Power data transaction risk discovery method, system, equipment and medium

Abstract

The invention provides an electric power data transaction risk discovery method, system and device and a medium, and the method comprises the steps: recognizing key nodes in an electric power data traffic panoramic view based on the connectivity of the electric power data traffic panoramic view formed by electric power data traffic through employing a heuristic algorithm based on iteration tabu search; key nodes are monitored, and abnormal data nodes are found by comparing changes of the key nodes; according to the method, through connectivity analysis of the panoramic view of the power data traffic, key nodes in a service link are identified, protection key points are defined, a data transaction source can be quickly and accurately traced when data transaction occurs, and through implementation of the method, the data security can be improved, and the security of the power data traffic is improved. The structure and the function of a service active link are protected, the serious influence of leakage of sensitive data on the service is prevented, and the problems that a panoramic view angle is lacked and a data transaction node is difficult to discover in the existing data transaction tracing process are solved.

Description

Technical Field

[0001] The present invention relates to the field of power system security, and in particular to a method, system, device and medium for discovering power data abnormality risks. Background Art

[0002] With the opening of online marketing channels, the business data involved covers multiple business types such as business expansion access, electricity bill settlement and asset management. The flow of data in the business interaction process is becoming more and more huge, and the business is complex and changeable. The data security issues and threats it faces are becoming more and more prominent and serious. Once a sensitive data leakage incident caused by external attacks, internal management or incorrect configuration occurs, it is very important to quickly and accurately trace the source of the data change.

[0003] In the existing technology, there is a lack of panoramic perspective in the process of tracing data anomalies, making it difficult to discover the source of data leakage, and it is impossible to grasp the overall operational security status of the business activities of the marketing online cooperation channels. It is difficult to quickly and accurately trace the source of data anomalies when data anomalies occur, and to conduct data anomaly tracing in the power marketing business. Summary of the invention

[0004] In order to solve the problem in the prior art of how to quickly and accurately trace the source of data changes when data changes occur, and to trace data changes in power marketing business, the present invention proposes a method for discovering risks of power data changes, including:

[0005] Based on the connectivity of the panoramic view of power data flow constituted by power data flow, a heuristic algorithm based on iterative taboo search is used to identify the key nodes in the panoramic view of power data flow;

[0006] The key nodes are monitored, and abnormal data nodes are discovered by comparing changes in the key nodes.

[0007] Preferably, the connectivity of the power data flow panoramic view formed by the power data flow adopts a heuristic algorithm based on iterative taboo search to identify key nodes in the power data flow panoramic view, including:

[0008] Based on the connectivity of the panoramic view of power data flow formed by power data flow, an initial matching scheme of key nodes in the panoramic view of power data flow is generated in a greedy adaptive manner;

[0009] Based on the connectivity of the panoramic view of the power data flow, an enhanced strategy based on taboo search is used to iteratively optimize the initial matching solution to identify key nodes in the panoramic view of the power data flow;

[0010] Among them, the connectivity of the panoramic view of the power data flow includes one or more of the following: average path length or clustering coefficient; the characteristics of the connectivity of the panoramic view of the power data flow include one or more of the following: out-degree or in-degree.

[0011] Preferably, the connectivity of the power data flow panoramic view formed by the power data flow is based on the greedy adaptive method to generate an initial matching scheme of key nodes in the power data flow panoramic view, including:

[0012] A1: Set the initial solution of the initial matching scheme of key nodes in the panoramic view of power data flow to an empty set;

[0013] A2: Based on the connectivity of the panoramic view of power data flow constituted by power data flow, a basic reduction method is used to delete all node sets in the panoramic view of power data flow that cannot be key nodes, to obtain a reduced graph, and a greedy algorithm is applied to randomly select a node whose greedy function value exceeds a threshold from the reduced graph to add to the initial solution, wherein the greedy function value includes: the flow size or priority of the node;

[0014] A3: In the reduced graph, randomly select a node, or select a node based on the out-degree or in-degree and according to a heuristic method, remove the node and its connected edges from the reduced graph, and add the node to the initial solution;

[0015] A4: Determine whether the reduction graph is empty. If so, take the current initial solution as the initial matching solution and end. Otherwise, proceed to step A2.

[0016] Preferably, the reduction rules of the basic reduction method include:

[0017] If the in-degree or out-degree of a node is 0, delete the node and its adjacent edges;

[0018] If the in-degree of a node is 1 and there is no self-loop, merge the node into its only predecessor node, or if the out-degree of a node is 1 and there is no self-loop, merge the node into its only successor node;

[0019] If a node has a self-loop, delete the node and its adjacent edges.

[0020] Preferably, the connectivity of the panoramic view of the power data flow is based on the use of a taboo search-based reinforcement strategy to iteratively optimize the initial matching solution to identify key nodes in the panoramic view of the power data flow, including:

[0021] B1: Based on the connectivity of the panoramic view of power data flow constituted by power data flow, a basic reduction method is used to delete all node sets in the panoramic view of power data flow that cannot be key nodes, and a reduced graph is obtained. The search depth threshold and the initial matching scheme are used as inputs, and the initial value of the number of iterations that cannot be optimized is set to 0, and the initial value of the set constituted by the key nodes is the initial matching scheme;

[0022] B2: The number of iterations cannot be optimized to increase by 1, and each node in the reduction graph is randomly sorted to obtain a node sequence;

[0023] B3: sequentially select a node from the node sequence and determine whether the node belongs to the initial matching solution. If so, proceed to step B5; otherwise, proceed to step B4;

[0024] B4: If the node can be directly placed in the initial matching solution, the insertion operator is executed to place the node in the initial matching solution, and based on the average path length and the clustering coefficient, it is determined whether the updated initial matching solution is better than the current set of key nodes. When the updated initial matching solution is better than the current set of key nodes, the set of key nodes is set as the updated initial matching solution and the number of iterations that failed to be optimized is set to 0, and the process goes to step B6;

[0025] B5: Based on the in-degree and out-degree, a scoring formula is used to calculate the neighbor node set of the node. If the neighbor node set is not empty, the exchange operator is executed to replace the node in the initial matching solution that conflicts with the node with the node;

[0026] B6: Determine whether all nodes in the reduced graph have been traversed. If so, execute B7; otherwise, go to B3;

[0027] B7: Determine whether the number of iterations that have not been optimized exceeds the search depth threshold. If so, output the set of current key nodes and end. Otherwise, proceed to step B2.

[0028] Preferably, the scoring formula is expressed as follows:

[0029]

[0030] In the formula, Score C (i) represents the score of node i, represents the in-degree of node i, Represents the out-degree of node i.

[0031] Preferably, judging whether the updated initial matching scheme is better than the current set of key nodes based on the average path length and the clustering coefficient includes:

[0032] Calculate the average path length and clustering coefficient of the updated initial matching solution, as well as the average path length and clustering coefficient of the set of current key nodes respectively;

[0033] If the average path length of the updated initial matching scheme is smaller than the average path length of the current set of key nodes, or the average clustering coefficient of the updated initial matching scheme is larger than the average clustering coefficient of the current set of key nodes, then the updated initial matching scheme is better than the current set of key nodes, where the average clustering coefficient is the average value of the clustering coefficients of each node.

[0034] Preferably, the calculation formula of the average path length is as follows:

[0035]

[0036] In the formula, L represents the average path length, N represents the number of nodes, and l ik Represents the distance from node i to node k.

[0037] Preferably, the clustering coefficient of a node is calculated as follows:

[0038]

[0039] In the formula, C i represents the clustering coefficient of node i, E represents the set of edges, d i represents the degree of node i, v i Represents node i, v j represents node j as a neighbor of node i, E i Represents the number of neighboring edges of node i.

[0040] Preferably, the connectivity of the power data flow panoramic view formed based on the power data flow, before identifying the key nodes in the power data flow panoramic view by using a heuristic algorithm based on iterative taboo search, further includes:

[0041] Abstract the power data flow into a panoramic view of the power data flow consisting of nodes and edges;

[0042] The maximum connected subgraph of the power data flow panoramic view is extracted, and the in-degree and out-degree of the power data flow panoramic view are calculated.

[0043] Based on the same inventive concept, the present application also provides a power data abnormality risk discovery system, including: a key node identification module and a data abnormality risk discovery module;

[0044] The key node identification module is used to identify the key nodes in the power data flow panoramic view based on the connectivity of the power data flow panoramic view formed by the power data flow, using a heuristic algorithm based on iterative taboo search;

[0045] The data change risk discovery module is used to monitor the key nodes and discover abnormal data nodes by comparing the changes in the key nodes.

[0046] Preferably, the key node identification module is specifically used for:

[0047] Based on the connectivity of the panoramic view of power data flow formed by power data flow, an initial matching scheme of key nodes in the panoramic view of power data flow is generated in a greedy adaptive manner;

[0048] Based on the connectivity of the panoramic view of the power data flow, an enhanced strategy based on taboo search is used to iteratively optimize the initial matching solution to identify key nodes in the panoramic view of the power data flow;

[0049] Among them, the connectivity of the panoramic view of the power data flow includes one or more of the following: average path length or clustering coefficient; the characteristics of the connectivity of the panoramic view of the power data flow include one or more of the following: out-degree or in-degree.

[0050] Preferably, the key node identification module generates an initial matching scheme of key nodes in the power data flow panoramic view in a greedy adaptive manner based on the connectivity of the power data flow panoramic view formed by the power data flow, including:

[0051] A1: Set the initial solution of the initial matching scheme of key nodes in the panoramic view of power data flow to an empty set;

[0052] A2: Based on the connectivity of the panoramic view of power data flow constituted by power data flow, a basic reduction method is used to delete all node sets in the panoramic view of power data flow that cannot be key nodes, to obtain a reduced graph, and a greedy algorithm is applied to randomly select a node whose greedy function value exceeds a threshold from the reduced graph to add to the initial solution, wherein the greedy function value includes: the flow size or priority of the node;

[0053] A3: In the reduced graph, randomly select a node, or select a node based on the out-degree or in-degree and according to a heuristic method, remove the node and its connected edges from the reduced graph, and add the node to the initial solution;

[0054] A4: Determine whether the reduction graph is empty. If so, take the current initial solution as the initial matching solution and end. Otherwise, proceed to step A2.

[0055] Preferably, the reduction rules of the basic reduction method in the key node identification module include:

[0056] If the in-degree or out-degree of a node is 0, delete the node and its adjacent edges;

[0057] If the in-degree of a node is 1 and there is no self-loop, merge the node into its only predecessor node, or if the out-degree of a node is 1 and there is no self-loop, merge the node into its only successor node;

[0058] If a node has a self-loop, delete the node and its adjacent edges.

[0059] Preferably, the key node identification module uses a taboo search-based reinforcement strategy to iteratively optimize the initial matching solution based on the connectivity of the power data flow panoramic view, and identifies the key nodes in the power data flow panoramic view, including:

[0060] B1: Based on the connectivity of the panoramic view of power data flow constituted by power data flow, a basic reduction method is used to delete all node sets in the panoramic view of power data flow that cannot be key nodes, and a reduced graph is obtained. The search depth threshold and the initial matching scheme are used as inputs, and the initial value of the number of iterations that cannot be optimized is set to 0, and the initial value of the set constituted by the key nodes is the initial matching scheme;

[0061] B2: The number of iterations cannot be optimized to increase by 1, and each node in the reduction graph is randomly sorted to obtain a node sequence;

[0062] B3: sequentially select a node from the node sequence and determine whether the node belongs to the initial matching solution. If so, proceed to step B5; otherwise, proceed to step B4;

[0063] B4: If the node can be directly placed in the initial matching solution, the insertion operator is executed to place the node in the initial matching solution, and based on the average path length and the clustering coefficient, it is determined whether the updated initial matching solution is better than the current set of key nodes. When the updated initial matching solution is better than the current set of key nodes, the set of key nodes is set as the updated initial matching solution and the number of iterations that failed to be optimized is set to 0, and the process goes to step B6;

[0064] B5: Based on the in-degree and out-degree, a scoring formula is used to calculate the neighbor node set of the node. If the neighbor node set is not empty, the exchange operator is executed to replace the node in the initial matching solution that conflicts with the node with the node;

[0065] B6: Determine whether all nodes in the reduced graph have been traversed. If so, execute B7; otherwise, go to B3;

[0066] B7: Determine whether the number of iterations that have not been optimized exceeds the search depth threshold. If so, output the set of current key nodes and end. Otherwise, proceed to step B2.

[0067] Preferably, the scoring formula in the key node identification module is expressed as follows:

[0068]

[0069] In the formula, Score C (i) represents the score of node i, represents the in-degree of node i, Represents the out-degree of node i.

[0070] Preferably, the key node identification module determines whether the updated initial matching scheme is better than the current set of key nodes based on the average path length and the clustering coefficient, including:

[0071] Calculate the average path length and clustering coefficient of the updated initial matching solution, as well as the average path length and clustering coefficient of the set of current key nodes respectively;

[0072] If the average path length of the updated initial matching scheme is smaller than the average path length of the current set of key nodes, or the average clustering coefficient of the updated initial matching scheme is smaller than the average clustering coefficient of the current set of key nodes, then the updated initial matching scheme is better than the current set of key nodes, where the average clustering coefficient is the average value of the clustering coefficients of each node.

[0073] Preferably, the calculation formula of the average path length in the key node identification module is as follows:

[0074]

[0075] In the formula, L represents the average path length, N represents the number of nodes, and l ik Represents the distance from node i to node k.

[0076] Preferably, the clustering coefficient of the nodes in the key node identification module is calculated as follows:

[0077]

[0078] In the formula, C i represents the clustering coefficient of node i, E represents the set of edges, d i represents the degree of node i, v i Represents node i, vj represents node j as a neighbor of node i, E i Represents the number of neighboring edges of node i.

[0079] Preferably, the system further comprises a feature extraction module of graph connectivity;

[0080] The feature extraction module of the graph connectivity is used to abstract the power data flow into a power data flow panoramic view composed of nodes and edges; extract the maximum connected subgraph of the power data flow panoramic view, and calculate the in-degree and out-degree of the power data flow panoramic view.

[0081] In another aspect, the present application further provides a computing device, comprising: at least one processor and a memory;

[0082] The memory is used to store one or more programs;

[0083] When the one or more programs are executed by the one or more processors, a method for discovering risks of power data changes as described above is implemented.

[0084] On the other hand, the present application also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed, the method for discovering the risk of abnormal changes in electric power data as described above is implemented.

[0085] Compared with the prior art, the present invention has the following beneficial effects:

[0086] The present invention provides a method, system, device and medium for discovering the risk of power data anomalies, including: based on the connectivity of the power data flow panoramic view composed of the power data flow, adopting a heuristic algorithm based on iterative taboo search to identify the key nodes in the power data flow panoramic view; monitoring the key nodes, and discovering abnormal data nodes by comparing the changes of the key nodes; the present invention identifies the key nodes in the business link by performing connectivity analysis on the power data flow panoramic view, clarifies the protection focus, and can quickly and accurately trace the source of the data anomaly when the data anomaly occurs. By implementing the method of the present invention, data security can be improved, the structure and function of the business activity link can be protected, and the leakage of sensitive data can be prevented from causing serious impact on the business, and the problems of lack of panoramic perspective and difficulty in discovering data anomaly nodes in the existing data anomaly tracing process are solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] Figure 1 A schematic diagram of a flow chart of a method for discovering power data anomaly risks provided by the present invention;

[0088] Figure 2 A panoramic view of the power data flow without modification provided by the present invention;

[0089] Figure 3 The present invention provides Figure 2 A panoramic view of power data flow after modification by Rule 1;

[0090] Figure 4 The present invention provides Figure 3 A panoramic view of power data flow after modification by Rule 2;

[0091] Figure 5 The present invention provides Figure 4 A panoramic view of power data flow after modification by Rule 3;

[0092] Figure 6 A schematic diagram of a connected subgraph in an undirected graph provided by the present invention;

[0093] Figure 7 A schematic diagram of the degrees of vertices in an undirected graph provided by the present invention;

[0094] Figure 8 A schematic diagram of the structure of a power data abnormality risk discovery system provided by the present invention;

[0095] Fig. 9 A schematic diagram of the structure of an example of a power data abnormality risk discovery system provided by the present invention;

[0096] Fig.10 The present invention provides a schematic diagram of the structure of an electronic device. DETAILED DESCRIPTION

[0097] In view of the risk identification of data anomalies in the power system, the present invention proposes an analysis method based on a power data flow topology map. In this method, the present invention first generates a data flow topology map based on the actual flow of data in the power system, in which the node represents the source or target of the data flow, and the edge represents the path of the data flow. This power data flow topology map reflects the data transmission relationship of data in the power network, rather than the network topology. By analyzing the connectivity characteristics of the data flow topology map, we can reveal the key nodes and paths in the power system. These features include constructing the maximum connected subgraph and calculating the node degree (i.e., in-degree and out-degree), etc. The changes in these features reflect the stability of the power data flow topology and possible security risks. Analysis based on the data flow topology map can also quickly locate abnormal nodes and paths, and deeply analyze the relationships and dependencies between them to determine the scope and path of risk propagation. By continuously monitoring the set of key nodes in the data flow topology map, we can promptly detect the anomalies of power data and take corresponding security measures.

[0098] This method based on data flow topology diagram provides an effective means for monitoring power system security and identifying risks. It can more accurately reflect the actual flow of power data and identify potential safety risks.

[0099] The technical problem to be solved by the present invention is how to quickly and accurately trace the source of data changes when data changes occur, and to trace data changes in power marketing business.

[0100] Embodiment 1:

[0101] A method for discovering power data anomaly risks, such as Figure 1 As shown, including:

[0102] Step 1: Based on the connectivity of the power data flow panoramic view formed by the power data flow, a heuristic algorithm based on iterative taboo search is used to identify key nodes in the power data flow panoramic view;

[0103] Step 2: Monitor key nodes and find abnormal data nodes by comparing changes in key nodes.

[0104] The purpose of this embodiment is to propose a method for discovering the risk of power data changes based on graph connectivity. This method addresses the complexity and security issues of a large number of user services and business data involved in online cooperation channels for power marketing, and solves the problem of difficulty in tracing sensitive data leakage incidents in the business chain. By constructing a panoramic view of the power data flow through the reachable paths of the business interfaces, and identifying key service interfaces based on graph connectivity path analysis, it is possible to identify key nodes in the business chain, clarify protection priorities, and quickly and accurately trace the source of data changes when they occur. This method can be applied to multiple business types in the power marketing business, including business expansion access, electricity bill settlement, and asset management.

[0105] By implementing this method, data security can be improved, the structure and function of the business activity link can be protected, the leakage of sensitive data can be prevented from causing serious impact on the business, and the problems of lack of panoramic perspective and difficulty in discovering data change nodes in the existing data change tracing process can be solved.

[0106] As an important means of assessing vulnerability, key node identification helps to discover important and vulnerable nodes in the power data center, and achieve accurate assessment of the vulnerability of the power data system.

[0107] The traditional key node detection problem model finds a set of key nodes for a given network, and removing them will significantly reduce the connectivity of the network. In addition, it is difficult for the key nodes in the model to match the actual key nodes of power data.

[0108] Therefore, in response to this problem, this embodiment transforms the key node identification problem in the power data network into the minimum feedback vertex set problem: given a directed graph G = (V, E), where V is a vertex set and E is an edge set, a feedback vertex set is a vertex subset. After removing the subset, there will be no loop in the graph. In this case, the feedback vertex set problem is to find a subset that meets the conditions and has the minimum number of vertices, that is, to remove as few vertices as possible to make the graph free of loops. This embodiment will use a heuristic algorithm based on iterative taboo search to solve this problem.

[0109] The present invention firstly constructs a panoramic view of data flow based on the reachable path of power data business; secondly, analyzes and identifies key nodes based on the graph connectivity path of the panoramic view of power data flow; finally, based on the changes of key nodes in the real-time power data middle platform represented by the graph, determines the source of the changes in marketing business data, completes the traceability of the changes, and improves the ability to quickly identify data changes.

[0110] The present invention helps power marketing 2.0 to form a risk discovery capability for power data changes based on graph connectivity, which can not only meet the data security risk discovery needs of online cooperation channels, but also achieve accurate tracing of change nodes after data changes.

[0111] Step 1 specifically includes:

[0112] 1-1: Generate initial match

[0113] A good initial matching solution can reduce the number of searches and quickly find the optimal match, thereby improving the efficiency of the algorithm. Consider using a greedy adaptive method to generate an initial matching solution (i.e., an initial solution), and the initial value of the initial solution is an empty set.

[0114] 1-1-1: Delete all vertex sets that cannot be key nodes through basic reduction method. The specific reduction method is as follows:

[0115] Rule 1: If the in-degree (out-degree) of a vertex v is 0, then v and all its adjacent edges can be removed from G without losing any optimal feedback vertex in G during the removal process. Then vertex v will be added to the redundant vertex set U r In. Figure 2 As shown, v1-v 10 There are 10 vertices, the in-degree of vertex v2 is 0, and the vertex v 10 The out-degree of is 0, and they and their adjacent edges can be deleted. The modified graph is Figure 3 .

[0116] Rule 2: If the in-degree (out-degree) of vertex v is 1 and there is no self-loop, then vertex v can be merged into the only predecessor (successor) vertex, and the best feedback vertex of G is not lost during the merging process. The merging process is that all edges connected to vertex v are connected to the only predecessor (successor) vertex of v. Then vertex v is also added to the redundant vertex set U r In. Figure 3 As shown in the figure, the vertex v1 with in-degree 1 can be merged with the predecessor vertex v4, and the vertex v7 with out-degree 1 can be merged with the successor vertex v5. The modified graph is Figure 4 .

[0117] Rule 3: If vertex v has a self-loop, then v and all its adjacent edges can be removed from G, and vertex v is added to the redundant key set C. r In. Figure 4 As shown in Figure 2, the vertex v3 with a self-loop and its adjacent edges can be deleted. The modified graph is Figure 5 .

[0118] Delete set U r and C r After that, the remaining vertex set in the original vertex set V is V0 = VU r -C r The remaining subgraph is the reduced graph G0 = (V0, E0), where E0 is the edge in the reduced graph G0. After reducing the graph G to the reduced graph G0, the greedy algorithm is applied to randomly select a vertex with a larger greedy function value (i.e., the greedy function value exceeds the preset threshold) from the graph G0 and add it to the initial solution π0. The greedy function and the calculated value can be defined, such as the traffic size of the node or the priority of the node.

[0119] 1-1-2: Randomly select a vertex or based on a heuristic method such as out-degree or in-degree, remove vertex v and its connected edges from the graph, and add vertex v to the feedback set.

[0120] 1-1-3: Repeat steps 1-1-1 and 1-1-2 until the graph G0 is reduced to nothing. The initial solution obtained at this time is the generated initial matching solution.

[0121] 1-2: ITS (i.e., enhanced strategy based on taboo search) algorithm optimizes matching

[0122] In the present invention, a strengthening strategy based on taboo search is adopted to find a better solution based on the initial matching solution. How to judge whether there is a ring in the graph directly affects the efficiency of the algorithm. According to the principle that there is a topological sequence (ordered sequence) in a directed acyclic graph, this problem is converted into an ordered sequence consisting of the longest acyclic subgraph. Each time a node is selected and moved to the ordered sequence, the nodes that destroy the pre-order relationship are deleted, and the number of nodes in the ordered sequence is increased until it can no longer be improved.

[0123] The main purpose of the enhanced strategy is to fully explore all feasible solutions (a legal feedback vertex set) near the initial solution constructed greedily to find a better improved solution. The present invention adopts a taboo search algorithm to implement the enhanced strategy, which avoids local optimality by accepting some non-optimal solutions and avoids repeating the same search path in the near future through a taboo table.

[0124] Strengthening strategy algorithm process:

[0125] 1-2-1: Given the reduced graph G0 = (V0, E0), the initial solution π0 and the search depth threshold ω = 30 as the algorithm input, set the number of iterations that failed to optimize No_Improvement and the global optimal solution π * The initial values ​​of are 0 and the initial solution π0;

[0126] 1-2-2: The number of iterations No_Improvement failed to be optimized and increased by 1. The vertices in V0 were randomly sorted to obtain a node sequence.

[0127] 1-2-3: Select a vertex S from the node sequence of V0 and determine whether the vertex S belongs to the initial solution π0;

[0128] 1-2-4: When vertex S does not belong to the initial solution π0, if vertex S can be directly put into the initial solution π0, execute the INSERT operator to put vertex S into the initial solution π0. When the updated initial solution π0 is better than the global optimal solution π * When π is set as the global optimal solution * The updated initial solution π0 is reset to 0. The enhancement strategy algorithm can be combined to determine whether the vertex S is directly placed in the initial solution π0. No constraints are imposed here. The updated initial solution π0 is better than the global optimal solution π * The condition is that the average path length of the initial solution π0 is less than the global optimal solution π * The average path length of the initial solution π0 is less than the global optimal solution π * The average clustering coefficient is the average clustering coefficient of each node. The clustering coefficient reflects the closeness of the connection between the neighboring nodes of a node. The graph with a small clustering coefficient is evenly distributed. When designing a power data network, it is hoped that the clustering coefficient is as small as possible to avoid the harm of large-scale network paralysis caused by the failure of an important node.

[0129] In practical applications, the change in the average clustering coefficient can be used as an important indicator to measure the connection density of a node set in a local area. When the average clustering coefficient of a set is high, it means that the neighbors of the nodes in the set have strong connectivity, forming a denser local structure. Extending the clustering coefficient of a single node to the comparison of node sets can be achieved by calculating the average clustering coefficient of the set or other statistics. This not only allows a reasonable comparison of the connectivity of different node sets, but also more comprehensively reflects the impact of each set on the entire network topology. The specific calculation process of the average path length and clustering coefficient is described in the following.

[0130] 1-2-5: When vertex S belongs to the initial solution π0, according to the scoring formula Calculate the set NM(S). If the set NM(S) is not empty, execute the SWAP operator to replace the conflicting vertex in the initial solution π0 (denoted as vertex U) with vertex S and put it into the initial solution π0; Score C (i) represents the score of node i, represents the in-degree of node i, represents the out-degree of node i, and the set NM(S) represents the set of neighboring nodes of vertex S. Vertex U conflicts with vertex S, which means that vertex U attempts to use the same edge or resource as vertex S.

[0131] 1-2-6: Repeat 1-2-3 to 1-2-5 until all vertices in V0 are traversed;

[0132] 1-2-7: Repeat 1-2-2 to 1-2-6 until the value of the counter No_Improvement exceeds the search depth threshold ω;

[0133] 1-2-8: Return the enhanced solution π, where the nodes in the solution π are the key nodes.

[0134] In step 1, the connectivity features of the panoramic view of power data flow are required, including the maximum connected subgraph, the out-degree or in-degree of the node, etc.

[0135] In step 1, the connectivity of the panoramic view of power data flow is also needed, such as the average path length or clustering coefficient. Among them, in the power data flow network, the average path length corresponds to the average value of the shortest path between any two nodes in the network. This parameter reflects the efficiency and speed of power data flow transmission in the network. The clustering coefficient reflects the connectivity density between a node and its neighboring nodes. If the clustering coefficient is high, it means that the connectivity of the local network where the node is located is very strong, and there are many redundant paths between nodes, which may cause a single point failure to have a greater impact on the entire network. When calculating the average path length, it is calculated in each largest connected subgraph of the panoramic view of power data flow.

[0136] The process of extracting connectivity features includes:

[0137] In order to accurately and quickly analyze the characteristics of the power data flow topology graph, the flow path of the power data center is abstracted. By analyzing the data flow of the power network, the power data flow topology is abstracted into a graph composed of vertices and edges. Then the characteristics of the graph connectivity are extracted. This includes constructing the maximum connected subgraph and calculating the node degree (i.e., in-degree and out-degree). The specific process is as follows:

[0138] Analyze the data flow path of the power system and abstract the power data flow topology into a graph consisting of vertices and edges:

[0139] Extract the features of graph connectivity, including constructing the maximum connected subgraph, calculating vertex degrees, etc.:

[0140] In order to better describe the characteristics of graph connectivity, relevant definitions are given here.

[0141] (1) Maximum connected subgraph: A maximal connected subgraph of an undirected graph is called a connected component (or connected branch). A connected graph has only one connected component, which is itself; a non-connected undirected graph has multiple connected components. Figure 6 As shown, the undirected graph is not a connected graph, but its two subgraphs (within the dotted line) meet the definition of a connected graph. They both satisfy the properties of a connected graph and are therefore connected components.

[0142] (2) Vertex degree

[0143] In an undirected graph, the degree of a vertex refers to the number of vertices adjacent to the vertex. For any vertex v in an undirected graph G = (V, E) i For i The degree is the degree of the connected vertex v i The number of edges is defined as K i .like Figure 7 As shown, for vertex v3, there are three edges (v3,v1), (v3,v2) and (v3,v4), so the degree of vertex v3 is 3. i The number of edges is v i The in-degree of v i The number of edges indicated is v i The out-degree of .

[0144] In summary, by extracting the connectivity features in the graph, a solid theoretical basis is provided for the quantitative calculation of connectivity and identification of key nodes in step 1.

[0145] After extracting the connectivity features in the power data flow topology map (i.e., the power data flow panoramic view), how to use the connectivity features to reasonably represent and calculate the connectivity of each node, and then reflect the degree of influence of the node on the connectivity of the topology map, which plays an important role in determining the accuracy of subsequent key node detection of power data.

[0146] Here, the calculated average path length and clustering coefficient are selected from the statistically observable features extracted from the actual power data flow topology graph as the quantitative observable state of node connectivity in the power data flow topology graph. The advantage of choosing this method is that these statistical features can not only fully reflect the connectivity of the nodes, but also effectively avoid the complexity of calculation.

[0147] Average path length L: In the power data flow network topology, the distance between two points is the number of edges contained in the shortest path connecting the two points. The average path length of the network refers to the average distance between all node pairs in the network. It indicates the degree of separation between nodes in the network and reflects the global characteristics of the network. The average path length L is defined as follows:

[0148]

[0149] N represents the number of nodes, l ik Represents the distance from node i to node k.

[0150] Clustering coefficient C: In a network, the clustering coefficient of a node refers to the ratio of the number of edges between all nodes adjacent to the node to the maximum possible number of edges between these adjacent nodes. i The clustering coefficient C i is defined as follows:

[0151]

[0152] E represents the set of edges, d i represents the degree of node i, v i Represents node i, v j represents node j as a neighbor of node i, E i Indicates the number of neighboring edges of node i. The clustering coefficient reflects the closeness of the connection between the neighboring nodes of a node. The graph with a small clustering coefficient is evenly distributed. When designing a power data network, it is hoped that the clustering coefficient is as small as possible to avoid the harm of large-scale network paralysis caused by the failure of an important node.

[0153] By calculating the connectivity of nodes in the network, the influence of nodes on network connectivity can be described more accurately. Furthermore, through the key node identification method based on the degree of connectivity influence, the differences between nodes in the power data network can be analyzed, and then the key node composition of the power data network can be obtained.

[0154] Step 2 specifically includes:

[0155] Real-time monitoring of key nodes in the power data center can be carried out to discover abnormal data nodes by comparing changes in key nodes in the power business data flow topology, thereby achieving rapid identification and tracing of data changes.

[0156] In summary, the method provided by the present invention can be applied to multiple business types in power marketing business, improve data security, and solve the problems in the existing data change tracing process. Through this method, power data changes can be effectively and timely discovered, and the changed nodes can be traced, protecting the security and function of the business activity link, and effectively solving the problem of tracing the source of power system changes.

[0157] Combination of panoramic view and graph connectivity analysis: The present invention provides a novel method to fully understand and monitor the flow of power data by visualizing the traffic path of power data services and combining it with graph connectivity analysis, which is uncommon in traditional power data monitoring.

[0158] Automatic identification and optimized matching of key nodes: Utilize advanced graph theory algorithms and ITS algorithms to automatically generate initial matching solutions and optimize matching strategies, reduce the number of searches and quickly locate the optimal match, which improves the efficiency and accuracy of key node identification.

[0159] Real-time data change monitoring and rapid response mechanism: Data anomalies can be identified by real-time monitoring of changes in key nodes, and data change traceability can be performed in combination with a rapid response mechanism. This greatly improves the ability to handle power data anomalies, thereby enhancing power data security and business continuity.

[0160] Applicability to multiple business types: The method provided by the present invention is not limited to a certain type of power data business, but can be widely applied to multiple business types of power marketing business, which makes it have broader application prospects and practicality.

[0161] Embodiment 2:

[0162] The present invention based on the same inventive concept also provides a power data abnormal risk discovery system, such as Figure 8 As shown, including:

[0163] Key node identification module and data change risk discovery module;

[0164] A key node identification module is used to identify key nodes in the power data flow panoramic view based on the connectivity of the power data flow panoramic view formed by the power data flow, and adopts a heuristic algorithm based on iterative taboo search;

[0165] The data change risk discovery module is used to monitor key nodes and discover abnormal data nodes by comparing changes in key nodes.

[0166] Preferably, the key node identification module is specifically used for:

[0167] Based on the connectivity of the panoramic view of power data flow formed by power data flow, an initial matching scheme of key nodes in the panoramic view of power data flow is generated in a greedy adaptive manner;

[0168] Based on the connectivity of the panoramic view of power data flow, the initial matching scheme is iteratively optimized using a taboo search-based reinforcement strategy to identify key nodes in the panoramic view of power data flow.

[0169] Among them, the connectivity of the panoramic view of power data flow includes one or more of the following: average path length or clustering coefficient; the characteristics of the connectivity of the panoramic view of power data flow include one or more of the following: out-degree or in-degree.

[0170] Preferably, the key node identification module generates an initial matching scheme of key nodes in the power data flow panoramic view in a greedy adaptive manner based on the connectivity of the power data flow panoramic view formed by the power data flow, including:

[0171] A1: Set the initial solution of the initial matching scheme of key nodes in the panoramic view of power data flow to an empty set;

[0172] A2: Based on the connectivity of the panoramic view of power data flow formed by power data flow, the basic reduction method is used to delete all the node sets in the panoramic view of power data flow that cannot be key nodes, and a reduced graph is obtained. The greedy algorithm is then applied to randomly select a node whose greedy function value exceeds a threshold from the reduced graph and add it to the initial solution, where the greedy function value includes: the flow size or priority of the node;

[0173] A3: In the reduced graph, randomly select a node, or select a node based on out-degree or in-degree and according to a heuristic method, remove the node and its connected edges from the reduced graph, and add the node to the initial solution;

[0174] A4: Determine whether the reduction graph is empty. If so, take the current initial solution as the initial matching solution and end. Otherwise, go to step A2.

[0175] Preferably, the reduction rules of the basic reduction method in the key node identification module include:

[0176] If the in-degree or out-degree of a node is 0, delete the node and its adjacent edges;

[0177] If the in-degree of a node is 1 and there is no self-loop, merge the node into its only predecessor node, or if the out-degree of a node is 1 and there is no self-loop, merge the node into its only successor node;

[0178] If a node has a self-loop, delete the node and its adjacent edges.

[0179] Preferably, the key node identification module uses a taboo search-based reinforcement strategy to iteratively optimize the initial matching solution based on the connectivity of the power data flow panoramic view, and identifies the key nodes in the power data flow panoramic view, including:

[0180] B1: Based on the connectivity of the panoramic view of power data flow constituted by power data flow, the basic reduction method is used to delete all node sets in the panoramic view of power data flow that cannot be key nodes, and a reduced graph is obtained. The search depth threshold and the initial matching scheme are used as input, and the initial value of the number of iterations that failed to be optimized is set to 0, and the initial value of the set constituted by the key nodes is the initial matching scheme;

[0181] B2: The number of iterations failed to be optimized and increased by 1. Each node in the reduced graph was randomly sorted to obtain a node sequence;

[0182] B3: Select a node from the node sequence in sequence and determine whether the node belongs to the initial matching solution. If so, proceed to step B5; otherwise, proceed to step B4;

[0183] B4: If the node can be directly placed in the initial matching solution, the insertion operator is executed to place the node in the initial matching solution, and based on the average path length and the clustering coefficient, it is determined whether the updated initial matching solution is better than the current set of key nodes. When the updated initial matching solution is better than the current set of key nodes, the set of key nodes is set as the updated initial matching solution and the number of iterations that failed to be optimized is set to 0, and the process goes to step B6;

[0184] B5: Based on the in-degree and out-degree, the scoring formula is used to calculate the neighbor node set of the node. If the neighbor node set is not empty, the exchange operator is executed to replace the node that conflicts with the node in the initial matching solution with the node;

[0185] B6: Determine whether all nodes in the reduction graph have been traversed. If so, execute B7, otherwise go to B3;

[0186] B7: Determine whether the number of iterations that have not been optimized exceeds the search depth threshold. If so, output the set of current key nodes and end. Otherwise, proceed to step B2.

[0187] Preferably, the scoring formula in the key node identification module is expressed as follows:

[0188]

[0189] In the formula, Score C (i) represents the score of node i, represents the in-degree of node i, Represents the out-degree of node i.

[0190] Preferably, the key node identification module determines whether the updated initial matching solution is better than the current set of key nodes based on the average path length and the clustering coefficient, including:

[0191] Calculate the average path length and clustering coefficient of the updated initial matching solution, as well as the average path length and clustering coefficient of the set of current key nodes respectively;

[0192] If the average path length of the updated initial matching scheme is smaller than the average path length of the current set of key nodes, or the average clustering coefficient of the updated initial matching scheme is smaller than the average clustering coefficient of the current set of key nodes, then the updated initial matching scheme is better than the current set of key nodes, where the average clustering coefficient is the average value of the clustering coefficients of each node.

[0193] Preferably, the calculation formula of the average path length in the key node identification module is as follows:

[0194]

[0195] In the formula, L represents the average path length, N represents the number of nodes, and l ik Represents the distance from node i to node k.

[0196] Preferably, the clustering coefficient of the nodes in the key node identification module is calculated as follows:

[0197]

[0198] In the formula, C i represents the clustering coefficient of node i, E represents the set of edges, d i represents the degree of node i, v i Represents node i, v j represents node j as a neighbor of node i, E i Represents the number of neighboring edges of node i.

[0199] Preferably, the system further comprises a feature extraction module of graph connectivity;

[0200] The feature extraction module of graph connectivity is used to abstract the power data flow into a power data flow panoramic view composed of nodes and edges; extract the maximum connected subgraph of the power data flow panoramic view, and calculate the in-degree and out-degree of the power data flow panoramic view.

[0201] Preferably, the system may further include a graph node connectivity calculation module for performing connectivity calculation, and when the key node identification module calculates the average path length and the clustering coefficient, the graph node connectivity calculation module may be called to perform calculations.

[0202] The power data change risk discovery system includes a key node identification module, a data change risk discovery module, a graph connectivity feature extraction module and a graph node connectivity calculation module. Fig. 9 shown.

[0203] Example 3

[0204] like Fig.10 As shown, the present invention also provides an electronic device, which may be a computer device, a single-chip device, an intelligent mobile device, etc. The electronic device in this embodiment may include a processor, a memory, a transceiver component, etc. The memory, the processor, and the transceiver component are connected via a bus; the memory may be used to store an execution program, and an exemplary execution program may include instructions; the processor is used to execute the instructions stored in the memory. The memory may also be used to store data, which may be called and / or modified when the instructions are executed.

[0205] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in a storage medium to implement the corresponding method flow or corresponding functions, so as to implement the steps of a method for discovering the risk of abnormal changes in power data in the above-mentioned embodiment.

[0206] Example 4

[0207] Based on the same inventive concept, the present invention also provides a readable storage medium, specifically an electronic device readable storage medium (Memory), which is a memory device in an electronic device for storing programs and data. It can be understood that the storage medium here can include both built-in storage media in electronic devices and, of course, extended storage media supported by electronic devices. The storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by a processor are also stored in the storage space, and these instructions can be one or more execution programs (including program codes). It should be noted that the storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor loads and executes one or more instructions stored in the storage medium, which can implement the steps of a method for discovering the risk of power data changes in the above embodiment.

[0208] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take 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.) containing computer-usable program code.

[0209] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks 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 produce 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 1 A device that provides the functions specified in a block or multiple blocks.

[0210] 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 comprising 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.

[0211] 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.

[0212] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention are included in the scope of the claims of the present invention to be approved.

Claims

1. A method for discovering power data abnormality risks, characterized in that: include: Based on the connectivity of the panoramic view of power data flow constituted by power data flow, a heuristic algorithm based on iterative taboo search is used to identify the key nodes in the panoramic view of power data flow; The key nodes are monitored, and abnormal data nodes are discovered by comparing changes in the key nodes.

2. The method according to claim 1, characterized in that The connectivity of the panoramic view of power data flow formed by the power data flow is based on the heuristic algorithm based on iterative taboo search to identify key nodes in the panoramic view of power data flow, including: Based on the connectivity of the panoramic view of power data flow formed by power data flow, an initial matching scheme of key nodes in the panoramic view of power data flow is generated in a greedy adaptive manner; Based on the connectivity of the panoramic view of the power data flow, an enhanced strategy based on taboo search is used to iteratively optimize the initial matching scheme to identify key nodes in the panoramic view of the power data flow; Among them, the connectivity of the panoramic view of the power data flow includes one or more of the following: average path length or clustering coefficient; the characteristics of the connectivity of the panoramic view of the power data flow include one or more of the following: out-degree or in-degree.

3. The method according to claim 2, characterized in that The connectivity of the panoramic view of power data flow formed by the power data flow is based on the greedy adaptive method to generate an initial matching scheme of key nodes in the panoramic view of power data flow, including: A1: Set the initial solution of the initial matching scheme of key nodes in the panoramic view of power data flow to an empty set; A2: Based on the connectivity of the panoramic view of power data flow constituted by power data flow, a basic reduction method is used to delete all node sets in the panoramic view of power data flow that cannot be key nodes, to obtain a reduced graph, and a greedy algorithm is applied to randomly select a node whose greedy function value exceeds a threshold from the reduced graph to add to the initial solution, wherein the greedy function value includes: the flow size or priority of the node; A3: In the reduced graph, randomly select a node, or select a node based on the out-degree or in-degree and according to a heuristic method, remove the node and its connected edges from the reduced graph, and add the node to the initial solution; A4: Determine whether the reduction graph is empty. If so, take the current initial solution as the initial matching solution and end. Otherwise, proceed to step A2.

4. The method according to claim 3, characterized in that The reduction rules of the basic reduction method include: If the in-degree or out-degree of a node is 0, delete the node and its adjacent edges; If the in-degree of a node is 1 and there is no self-loop, merge the node into its only predecessor node, or if the out-degree of a node is 1 and there is no self-loop, merge the node into its only successor node; If a node has a self-loop, delete the node and its adjacent edges.

5. The method according to claim 2, characterized in that The connectivity of the panoramic view of the power data flow is based on the use of a taboo search-based reinforcement strategy to iteratively optimize the initial matching solution, and identify key nodes in the panoramic view of the power data flow, including: B1: Based on the connectivity of the panoramic view of power data flow constituted by power data flow, a basic reduction method is used to delete all node sets in the panoramic view of power data flow that cannot be key nodes, and a reduced graph is obtained. The search depth threshold and the initial matching scheme are used as inputs, and the initial value of the number of iterations that cannot be optimized is set to 0, and the initial value of the set constituted by the key nodes is the initial matching scheme; B2: The number of iterations cannot be optimized to increase by 1, and each node in the reduction graph is randomly sorted to obtain a node sequence; B3: sequentially select a node from the node sequence and determine whether the node belongs to the initial matching solution. If so, proceed to step B5; otherwise, proceed to step B4; B4: If the node can be directly placed in the initial matching solution, the insertion operator is executed to place the node in the initial matching solution, and based on the average path length and the clustering coefficient, it is determined whether the updated initial matching solution is better than the current set of key nodes. When the updated initial matching solution is better than the current set of key nodes, the set of key nodes is set as the updated initial matching solution and the number of iterations that failed to be optimized is set to 0, and the process goes to step B6; B5: Based on the in-degree and out-degree, a scoring formula is used to calculate the neighbor node set of the node. If the neighbor node set is not empty, the exchange operator is executed to replace the node in the initial matching solution that conflicts with the node with the node; B6: Determine whether all nodes in the reduced graph have been traversed. If so, execute B7; otherwise, go to B3; B7: Determine whether the number of iterations that have not been optimized exceeds the search depth threshold. If so, output the set of current key nodes and end. Otherwise, proceed to step B2.

6. The method according to claim 5, characterized in that The scoring formula is expressed as follows: In the formula, Score C (i) represents the score of node i, represents the in-degree of node i, Represents the out-degree of node i.

7. The method according to claim 5, characterized in that The step of judging whether the updated initial matching scheme is better than the current set of key nodes based on the average path length and the clustering coefficient includes: Calculate the average path length and clustering coefficient of the updated initial matching solution, as well as the average path length and clustering coefficient of the set of current key nodes respectively; If the average path length of the updated initial matching scheme is smaller than the average path length of the current set of key nodes, or the average clustering coefficient of the updated initial matching scheme is smaller than the average clustering coefficient of the current set of key nodes, then the updated initial matching scheme is better than the current set of key nodes, where the average clustering coefficient is the average value of the clustering coefficients of each node.

8. The method according to claim 7, characterized in that The average path length is calculated as follows: In the formula, L represents the average path length, N represents the number of nodes, and l ik Represents the distance from node i to node k.

9. The method according to claim 7, characterized in that The calculation formula of the node clustering coefficient is as follows: In the formula, C i represents the clustering coefficient of node i, E represents the set of edges, d i represents the degree of node i, v i Represents node i, v j represents node j as a neighbor of node i, E i Represents the number of neighboring edges of node i.

10. The method according to claim 1, characterized in that The connectivity of the panoramic view of power data flow formed by the power data flow, before identifying the key nodes in the panoramic view of power data flow by using a heuristic algorithm based on iterative taboo search, further includes: Abstract the power data flow into a panoramic view of the power data flow consisting of nodes and edges; The maximum connected subgraph of the power data flow panoramic view is extracted, and the in-degree and out-degree of the power data flow panoramic view are calculated.

11. A power data abnormality risk discovery system, characterized in that: include: Key node identification module and data change risk discovery module; The key node identification module is used to identify the key nodes in the power data flow panoramic view based on the connectivity of the power data flow panoramic view formed by the power data flow, using a heuristic algorithm based on iterative taboo search; The data change risk discovery module is used to monitor the key nodes and discover abnormal data nodes by comparing the changes in the key nodes.

12. The system according to claim 11, characterized in that The key node identification module is specifically used for: Based on the connectivity of the panoramic view of power data flow formed by power data flow, an initial matching scheme of key nodes in the panoramic view of power data flow is generated in a greedy adaptive manner; Based on the connectivity of the panoramic view of the power data flow, an enhanced strategy based on taboo search is used to iteratively optimize the initial matching scheme to identify key nodes in the panoramic view of the power data flow; Among them, the connectivity of the panoramic view of the power data flow includes one or more of the following: average path length or clustering coefficient; the characteristics of the connectivity of the panoramic view of the power data flow include one or more of the following: out-degree or in-degree.

13. The system of claim 12, wherein: The key node identification module generates an initial matching scheme of key nodes in the power data flow panoramic view in a greedy adaptive manner based on the connectivity of the power data flow panoramic view formed by the power data flow, including: A1: Set the initial solution of the initial matching scheme of key nodes in the panoramic view of power data flow to an empty set; A2: Based on the connectivity of the panoramic view of power data flow constituted by power data flow, a basic reduction method is used to delete all node sets in the panoramic view of power data flow that cannot be key nodes, to obtain a reduced graph, and a greedy algorithm is applied to randomly select a node whose greedy function value exceeds a threshold from the reduced graph to add to the initial solution, wherein the greedy function value includes: the flow size or priority of the node; A3: In the reduced graph, randomly select a node, or select a node based on the out-degree or in-degree and according to a heuristic method, remove the node and its connected edges from the reduced graph, and add the node to the initial solution; A4: Determine whether the reduction graph is empty. If so, take the current initial solution as the initial matching solution and end. Otherwise, proceed to step A2.

14. The system of claim 13, wherein: The reduction rules of the basic reduction method in the key node identification module include: If the in-degree or out-degree of a node is 0, delete the node and its adjacent edges; If the in-degree of a node is 1 and there is no self-loop, merge the node into its only predecessor node, or if the out-degree of a node is 1 and there is no self-loop, merge the node into its only successor node; If a node has a self-loop, delete the node and its adjacent edges.

15. The system of claim 12, wherein: The key node identification module uses a taboo search-based reinforcement strategy to iteratively optimize the initial matching solution based on the connectivity of the power data flow panoramic view, and identifies the key nodes in the power data flow panoramic view, including: B1: Based on the connectivity of the panoramic view of power data flow constituted by power data flow, a basic reduction method is used to delete all node sets in the panoramic view of power data flow that cannot be key nodes, and a reduced graph is obtained. The search depth threshold and the initial matching scheme are used as inputs, and the initial value of the number of iterations that cannot be optimized is set to 0, and the initial value of the set constituted by the key nodes is the initial matching scheme; B2: The number of iterations cannot be optimized to increase by 1, and each node in the reduction graph is randomly sorted to obtain a node sequence; B3: sequentially select a node from the node sequence and determine whether the node belongs to the initial matching solution. If so, proceed to step B5; otherwise, proceed to step B4; B4: If the node can be directly placed in the initial matching solution, the insertion operator is executed to place the node in the initial matching solution, and based on the average path length and the clustering coefficient, it is determined whether the updated initial matching solution is better than the current set of key nodes. When the updated initial matching solution is better than the current set of key nodes, the set of key nodes is set as the updated initial matching solution and the number of iterations that failed to be optimized is set to 0, and the process goes to step B6; B5: Based on the in-degree and out-degree, a scoring formula is used to calculate the neighbor node set of the node. If the neighbor node set is not empty, the exchange operator is executed to replace the node in the initial matching solution that conflicts with the node with the node; B6: Determine whether all nodes in the reduced graph have been traversed. If so, execute B7; otherwise, go to B3; B7: Determine whether the number of iterations that have not been optimized exceeds the search depth threshold. If so, output the set of current key nodes and end. Otherwise, proceed to step B2.

16. The system of claim 15, wherein: The expression of the scoring formula in the key node identification module is as follows: In the formula, Score C (i) represents the score of node i, represents the in-degree of node i, Represents the out-degree of node i.

17. The system of claim 15, wherein: The key node identification module determines whether the updated initial matching solution is better than the current set of key nodes based on the average path length and the clustering coefficient, including: Calculate the average path length and clustering coefficient of the updated initial matching solution, and the average path length and clustering coefficient of the set of current key nodes respectively; If the average path length of the updated initial matching scheme is smaller than the average path length of the current set of key nodes, or the average clustering coefficient of the updated initial matching scheme is larger than the average clustering coefficient of the current set of key nodes, then the updated initial matching scheme is better than the current set of key nodes, where the average clustering coefficient is the average value of the clustering coefficients of each node.

18. The system of claim 17, wherein: The calculation formula of the average path length in the key node identification module is as follows: In the formula, L represents the average path length, N represents the number of nodes, and l ik Represents the distance from node i to node k.

19. The system of claim 17, wherein: The calculation formula of the clustering coefficient of the nodes in the key node identification module is as follows: In the formula, C i represents the clustering coefficient of node i, E represents the set of edges, d i represents the degree of node i, v i Represents node i, v j represents node j as a neighbor of node i, E i Represents the number of neighboring edges of node i.

20. The system of claim 11, wherein: The system also includes a feature extraction module of graph connectivity; The feature extraction module of the graph connectivity is used to abstract the power data flow into a panoramic view of the power data flow consisting of nodes and edges; The maximum connected subgraph of the power data flow panoramic view is extracted, and the in-degree and out-degree of the power data flow panoramic view are calculated.

21. A computer device, characterized in that: include: at least one processor and memory; The memory and the processor are connected via a bus; The memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, the power data change risk discovery method as described in any one of claims 1 to 10 is implemented.

22. A computer-readable storage medium, characterized in that: An execution program is stored thereon, and when the execution program is executed, the method for discovering the risk of abnormal changes in electric power data as described in any one of claims 1 to 10 is implemented.

Images