is used to characterize a probability of failure of the corresponding device and an importance of the corresponding device in the power grid, and the first preset model is used to simulate a propagation path of the fault in the power grid; The third determining unit is used to determine the fault risk of the power grid according to the L characteristic vectors.

9. A computer program product, characterized in that The computer program product comprises a computer program, wherein when the computer program is run, the computer program product is controlled to execute the method for determining the power grid failure risk according to any one of claims 1 to 7.

10. An electronic device, characterized in that: It includes one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method for determining the power grid failure risk as described in any one of claims 1 to 7.

Images