Electric power communication network toughness evaluation method and system considering network topology association

Abstract

The invention belongs to the technical field of electric power communication network risk assessment, and particularly relates to an electric power communication network toughness assessment method and system considering network topology association, and the method comprises the steps: obtaining an electric power communication super network based on network topology; carrying out super-core decomposition on the obtained electric power communication super-network, and calculating the core centrality of the super-network; performing risk prediction on topological nodes of the power communication network by adopting a power communication risk prediction method, and calculating a damage probability and a damage form of the power communication network; determining node importance according to the obtained core centrality, performing node repair by combining the node importance and a damage form, and calculating a performance time history response function of the electric power communication network according to a smooth state of a network topology link; and obtaining a damage toughness evaluation value of the power communication network based on the obtained performance time history response function, and completing the evaluation of the toughness of the power communication network.

Description

Technical Field

[0001] The present invention belongs to the technical field of risk assessment of electric power communication networks, and in particular relates to a method and system for assessing the resilience of electric power communication networks taking into account network topology associations. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Optical fiber communication has become an indispensable and important part of the safe and stable operation of the power system, thus realizing real-time communication of each site and ensuring the normal operation of the power grid automation. With the advancement of the power market reform and the process of electric energy substitution, the surge in social electricity consumption and the development of differentiated innovative businesses have put forward higher requirements for the timeliness and reliability of optical fiber communication. Due to the characteristics of power communication such as wide coverage, multiple links, strong equipment correlation, and huge overall architecture, the current monitoring technology is difficult to identify and warn abnormalities in a timely manner, resulting in high pressure on operation and maintenance, difficult troubleshooting and positioning, and low processing efficiency.

[0004] The traditional "post-processing" communication risk management model is difficult to adapt to the ever-increasing business carrying capacity requirements of the power communication network. Therefore, an effective "pre-prediction and prevention" communication risk management method is urgently needed to provide intelligent operation and maintenance support for the new generation of communication management systems. Therefore, relevant scholars have proposed a power communication risk prediction method, which uses the advantage of graph convolutional neural networks that have stronger expression capabilities for graph structured data, effectively mines the characteristics of power flow propagation in the power grid, identifies the vulnerable links in the network, and has the ability to warn of risks and respond quickly when facing abnormal events, ensuring that the power communication network is in a reliable and stable state for a long time.

[0005] Resilience refers to the ability of a system to absorb external changes and disturbances and recover quickly from them. Examining the ability of the power communication network to withstand damage or quickly recover to normal state helps determine how the power grid performs in natural disasters or man-made attacks, so as to take preventive measures to reduce the impact of hazardous events on power grid operations. Node importance is an important basis for evaluating the resilience of power network damage. Existing methods do not consider the impact of business needs on node importance, resulting in resilience assessment results that cannot meet actual application needs. Summary of the invention

[0006] In order to solve the above problems, the present invention proposes a method and system for evaluating the resilience of power communication networks taking into account network topology associations. It takes providing technical support for network fault prevention and risk reduction as a starting point, and improves the stability and reliability of the power grid from the perspective of considering the differentiated communication needs of different businesses, thereby providing strong technical support for ensuring the security of the power grid and improving the service quality of power grid enterprises.

[0007] According to some embodiments, a first solution of the present invention provides a method for predicting power communication risks, which adopts the following technical solution:

[0008] A method for predicting power communication risks, comprising:

[0009] Acquire a power communication graph, and obtain an attribute information matrix and a local information representation matrix of the power communication graph;

[0010] Performing data augmentation on the acquired power communication graph to obtain an augmented power communication graph, an augmented attribute information matrix, and an augmented local information representation matrix;

[0011] According to the obtained attribute information matrix, local information representation matrix, augmented attribute information matrix and augmented local information representation matrix, combined with the introduction of multi-head attention memory spatiotemporal convolutional neural network, a comprehensive representation matrix and an augmented comprehensive representation matrix are obtained;

[0012] A contrast loss function between the comprehensive representation matrix obtained by using the projection head mapping calculation and the augmented comprehensive representation matrix;

[0013] Based on the comprehensive representation matrix, preliminary risk prediction of power communication nodes is performed and preliminary prediction loss function is calculated;

[0014] The obtained contrast loss function and preliminary prediction loss function are jointly optimized to obtain the risk prediction loss function, thus completing the risk prediction of power communication.

[0015] As a further technical limitation, data augmentation is performed on the constructed power communication graph to obtain an augmented power communication graph and its attribute information matrix; one-dimensional convolution is used to mine local dependency information of the attribute information matrix and the augmented attribute information matrix, respectively. The one-dimensional convolution slides on all attribute information matrices along the time axis to capture time dependency, and obtains a local information representation matrix and an augmented local information matrix for representing the local information of the power communication graph and the augmented local information of the power communication graph, respectively.

[0016] Furthermore, the contrast loss function is related to the comprehensive representation matrix feature sequence, the augmented comprehensive representation matrix feature sequence, the multi-head attention network, and the similarity measurement function; the preliminary prediction loss function is calculated by the binary cross entropy loss; and the risk prediction loss function is the weighted sum of the preliminary prediction loss function and the contrast loss function.

[0017] According to some embodiments, a second solution of the present invention provides a method for evaluating the resilience of a power communication network taking into account network topology association, using the following technical solution:

[0018] A method for evaluating the resilience of a power communication network taking into account network topology associations comprises:

[0019] Obtaining a power communication supernetwork based on network topology;

[0020] Perform hypercore decomposition on the obtained power communication hypernetwork and calculate the core centrality of the hypernetwork;

[0021] The power communication risk prediction method described in the first solution is used to predict the risk of the topological nodes of the power communication network, and the damage probability and damage form of the power communication network are calculated;

[0022] The node importance is determined according to the obtained core centrality, and the node is repaired in combination with the node importance and damage form. The performance time-history response function of the power communication network is calculated according to the smooth state of the network topology link.

[0023] Based on the obtained performance time-history response function, the damage resilience assessment value of the power communication network is obtained, and the assessment of the resilience of the power communication network is completed.

[0024] As a further technical limitation, in the process of obtaining the power communication super network based on network topology, according to the geographical wiring diagram of the power communication network, without considering the distribution network communication system, a power communication network topology model is constructed in which the communication systems of each plant and dispatching center are equivalent to nodes, and the nodes in the power communication network topology model are mapped to nodes in the super network. The combination of nodes in the complete business link is mapped to hyper edges until all nodes and lines are retrieved. A super network is obtained based on all the mapped nodes and hyper edges, that is, a current communication super network based on network topology is obtained.

[0025] As a further technical limitation, based on the risk prediction results of the topological nodes of the power communication network, the vulnerable nodes in the power communication network are identified, and the damage probability of the topological nodes of the power communication network is obtained, and the obtained node damage probability is converted into the damage form of the power communication network based on the Monte Carlo simulation method; the repair budget and repair resources are allocated according to the node importance and the damage form, and the nodes are repaired until the network resilience is restored to a preset state; in the node repair process, the real-time performance of the power communication network is obtained according to the smooth state of the network topological links, and the performance time response function of the power communication network is calculated according to the connectivity state of the power communication links.

[0026] According to some embodiments, a third solution of the present invention provides a power communication network resilience assessment system taking into account network topology association, which adopts the following technical solution:

[0027] A power communication network resilience assessment system taking into account network topology association, comprising:

[0028] An acquisition module configured to acquire a power communication super network based on a network topology;

[0029] A calculation module is configured to perform a hypercore decomposition on the acquired power communication supernetwork and calculate the core centrality of the supernetwork; perform risk prediction of topological nodes of the power communication network using the power communication risk prediction method described in the first solution, and calculate the damage probability and damage form of the power communication network;

[0030] A repair module is configured to determine the node importance according to the obtained core centrality, perform node repair in combination with the node importance and damage form, and calculate the performance time-course response function of the power communication network according to the unimpeded state of the network topology link;

[0031] An evaluation module is configured to obtain an evaluation value of the damage resilience of the electric power communication network based on the obtained performance time-history response function, thereby completing the evaluation of the resilience of the electric power communication network.

[0032] According to some embodiments, a fourth solution of the present invention provides a computer-readable storage medium, which adopts the following technical solution:

[0033] A computer-readable storage medium stores a program thereon, which, when executed by a processor, implements the steps in a method for assessing the resilience of a power communication network taking into account network topology associations as described in the second solution of the present invention.

[0034] According to some embodiments, a fifth solution of the present invention provides an electronic device, which adopts the following technical solution:

[0035] An electronic device comprises a memory, a processor and a program stored in the memory and running on the processor, wherein when the processor executes the program, the steps in a method for assessing the resilience of a power communication network taking into account network topology association as described in the second solution of the present invention are implemented.

[0036] According to some embodiments, a sixth solution of the present invention provides a computer program product, which adopts the following technical solution:

[0037] A computer program product includes software code, wherein the program in the software code executes the steps in a method for assessing the resilience of a power communication network taking into account network topology association as described in the second embodiment of the present invention.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] Based on power communication data, the present invention performs data augmentation on the original data graph, uses one-dimensional convolution to mine the local dependency information of the original graph and the augmented graph, and uses a spatiotemporal graph convolutional neural network with a multi-head attention mechanism to capture the global dependency of the data and generate a comprehensive representation matrix; uses a projection head to map features to a contrastive learning feature space, combines contrast loss and prediction loss to optimize model parameters, thereby improving the performance of power communication risk prediction;

[0040] Starting from the perspective of business needs, the present invention uses hypergraph theory to model the power network based on business, and performs hypercore decomposition to generate power unit importance, identify key nodes that affect network stability and propagation characteristics from a business perspective, and then realize repair resource scheduling, reducing the waste of repair resources; combined with the main and backup routing schemes and priority weights of the business, the performance time-history response function curve is drawn, and the dynamic evaluation results of the power network damage resilience are output, which helps to enhance the flexibility of taking preventive measures or response strategies and improve the resilience of the power communication network. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The drawings in the specification that constitute a part of this embodiment are used to provide a further understanding of this embodiment. The schematic embodiments of this embodiment and their descriptions are used to explain this embodiment and do not constitute improper limitations on this embodiment.

[0042] Figure 1 This is a flow chart of the power communication risk prediction method in Embodiment 1 of the present invention;

[0043] Figure 2 This is a flow chart of constructing a business-based power super network in Embodiment 2 of the present invention;

[0044] Figure 3 This is a flow chart of a method for evaluating the resilience of a power communication network taking into account network topology association in Embodiment 2 of the present invention;

[0045] Figure 4 This is a structural block diagram of a power communication network resilience assessment system taking into account network topology association in Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0046] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0047] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0048] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0049] In the present invention, terms such as "upper", "lower", "left", "right", "front", "back", "vertical", "horizontal", "side", "bottom", etc. indicate directions or positional relationships based on the directions or positional relationships shown in the accompanying drawings. They are relational words determined only for the convenience of describing the structural relationships of the various parts or elements of the present invention, and do not specifically refer to any part or element in the present invention and should not be understood as limitations on the present invention.

[0050] In the present invention, terms such as "fixed connection", "connected", "connection", etc. should be understood in a broad sense, indicating that it can be fixedly connected, integrally connected or detachably connected; it can be directly connected or indirectly connected through an intermediate medium. For relevant scientific research or technical personnel in this field, the specific meanings of the above terms in the present invention can be determined according to specific circumstances, and they cannot be understood as limitations on the present invention.

[0051] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0052] Embodiment 1

[0053] Embodiment 1 of the present invention introduces a method for predicting power communication risks.

[0054] like Figure 1 A power communication risk prediction method is shown, comprising:

[0055] Collect relevant power communication data for massive power information, pre-process the acquired power communication time series data, and construct a power communication graph;

[0056] The original power communication graph G is augmented and the edges are randomly deleted to obtain the augmented power communication graph G. aug , and the power communication graph G and the augmented power communication graph G are obtained at the same time aug The corresponding attribute information matrix X and augmented attribute information matrix X aug ;

[0057] Based on the power communication graph G and the augmented power communication graph G aug, one-dimensional convolution is used to mine local dependency information. The one-dimensional convolution slides on the attribute information matrix of all nodes along the time axis to capture the time dependency and obtain the local information representation matrix Y and the augmented local information representation matrix Y aug ;

[0058] According to the attribute information matrix X, the local information representation matrix Y, and the augmented attribute information matrix X aug And the augmented local information representation matrix Y aug , a spatiotemporal graph convolutional neural network with a multi-head attention mechanism is used to capture the global dependency of the data, and the comprehensive representation matrix H and the augmented comprehensive representation matrix H are obtained. aug ;

[0059] Based on the comprehensive representation matrix H and the augmented comprehensive representation matrix H aug , introduce the projection head, map the features to the contrastive learning feature space, make the difference between positive and negative samples more obvious, and calculate the contrastive loss function L CL ;

[0060] Use the comprehensive representation matrix H to construct the activation function, predict node risk, and calculate the initial prediction loss function L pred ;

[0061] For the obtained contrast loss function L CL And the initial prediction loss function L pred Through joint optimization, we obtain the risk prediction loss function and complete the risk prediction of power communication.

[0062] As one or more implementation methods, the data of the power communication system of a province of the State Grid is collected, and according to the obtained power communication geographical wiring diagram, the communication system of the high-voltage transmission network of 220kV and above is considered, and the communication system of the distribution network is not considered, and a simplified power communication network topology model is established. Its topology model regards the communication systems of each plant (including power plants, substations, and converter stations) and the dispatching center as equivalent to the same nodes; the communication lines between nodes are regarded as edges, the differences between the communication lines are ignored, and the communication lines are considered to be bidirectional, that is, all edges are undirected edges, and the power communication time series data is constructed as a phaseless and weightless graph, that is, G (y) =(V,E,A,X); where represents a node, n is the number of nodes; E represents an edge; Represents the adjacency matrix of graph G; It represents the attribute information matrix of the node at time t, m is the dimension of the attribute, and the attribute information is divided into node information and channel information. Among them, the node information includes the network packet loss rate, end-to-end communication delay, service type, service volume, transmission time, number of node connections, node type, etc. of the node; the channel information includes the sending node, receiving node, channel number, channel type, channel bandwidth, flow, etc.

[0063] In this embodiment, the preprocessing of the acquired power communication time series data at least includes data cleaning, missing data completion, data definition and storage.

[0064] As one or more implementations, the local information representation matrix Y obtained is Among them, * represents a one-dimensional convolution operation, ⊕ represents an XOR operation, and W y is the convolution kernel, b u represents the bias, δ(·) represents the ReLU activation function, and thus the local information representation matrix Y of the original graph is obtained.

[0065] For the augmented attribute information matrix X aug Perform the same operation to obtain the augmented local information representation matrix Y aug The specific process is consistent with the above process, and this embodiment will not be repeated here.

[0066] As one or more implementation modes, based on the attribute information matrix X and the local information representation matrix Y, a spatiotemporal graph convolutional neural network with a multi-head attention mechanism is used to capture the global dependency of the data, and the spatiotemporal information of the power communication data is fully represented and learned.

[0067] The graph convolution method is Among them, A is the adjacency matrix of the power communication graph, and to introduce the node characteristics, let Degree Matrix W (l) is the weight matrix of the lth layer; δ(·) is the ReLU activation function.

[0068] In this embodiment, the multi-head attention mechanism is an extended form of the attention mechanism, which can enhance the model's expressiveness and diversity for different representation spaces. Each head of the multi-head attention is an independent attention mechanism, which processes inputs in parallel and generates its own attention output. Its single-layer input is a node attribute information matrix. The output is represented as a new attribute information feature set, where x i and x′ i Has different dimensions; The attention coefficient after softmax normalization is Among them, k∈N(i) represents the neighboring nodes of node i, the learnable weight matrix W realizes the transformation of m-dimensional input features to m′-dimensional output features; α is the learnable weight vector; LeakyReLU is the nonlinear activation function.

[0069] Combined with the attention coefficient α ij, perform neighborhood feature aggregation, σ(·) represents the eLU activation function, and its aggregation method is x′ i =σ(∑ j∈N(i) α ij Wx j ); For a self-attention layer with P attention heads (P>>2), the average method is used as the output, that is, The output of a spatiotemporal convolution block is obtained by using the attribute information matrix X and the local information representation matrix Y, that is, After stacking two spatiotemporal convolution blocks to fuse the features from the spatial and temporal domains, the comprehensive representation matrix H of the power communication graph is obtained.

[0070] Performing the same operation on the augmented power communication graph, we can obtain the augmented comprehensive representation matrix H aug The specific process is consistent with the above process, and this embodiment will not be repeated here.

[0071] As one or more implementations, based on the comprehensive representation matrix H and the augmented comprehensive representation matrix H aug , introduces a projection head to map features to the contrastive learning feature space, making the difference between positive and negative samples more obvious. For a given feature sequence Z 0 =[H,A] and Specifically:

[0072] Z l =LayerNorm(Z l-1 +MHSA(Z l-1 ))

[0073] Z l =LayerNorm(Z l +FFN(Z l-1 ))

[0074]

[0075]

[0076] Among them, l represents the number of stacked layers, MHSA() represents multi-head self-attention network, FFN() represents feedforward network, and LayerNorm() represents layer normalization.

[0077] The contrast loss function is calculated by the InfoNCE loss function, that is Here, sim(·) represents the similarity metric function and τ represents the temperature hyperparameter, which is used to adjust the smoothness of the distribution.

[0078] As one or more implementation methods, the risk prediction of the power communication node is performed according to the comprehensive representation matrix H, and the result prediction is performed by inputting the sigmoid layer, that is, the risk prediction result of the power communication node for Among them, FC() represents a fully connected operation.

[0079] This embodiment uses binary cross entropy loss to calculate the initial prediction loss function L pred ,Right now in represents the predicted risk probability of the i-th node, y i is the true label.

[0080] As one or more implementations, the contrast loss function L CL And the initial prediction loss function L pred Joint optimization is performed to ensure that the model has good generalization performance and prediction accuracy while learning feature representation, that is, ω c (t) = 1-ω p (t); L(t) = ω p (t)L pred (t)+ω c (t)L CL (t); where ω * (t) is the loss weight at time t, β is the hyperparameter that controls the degree of loss weight update, and m ω is the momentum coefficient hyperparameter.

[0081] After completing the power communication risk prediction, this embodiment outputs the prediction results of the experimental sample set, compares them with the actual situation, and feeds back to update the underlying data information, thereby continuously optimizing the data weight values ​​in the model and continuously improving the power communication risk prediction.

[0082] This embodiment uses power communication data as the basis, performs data augmentation on the original data graph, uses one-dimensional convolution to mine the local dependency information of the original graph and the augmented graph, and adopts a spatiotemporal graph convolutional neural network that introduces a multi-head attention mechanism to capture the global dependency of the data and generate a comprehensive representation matrix; uses a projection head to map features to a contrastive learning feature space, combines contrast loss and prediction loss to optimize model parameters, and thus improves the performance of power communication risk prediction.

[0083] Embodiment 2

[0084] Embodiment 1 of the present invention introduces a method for evaluating the resilience of a power communication network taking into account network topology associations.

[0085] like Figure 3 A method for evaluating the resilience of a power communication network taking into account network topology association is shown, comprising:

[0086] According to the hypergraph theory, a business-based power hypernetwork is constructed, whose nodes are power units represented by substations and dispatching centers in the power communication network, and whose hyperedges are the collection of all nodes in a complete business link;

[0087] According to the constructed business-based power super network, the super core decomposition is carried out, the core centrality of the nodes is calculated, and the importance of the power unit is obtained;

[0088] Conduct vulnerability analysis on nodes based on risk prediction methods, obtain node damage probability, and determine the damage form of the power network;

[0089] According to the damage form of the power network, the repair resources are allocated using the order of importance of the power units, the damaged nodes are repaired, and the iterative process is performed until the resource consumption reaches the upper limit or the control time is reached;

[0090] According to the input repair conditions, check the smooth status of the main route and backup route of the service, set the performance time response function (PRF) from the dual dimensions of service and structure, and obtain the real-time performance of each service;

[0091] Based on the real-time performance of each service, the service priority is used to weight the total performance of the power network at that moment and draw the PRF curve;

[0092] Based on the drawn PRF curve, the damage resilience assessment value of the power network is determined to complete the network resilience assessment.

[0093] As one or more implementation methods, the process of constructing a service-based power super network in this embodiment is as follows: Figure 2 As shown, specifically:

[0094] (1) Based on the geographical wiring diagram of the power communication network of a province of the State Grid, a simplified power communication network topology model is established by considering the communication system of the high-voltage transmission network of 220kV and above, and ignoring the communication system of the distribution network. The topology model regards the communication systems of each plant (including power plants, substations, and converter stations) and the dispatching center as equivalent to the same nodes; the communication lines between nodes are regarded as edges, the differences between the communication lines are ignored, and the communication lines are considered to be bidirectional, that is, all edges are undirected edges. The topology model is abstracted as an undirected graph G = (V, E), where V is the node set and E is the edge set. Define the set of service routes carried in the network in Indicates the mth routing solution for service s

[0095] (2) Based on the power communication network topology model established in (1), its nodes are mapped to nodes in the super network, and the set of nodes in a complete service link is mapped to a hyper edge. Define the service-based power super network as H = (V′, E′), where V′ is the node set, which can be obtained by V′ = V; E′ is the edge set, and each hyper edge e∈E′ consists of the nodes it connects, that is, Traversing a complete business chain If the node v i and v j In the same business link, it is added to the corresponding hyperedge until all nodes and lines are retrieved.

[0096] As one or more implementations, a hypergraph is a high-order generalization of a network, in which nodes can interact in hyperedges of arbitrary size, and high-order interactions produce new structures and phenomena, so that the characteristics of the network can be better represented by a hypergraph. Based on the constructed business-based power hypernetwork, a hypercore decomposition is performed to help identify key nodes that affect the stability and propagation characteristics of the power communication network, so as to more centrally study the core parts of the network. The greater the importance of the power unit, the greater the influence of the representative node on the overall power communication network at the business level. The core centrality of the hypergraph is used to characterize the importance of the node, and the specific method is as follows:

[0097] Set the parameters of the hypercore decomposition: the number of hyperedges k and the hyperedge degree c, then all nodes in the (k,c)-hypercore belong to at least k hyperedges with hyperedge degree greater than or equal to c. Then define the node v i The c-shell index is C c (i) = k, indicating node v i Connects k hyperedges with degree c. S(k,c) represents all hyperedges that satisfy C c (i) = k nodes. Therefore, the node importance is defined as its core centrality I(i) in the hypergraph, that is, Where M is the maximum value of the hyperedge degree in the hypergraph; g(c) is an arbitrary weight function, indicating that high-order effects of different scales have different influences. Usually, g(c) can be simply set to 1. The maximum value of k for which the shell S(k,c) is not empty.

[0098] As one or more implementation methods, the power communication risk prediction method in Example 1 is used to predict the risk of each node, identify the vulnerable links in the power network, and obtain the damage probability of the node. Using the obtained node damage probability, the Monte Carlo simulation method is used to convert the damage probability of each unit into the damage form of the network.

[0099] In this embodiment, the higher the importance of a node, the stronger its influence in the network, and the more serious the adverse consequences caused by damage. In the process of damage repair, due to the large number of network nodes and limited repair resources, priority should be given to nodes with high importance. Based on the determined damage form of the power network, specific repair budgets b and repair resources r are allocated according to the order of the obtained node importance I(i), and the nodes are repaired. Iterate multiple times until the resource consumption reaches the upper limit or the control time is reached, so that the network resilience is restored to the best achievable state.

[0100] This embodiment performs restoration process simulation based on the input repair conditions; checks each service link set P s The smooth state of the business can obtain the real-time performance of the business s (t), where F s (t) represents the performance time response function (PRF) of service s in the power communication network at time t, that is, Among them, P s Represents the routing set of service s, Indicates P s The currently unblocked links in W reflect the connectivity status of business links in the structural dimension; s is the set of power supply and demand node pairs in the sub-network involved in the business, Q ω is the power transmission from point to ω, Z ω It is the equivalent impedance between point pairs ω, that is, the impedances of various electrical equipment in the power grid, such as generators, transformers, transmission lines, etc., are combined and calculated according to certain rules. An equivalent impedance is used to represent the resistance of the entire power grid or a part of the power grid to the current under specific operating conditions, which reflects the transmission efficiency of the power grid from a functional dimension.

[0101] Based on the real-time performance of each business s (t), calculate the total performance F(t) of the power communication network at time t, where θ s Represents the priority weight coefficient of service s, that is, F(t) = ∑θ s F s (t).

[0102] Record data and draw network performance time response function (PRF) curve to obtain dynamic performance of network performance.

[0103] According to the drawn PRF curve, the damage resilience assessment value of the power network is calculated to complete the assessment of network resilience. The damage resilience of the power network can be expressed as Where F(t) is the PRF of the power communication network; N(t) is the efficiency function of the network during normal communication, which is generally simplified to a constant; t 0 is the starting control time, t RTo end the control time.

[0104] From the perspective of business needs, this embodiment uses hypergraph theory to model the power network based on business, and performs hypercore decomposition to generate power unit importance, identify key nodes that affect network stability and propagation characteristics from a business perspective, and then realize repair resource scheduling, reducing the waste of repair resources; combined with the main and backup routing schemes and priority weights of the business, draw the performance time-history response function curve, and output the dynamic evaluation results of the power network damage resilience, which helps to enhance the flexibility of taking preventive measures or response strategies and improve the resilience of the power communication network.

[0105] Embodiment 3

[0106] Embodiment 2 of the present invention introduces a power communication network resilience assessment system taking into account network topology associations.

[0107] like Figure 4 A power communication network resilience assessment system considering network topology association is shown, comprising:

[0108] An acquisition module configured to acquire a power communication super network based on a network topology;

[0109] A calculation module is configured to perform a hypercore decomposition on the acquired power communication supernetwork and calculate the core centrality of the supernetwork; perform risk prediction of topological nodes of the power communication network using the power communication risk prediction method described in the first embodiment, and calculate the damage probability and damage form of the power communication network;

[0110] A repair module is configured to determine the node importance according to the obtained core centrality, perform node repair in combination with the node importance and damage form, and calculate the performance time-course response function of the power communication network according to the unimpeded state of the network topology link;

[0111] An evaluation module is configured to obtain an evaluation value of the damage resilience of the electric power communication network based on the obtained performance time-history response function, thereby completing the evaluation of the resilience of the electric power communication network.

[0112] The detailed steps are the same as those of the power communication network resilience assessment method taking into account network topology association provided in Example 2, and will not be repeated here.

[0113] Embodiment 4

[0114] Embodiment 4 of the present invention provides a computer-readable storage medium.

[0115] A computer-readable storage medium stores a program thereon, which, when executed by a processor, implements the steps in a method for evaluating the resilience of a power communication network taking into account network topology association as described in Embodiment 2 of the present invention.

[0116] The detailed steps are the same as those of the power communication network resilience assessment method taking into account network topology association provided in Example 2, and will not be repeated here.

[0117] Embodiment 5

[0118] A fifth embodiment of the present invention provides an electronic device.

[0119] An electronic device comprises a memory, a processor and a program stored in the memory and running on the processor, wherein when the processor executes the program, the steps in a method for assessing the resilience of a power communication network taking into account network topology association as described in Embodiment 2 of the present invention are implemented.

[0120] The detailed steps are the same as the power communication network resilience assessment method taking into account network topology association provided in Example 2, and will not be repeated here.

[0121] Embodiment 6

[0122] Embodiment 6 of the present invention provides a computer program product.

[0123] A computer program product includes software code, wherein the program in the software code executes the steps in a method for assessing the resilience of a power communication network taking into account network topology association as described in Embodiment 2 of the present invention.

[0124] The detailed steps are the same as those of the power communication network resilience assessment method taking into account network topology association provided in Example 2, and will not be repeated here.

[0125] The above description is only a preferred embodiment of the present embodiment and is not intended to limit the present embodiment. For those skilled in the art, the present embodiment may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present embodiment shall be included in the protection scope of the present embodiment.

Claims

1. A method for predicting power communication risks, characterized in that: include: Acquire a power communication graph, and obtain an attribute information matrix and a local information representation matrix of the power communication graph; Performing data augmentation on the acquired power communication graph to obtain an augmented power communication graph, an augmented attribute information matrix, and an augmented local information representation matrix; According to the obtained attribute information matrix, local information representation matrix, augmented attribute information matrix and augmented local information representation matrix, combined with the introduction of multi-head attention memory spatiotemporal convolutional neural network, a comprehensive representation matrix and an augmented comprehensive representation matrix are obtained; A contrast loss function between the comprehensive representation matrix obtained by using the projection head mapping calculation and the augmented comprehensive representation matrix; Based on the comprehensive representation matrix, preliminary risk prediction of power communication nodes is performed and preliminary prediction loss function is calculated; The obtained contrast loss function and preliminary prediction loss function are jointly optimized to obtain the risk prediction loss function, thus completing the risk prediction of power communication.

2. A method for predicting power communication risks as claimed in claim 1, characterized in that: Data augmentation is performed on the constructed power communication graph to obtain the augmented power communication graph and its attribute information matrix. One-dimensional convolution is used to mine the local dependency information of the attribute information matrix and the augmented attribute information matrix respectively. The one-dimensional convolution slides on all attribute information matrices along the time axis to capture the time dependency, and the local information representation matrix and augmented local information matrix used to represent the local information of the power communication graph and the augmented local information of the power communication graph are obtained respectively.

3. A method for predicting power communication risks as claimed in claim 2, characterized in that: The contrast loss function is related to the comprehensive representation matrix feature sequence, the augmented comprehensive representation matrix feature sequence, the multi-head attention network, and the similarity measurement function; the preliminary prediction loss function is calculated by the binary cross entropy loss; the risk prediction loss function is the weighted sum of the preliminary prediction loss function and the contrast loss function.

4. A method for evaluating the resilience of a power communication network taking into account network topology association, characterized in that: include: Obtaining a power communication supernetwork based on network topology; Perform hypercore decomposition on the obtained power communication hypernetwork and calculate the core centrality of the hypernetwork; Using the power communication risk prediction method as described in any one of claims 1 to 3 to perform risk prediction of topological nodes of the power communication network, and calculate the damage probability and damage form of the power communication network; The node importance is determined according to the obtained core centrality, and the node is repaired in combination with the node importance and damage form. The performance time-history response function of the power communication network is calculated according to the smooth state of the network topology link. Based on the obtained performance time-history response function, the damage resilience assessment value of the power communication network is obtained, and the assessment of the resilience of the power communication network is completed.

5. A method for evaluating the resilience of a power communication network taking into account network topology association as claimed in claim 4, characterized in that: In the process of obtaining the power communication super network based on network topology, according to the geographical wiring diagram of the power communication network, without considering the distribution network communication system, a power communication network topology model is constructed in which the communication systems of various power plants and dispatching centers are equivalent to nodes, the nodes in the power communication network topology model are mapped to the nodes in the super network, and the combination of nodes in the complete business link is mapped to hyper edges until all nodes and lines are retrieved. A super network is obtained according to all the mapped nodes and hyper edges, that is, a current communication super network based on network topology is obtained.

6. A method for evaluating the resilience of a power communication network taking into account network topology association as claimed in claim 4, characterized in that: According to the risk prediction results of the topological nodes of the power communication network, the vulnerable nodes in the power communication network are identified, and the damage probability of the topological nodes of the power communication network is obtained. Based on the Monte Carlo simulation method, the obtained node damage probability is converted into the damage form of the power communication network; the repair budget and repair resources are allocated according to the node importance and damage form, and the nodes are repaired until the network resilience is restored to the preset state; During the node repair process, the real-time performance of the power communication network is obtained according to the smooth status of the network topology links, and the performance time-history response function of the power communication network is calculated based on the connectivity status of the power communication links.

7. A power communication network resilience assessment system taking into account network topology association, characterized in that: include: An acquisition module configured to acquire a power communication super network based on a network topology; A calculation module, configured to perform a super-core decomposition on the acquired power communication super-network and calculate the core centrality of the super-network; Using the power communication risk prediction method as described in any one of claims 1 to 3 to perform risk prediction of topological nodes of the power communication network, and calculate the damage probability and damage form of the power communication network; A repair module is configured to determine the node importance according to the obtained core centrality, perform node repair in combination with the node importance and damage form, and calculate the performance time-history response function of the power communication network according to the unimpeded state of the network topology link; An evaluation module is configured to obtain an evaluation value of the damage resilience of the electric power communication network based on the obtained performance time-history response function, thereby completing the evaluation of the resilience of the electric power communication network.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of a method for evaluating the resilience of a power communication network taking into account network topology associations as described in any one of claims 4 to 6 are implemented.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the program, the steps of a method for evaluating the resilience of a power communication network taking into account network topology associations are implemented as described in any one of claims 4-6.

10. A computer program product comprising software code, characterized in that The program in the software code executes the steps of a method for evaluating the resilience of a power communication network taking into account network topology association as described in any one of claims 4-6.

Images