Multipoint network attack-oriented complex power distribution system vulnerability identification method

Abstract

The invention relates to the technical field of power distribution system security, and discloses a complex power distribution system vulnerability identification method for multipoint network attacks, which comprises the following steps of: obtaining topological data of a power distribution network and information of a target user, and constructing a tree topology model with complete attributes; classifying the target users; corresponding switch search strategies are adopted for different types of target users, and a preliminary planning power failure scheme is generated; performing joint optimization on the preliminarily planned power failure scheme to generate an optimized switch combination; and outputting the optimized switch combination as a system vulnerability set. According to the method, the tree topology model of the core attribute set containing the depth is constructed, the users are classified according to the tree topology model, the corresponding switch search strategy is adopted, then through joint optimization, the cut-off effect is guaranteed, meanwhile, the optimized switch combination with the smaller operation number and the smaller additional influence is generated, and the accuracy of vulnerability identification is improved.

Description

Technical Field

[0001] This invention relates to the field of power distribution system security technology, specifically to a method for identifying vulnerabilities in complex power distribution systems in response to multi-point network attacks. Background Technology

[0002] With the deep application of information and communication technologies in power systems, the intelligence and automation levels of distribution networks have been improved. While the integration of the information layer and the physical layer enhances the operational efficiency of distribution networks, it also introduces new cybersecurity risks. Attackers can exploit information network paths to disrupt the normal operation of the power physical system by tampering with control commands or interfering with status awareness, thereby threatening power supply reliability.

[0003] To address these risks, existing technologies have focused on methods for identifying vulnerable points in power systems. Some methods assess and identify critical nodes in the power grid by analyzing electrical characteristics such as the system's voltage change rate and branch power flow. Other methods combine the topological properties and electrical characteristics of complex networks to construct comprehensive evaluation models that rank the importance of nodes. Still other research has incorporated deep learning techniques to identify critical nodes by fusing multi-dimensional features such as topology, electrical distance, and voltage.

[0004] However, most existing technologies focus on assessing the importance of individual nodes or the overall stability of the power grid, lacking effective identification methods for scenarios where attackers launch coordinated attacks against multiple users in the distribution network. Existing methods fail to fully consider the complex power supply patterns constituted by main feeders, branch feeders, and backup transfer paths in the distribution network, and do not provide solutions for generating a switch operation combination that can both ensure the disconnection of power to all target users and control additional losses. Therefore, identifying a switch combination capable of accurately interrupting power supply to multiple users is a problem that needs to be solved in this field. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for identifying vulnerabilities in complex power distribution systems in the face of multi-point network attacks. This method solves the problem of finding a precise switch combination that can both ensure the interruption of power supply to multiple target users and control additional effects under complex power supply modes.

[0006] To address the aforementioned technical problems, this invention provides a method for identifying vulnerabilities in complex power distribution systems against multi-point network attacks, comprising the following steps: S01. Obtain the topology data of the distribution network and the information of the target users, and construct a tree topology model with complete attributes that can characterize the connection relationship of each feeder. S02. Based on the tree topology model, classify the target users according to their location attributes in the tree topology model and whether they have a transfer path; S03. For different categories of target users, adopt the corresponding switch search strategy to generate a preliminary power outage plan containing all necessary operations; S04. Perform joint optimization on the preliminary planned power outage scheme to generate an optimized switch combination; S05. Output the optimized switch combination as a set of system vulnerabilities.

[0007] As a further optimization of the technical solution of the present invention, in step S01, when constructing the tree topology model, a core attribute set is filled for each node in the model. This core attribute set includes at least: The feeder number used to identify the feeder to which a node belongs, the globally unique identifier of the node, the parent node ID of the direct superior node, the depth representing the number of edges in the path from the root node to the current node, the line hierarchy code reflecting the precise location of the node in the topology, and the node type identifying the type of equipment.

[0008] Specifically, when the node type is a feeder switch, a cross-tree reference field is also defined in the core attribute set. This cross-tree reference field records a feeder number and a globally unique identifier for a corresponding connected switch, wherein the feeder is a feeder connected to the feeder to which the feeder itself belongs through the feeder switch. This design enables the model to accurately represent the potential connection relationships across feeders.

[0009] In step S02, the specific process of classifying target users is as follows: In the tree topology model, the continuous path with the largest line capacity is defined as the main feeder, and the remaining lines directly connected to the main feeder are defined as branch feeders. Based on this division, target users are initially classified into main feeder users and branch feeder users. It is determined whether the target user has a backup transfer path, and the transfer type is distinguished as internal feeder transfer or cross-feeder transfer. The location attribute and transfer path attribute are cross-combined to classify the target users into branch feeder users without transfer paths, branch feeder users with internal feeder transfer, branch feeder users with cross-feeder transfer, main feeder users without transfer paths, and main feeder users with transfer paths.

[0010] In step S03, corresponding switch search strategies are adopted for different categories of users.

[0011] For branch feeder users without a transfer path, calculate the target user clustering degree on their respective branch feeder. Target user clustering degree The calculation method is as follows: .in, This represents the total number of users on this branch feeder. This refers to the number of target users on the same branch feeder. If... If the percentage is greater than 50%, it is determined to be the target user of the cluster, and the nearest common upstream network reachable circuit breaker is determined as the operating switch; if If the percentage is no more than 50%, it is determined to be a dispersed target user, and each dispersed target user independently searches for and determines the nearest upstream network reachable circuit breaker as the operating switch.

[0012] For target users with a power transfer path, search and determine the first set of operating switches to cut off their main power supply path; simultaneously, on their power transfer path, based on the principle of minimizing additional losses, search and determine the second set of operating switches to cut off the backup power supply path. Combine the two sets of operating switches to obtain a complete set of operating switches.

[0013] In step S04, the preliminary planned power outage scheme is jointly optimized. The optimization process includes redundant switch reduction based on the topology level. The rule is: when a main feeder switch in the scheme is located upstream of a branch feeder switch, the main feeder switch is retained, and the branch feeder switch is removed from the scheme.

[0014] Furthermore, the structure of the switch combination is optimized. Based on the parent-child node attributes between switch nodes, the system traverses upwards to identify whether there exists a higher-level switch whose operation can cover the control range of multiple switches in the current combination, and whose downstream has no other uninterrupted power supply path. If such a higher-level switch exists, it replaces the original multiple switches.

[0015] During the optimization process, the principle of minimizing additional losses is also followed for selection. When multiple feasible switching reduction or alternative solutions exist, the total additional losses of each solution are calculated and compared, and the solution with the minimum total additional loss is selected first. Total Additional Loss The calculation method is as follows: ,in, For the first in the plan Load loss to non-target users caused by a single switch operation.

[0016] In step S05, the specific output includes: a target switch set containing globally unique identifiers of all switches constituting the vulnerability, the total additional loss value associated with the optimized switch combination, and a set of coordinated switch action sequences.

[0017] This invention provides a method for identifying vulnerabilities in complex power distribution systems in response to multi-point network attacks. It offers the following advantages: 1. This invention optimizes the initial planned power outage scheme by eliminating redundant switch operations and simplifying the switch combination structure. Based on the principle of minimizing additional losses, it can generate an optimized switch combination with fewer operations and a smaller impact on non-target users while ensuring that the power supply path to all target users is cut off, thereby improving the accuracy of vulnerability identification.

[0018] 2. This invention classifies target users based on their location attributes in the tree topology model and whether they have a transfer path, and adopts corresponding switch search strategies for different categories of users, such as branch feeder users without transfer paths and target users with transfer paths. It can generate preliminary planned power outage schemes containing all necessary operations for the complex power supply modes of various users, ensuring the comprehensiveness of cutting off all power supply paths of the target users.

[0019] 3. This invention constructs a model that can accurately characterize the connection relationship of each feeder, the topological hierarchy relationship, and the cross-feeder transfer path of the distribution network by filling each node in the tree topology model with a core attribute set containing depth and line hierarchical coding, and defining a cross-tree reference field for the transfer switch. This provides a structured data foundation for subsequent user classification and switch search strategies. Attached Figure Description

[0020] Figure 1 This is a diagram of a distribution network feeder tree structure according to an embodiment of the present invention; Figure 2 This is a distribution network planned power outage attack decision diagram according to an embodiment of the present invention; Figure 3 This is a flowchart illustrating the vulnerability identification process for complex power distribution systems against multi-point network attacks, according to an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] This invention provides a method for identifying vulnerabilities in complex power distribution systems against multi-point network attacks, which can run on computing devices such as servers and computer terminals. The hardware environment of this device may include a processor, a memory (for storing program instructions), and related communication interfaces.

[0023] See attached document Figure 1 -Appendix Figure 3This invention provides a method for identifying vulnerabilities in complex power distribution systems in response to multi-point network attacks, comprising the following steps: S01. Obtain the topology data of the distribution network and the target important user information, and construct a tree topology model with complete attributes that can characterize the connection relationship of each feeder. S02. Based on the tree topology model, all target users are automatically classified according to their location attributes in the topology (such as whether the feeder they belong to is the main feeder or a branch feeder) and whether they have a transfer path. S03. For different user groups, adopt a differentiated switch search strategy to generate a preliminary power outage plan that includes all necessary operations. S04. Jointly optimize the preliminary planned power outage scheme by eliminating redundant switch operations, simplifying the switch combination structure and minimizing additional losses to generate an optimized switch combination. S05. Output the optimized switch combination as a set of system vulnerabilities.

[0024] The specific technical implementation methods involved in each of the above steps will be explained in detail below.

[0025] See attached document Figure 1 In the initial stage of the vulnerability identification method, it is necessary to transform the topology information of the actual distribution network into a tree-like data structure with rich attributes, which can be efficiently analyzed by computers. Step S01 provides a unified and standardized data foundation for all subsequent analysis and calculations. This step may specifically include: S101. Receive and process the raw topology data of the distribution network. This data can come from a distribution automation system database containing equipment ledgers and line parameters, or from data files conforming to the Common Information Model (CIM) standard. To facilitate subsequent calculations, the raw data first needs to be preprocessed. For example, since the original equipment number (OID) itself is not globally unique, the equipment type code (CID) can be concatenated with the OID to generate a globally unique node ID. The CID is used to distinguish equipment types; for example, 108 represents a substation outgoing circuit breaker, and 114 represents a switch. Furthermore, the status of key equipment for load transfer is standardized and labeled; for example, a normally open switch in the open position under normal conditions is marked with a status of 0.

[0026] S102. Represent the preprocessed connection information in a structured manner. Mathematically, a structure containing... A power distribution network with multiple switching nodes can have its interconnections defined by an adjacency matrix. Characterization: ; in, to The elements within the matrix represent all relevant switching equipment in the distribution network structure. This defines the connection relationship between any two switches: ; S103. Based on structured connections, a tree topology model of the distribution network is constructed using a depth-first search algorithm. This process first initializes an empty dictionary as an adjacency list and traverses all edge relationships, recording the direct connections between all nodes. Then, all substation outgoing circuit breakers are identified and marked as root nodes. Starting from each root node, the depth-first search function is recursively called for depth-first traversal. During the traversal, a set of visited nodes is used to record processed nodes to prevent duplicate visits and loop formation. When the recursion reaches a leaf node with no unvisited neighbors, the call begins to return layer by layer, and the parent node adds the returned subtree to its own child node list. This process continues until all reachable nodes have been visited, ultimately forming one or more complete tree topologies rooted at substation outgoing circuit breakers.

[0027] To more clearly describe this tree-like topology model, the following will combine it with the appendix. Figure 1 The node structure of the model is explained. In the final tree topology, different nodes play distinct roles: the root node (such as a substation outgoing circuit breaker) is located at the top level of the topology, serving as the starting point for traversal, and its parent node attribute is empty; internal nodes act as bridges between higher and lower levels, serving as both child nodes of higher-level nodes and parent nodes of lower-level nodes, forming branches of the tree and playing a crucial role in maintaining topological connectivity; leaf nodes are located at the ends of each branch, no longer possessing any child nodes, and typically represent end users or transfer switches, marking the end point of the path.

[0028] Please refer to the appendix for details. Figure 1 In this topology, node B (the substation outgoing circuit breaker) serves as the root node, representing the starting point and overall control point of the feeder. Nodes S1 to S11 are internal nodes, forming the branching structure of the topology. Nodes U1 to U6 are leaf nodes, typically representing the distribution transformers or users at the end of the feeder, and are the key targets of this analysis.

[0029] S104. To accurately characterize the topology and functional attributes of the distribution network, the following core attribute set is calculated and populated for each node in the constructed tree model: Feeder number: A unique code that identifies the feeder to which a node belongs; Node ID: A globally unique identifier for a node, ensuring its uniqueness and addressability within the system; Parent Node ID: Records the ID of this node's direct parent node. For the root node, this attribute is empty; Depth: Represents the number of edges traversed in the shortest path from the root node to the current node. The depth of the root node is defined as 0; Line hierarchical coding: a hierarchical coding system that reflects the precise location of nodes in the feeder topology. The main feeder is the first level (e.g., F1), its direct branches are the second level (e.g., F1-1), and subsequent branches follow the same pattern (e.g., F1-1-1). Parent Line Code: The line code of the branch to which the current node belongs. This attribute is empty for the main feeder node. Node type: Identifies the type of equipment at the node, such as circuit breaker, sectionalizing switch, load switch, etc. This attribute is crucial for determining whether the switchgear is equipped with a feeder terminal unit (FTU).

[0030] S105. Special leaf nodes acting as transfer switches are processed to accurately characterize cross-feeder connections. These nodes are normally disconnected, belong to the current feeder topology tree, but are physically connected to another feeder. To incorporate this potential connectivity into the model, a cross-tree reference field needs to be defined in the attributes of this transfer switch node. This field records the feeder number on the opposite side it connects to and the node ID of the switch connected to the opposite side. To ensure topology consistency, the switch on the opposite side also needs to establish a reverse reference, thus forming a complete bidirectional traceable connection record. This reference relationship is a key feature for subsequently identifying transfer logic and determining the user's transfer capability. When a power outage occurs in the main tree, closing this transfer switch can realize cross-feeder transfer of the load, restoring power supply to the upstream part of the network. Therefore, accurately characterizing this potential branch relationship is crucial for simulating complex power outage and power restoration scenarios.

[0031] After constructing the tree topology model of the distribution network in step S01, step S02 automatically classifies all important target users to be analyzed. This classification forms the basis for implementing the differentiated switch group search strategy in the subsequent step S03. This step combines the topological location attributes of target users with the power supply path characteristics to achieve accurate grouping. Specifically, it may include: S201. Perform topology hierarchical division of the feeders in the distribution network to distinguish between main feeders and branch feeders. In each topology tree constructed in step S01, the continuous path with the largest line capacity and the longest path is determined as the main feeder. In specific implementation, starting from the root node of the tree (the circuit breaker of the substation), all paths to the leaf nodes are traversed, and the electrical parameters of the path (such as line capacity) and the topology length (such as the number of nodes passed through) are comprehensively considered to determine the main road. All nodes (switches) located on the main feeder path are defined as main feeder switches. In addition to the main feeder, all paths that branch directly from the main feeder switch node are classified as branch feeders.

[0032] S202. Based on the hierarchical division results of the feeder, the target users are initially classified by querying the upstream connections of user nodes in the topology tree. Users directly connected to the main feeder switch are defined as main feeder users; correspondingly, users located on each branch feeder are classified as branch feeder users.

[0033] S203. Determine whether each target user has an alternative transfer path and distinguish the transfer type. This determination process relies on the cross-tree reference attribute defined in step S105. The specific implementation method is as follows: For a target user, the algorithm checks whether there is a power transfer switch downstream of the power supply path that serves as a leaf node.

[0034] If a power transfer switch exists, read the corresponding cross-tree reference field for the opposite connected switch. If this reference is identical to the current power transfer switch, then the user has an internal power transfer path within the feeder.

[0035] If the switch connected to the opposite side is different from this transfer switch, it is determined that the user has a cross-feeder transfer path.

[0036] If a user's entire downstream path does not contain any transfer switches, then the user is determined to be a user without a transfer path.

[0037] S204. The above topology location attributes (main feeder / branch feeder) and transfer path attributes (no transfer / internal transfer / cross-feeder transfer) are cross-combined to perform a final refined classification of all target users. Through this step, all input target users are divided into one of the following five independent categories: Branch feeder users without a transfer path; Branch feeder users supplied from within the feeder; Branch feeder users supplied across feeders; Main feeder users without a transfer path; Main feeder users with a transfer path.

[0038] After completing the above steps, the system will assign a clear category label to each target user, providing clear input for adopting a differentiated switch search strategy that matches different user groups in step S03.

[0039] See attached document Figure 2 After classifying the target users, this method employs a differentiated intelligent search strategy for each of the five user groups identified in step S02 to determine the set of switches required for precise power outage. This step may specifically include: S301. For branch feeder user groups without a transfer path, this method introduces the target user clustering degree (…). This factor is used as a decision factor to determine whether the target users on this branch feeder are clustered or dispersed. The formula for calculating this decision factor is: ; in, This represents the total number of users on this branch feeder. This refers to the number of key users on the target feeder line of this branch.

[0040] according to The calculation results are used to execute different search logics: like Then the target user group is determined to be a clustered target user group. At this point, the method will start from these user nodes and search upstream along the topology tree until they find their nearest common upstream network reachable circuit breaker. This strategy aims to isolate the clustered user group with minimal operations.

[0041] like If so, the target user group is determined to be a dispersed target user group. At this point, the method will independently search upstream along the topology tree for the nearest network-reachable circuit breaker for each dispersed target user. This strategy aims to accurately isolate dispersed targets and avoid unnecessary collateral damage to non-target users.

[0042] The above search logic can be expressed by the following formula: ; in, To gather the target user group, For a dispersed target user group, For target user concentration, This refers to the number of key users on the target feeder line of this branch.

[0043] The switches found were all included in a group called In the set of switches.

[0044] S302. For a group of users on branch feeders that transfer power within a feeder, the determination criterion is that the downstream transfer switch of the target user and the corresponding connecting switch on the opposite side have the same feeder number. For such users, the method treats the two associated branch feeders providing main power and backup power as a whole, searches upstream for the nearest common network reachable circuit breaker of these two branch feeders as the operating point, and incorporates the result into the switch set. .

[0045] S303. For branch feeder user groups that transfer power across feeders, the determination condition is that the feeder number of the transfer switch and the connecting switch on the opposite side are different. This strategy is executed in two steps: First, based on the user clustering degree in step S301... The logic determines the disconnection scheme for cutting off its main power supply path. Then, to cut off its backup power supply path, the method searches for and determines an optimal circuit breaker as an additional disconnection point on its transfer path, with the goal of minimizing additional losses. All switches determined in these two steps are included in the switch set. .

[0046] S304. For main feeder user groups without a transfer path, their structure is relatively simple due to their location on the main feeder, so there is no need to determine the clustering of user distribution. The method directly searches for the nearest upstream network reachable circuit breaker for each target user in the group and incorporates the results into the switch set. .

[0047] S305. For a group of customers on a main feeder with a power transfer path, the method employs a combined strategy. First, following the logic of step S304, a nearest disconnection operation is performed on the target customer on the main feeder to cut off the main power supply path. Then, similar to the second step of step S303, an additional search is conducted on the power transfer path to identify a circuit breaker with the minimum additional loss as the break point, thus cutting off the backup power supply path. The final result is incorporated into the switch set. .

[0048] S306. After executing the above-mentioned corresponding search strategies for all target user groups, the switch sets obtained from each strategy are... , , , , Merge the results. Through the union operation, a preliminary, unoptimized planned power outage scheme is generated. (Hereinafter referred to as the preliminary power outage plan) Its expression is: ; This preliminary power outage plan In theory, it ensures that all power supply paths (including main power supply and backup power transfer) for all target users are covered.

[0049] The preliminary power outage plan generated in step S03 Subsequently, especially in multi-source urban distribution networks with complex power transfer capabilities, the power supply modes of critical target users exhibit high complexity and diversity. To block all potential power supply paths (including the main power supply and multiple backup power transfer paths), preliminary power outage plans are being developed. As a union of multiple independent strategies, the switch set often becomes excessively large, containing a large number of redundant operations. Therefore, this embodiment introduces a joint optimization algorithm for switch groups to optimize the initial power outage planning. Simplify and refine. This step aims to strike a balance between minimizing additional overhead and reducing the complexity of attack implementation, and may specifically include: S401. Preliminary power outage plan based on topology hierarchy. Redundant switches are reduced. This process involves systematically identifying and eliminating switches based on the upstream and downstream relationships between the main feeder and branch feeders, as well as the switch hierarchy on the same feeder. Specific rules are as follows: When a main feeder switch is located upstream of a branch feeder switch, operating the main feeder switch already covers the downstream branch feeder switch. Therefore, the main feeder switch should be retained, and a preliminary power outage plan should be developed. Remove the branch feeder switch. For example, if the main feeder switch... Located in the branch feeder switch Upstream, the switch combination is updated to (hereinafter referred to as) (This refers to a switch combination that has undergone redundancy reduction). Conversely, if the main feeder switch is located downstream of the branch feeder switch, they control different network areas, and in this case, both switches must be retained.

[0050] When multiple target switches exist on the same feeder (whether it is a main feeder or a branch feeder), retain the topmost switch in the topology hierarchy and remove all other target switches downstream of it.

[0051] In scenarios involving cross-feeder transfer paths, to avoid accidentally reserving a backup power supply path for the target user due to the accidental removal of a switch, an additional verification rule is added: Before planning to remove a switch (especially a transfer switch) on a branch feeder, it is necessary to determine whether the upstream of the transfer path it is connected to already exists in the initially planned power outage scheme. The circuit breaker in the circuit. If it exists, it means that the transfer path has been cut off and the branch feeder switch can be safely removed; if it does not exist, the branch feeder switch must be kept in the combination with the main feeder switch.

[0052] S402, For switch combinations that have undergone redundancy reduction Structural optimization is performed. This step aims to simplify operation combinations by finding higher-level switches to replace multiple lower-level switches. Specifically, based on the parent-child node attributes between switch nodes, the topology tree is traversed upwards to identify whether a switch exists whose operation can cover the control range of multiple switches in the current combination. For example, if the switch... The operation can be disconnected and All downstream target users under control, and Downstream, besides leading to and If there are no other uninterrupted power supply paths besides the one that is connected, then it can be used. Alternative and The combination of . This process can be represented as: , (hereinafter referred to as (This is a structurally optimized switch combination). This optimization simplifies the switch combination without introducing additional load losses.

[0053] S403. In the multiple optimization stages, additional losses are assessed and optimal solutions are selected. When multiple feasible reduction or alternative solutions emerge in step S401 or step S402, this method will determine the solution based on the total additional losses. The selection should be based on the principle of minimization. Total additional loss refers to the sum of load losses borne by all non-target users due to the switching operation. Its calculation formula is: ; in, This is a structurally optimized switch combination. It is a switch in the combination. It is an operating switch The additional loss contribution value brought about. Specifically, it refers to the reason for disconnecting the switch. The total load of all non-critical users in the area causing the power outage. This was determined by calculating and comparing different solutions. The value is selected based on the switch combination that has the least impact on non-target users.

[0054] Through the iterative process of redundancy reduction, structural optimization, and loss assessment described above, a final planned power outage scheme with fewer operations, lower additional losses, and effective isolation of all target users is generated.

[0055] After completing a series of optimizations in step S04, the final data output stage begins. Step S05 organizes the calculated final planned power outage scheme and outputs it as a set of system vulnerability points, providing a decision-making basis for subsequent grid protection and resilience improvement. This step may specifically include: S501, The final structurally optimized switch combination obtained after redundancy reduction, structural optimization, and loss assessment in step S04. This was determined to be the final result of this identification task. This switch combination constitutes the final planned power outage scheme for multi-point user network attacks, which this invention aims to identify. The final planned power outage scheme is defined as a set of key vulnerabilities in the power distribution system under this simulated attack scenario.

[0056] S502. Output the final planned power outage scheme in structured data format. The output may include the following information: A target set of switches, which contains, in list form, unique identifiers for all switches that constitute a vulnerability.

[0057] Total additional loss associated with this scheme This value is calculated in step S403 to quantify the extent of collateral impact on non-target users caused by implementing this plan.

[0058] A coordinated sequence of switching actions. This sequence is generated in step S04 after an optimization decision is made based on the topological location of the target switch in the network, and clarifies the order of switching operations when executing the power outage plan.

[0059] By outputting the final planned power outage scheme, this method provides power grid managers with a set of critical equipment nodes that can be exploited under specific attack intentions. Based on this, precise protection resource allocation strategies can be formulated in advance for this set of vulnerabilities, or corresponding contingency plan drills can be conducted, thereby effectively improving the operational resilience and proactive defense capabilities of the distribution network in the face of such targeted cyberattacks.

Claims

1. A method for identifying vulnerable points of a complex power distribution system against multi-point cyber attacks, characterized in that, The method comprises the following steps: S01, acquiring topology data of a power distribution network and information of a target user, and constructing a tree-type topology model capable of representing connection relationships of each feeder and having complete attributes; S02, based on the tree-type topology model, classifying the target user according to a position attribute of the target user in the tree-type topology model and whether the target user has a transfer path; S03, for the target users of different categories, generating a preliminary planned power outage scheme including all necessary operations by using a corresponding switch searching strategy; S04, jointly optimizing the preliminary planned power outage scheme to generate an optimized switch combination; S05, outputting the optimized switch combination as a system vulnerability set.

2. The method of claim 1, wherein the method is characterized by, The S02 step specifically comprises: In the tree-type topology model, a continuous path with the largest line capacity is defined as a main feeder, and the remaining lines directly connected to the main feeder are defined as branch feeders, and the target users are preliminarily classified into main feeder users and branch feeder users according to the main feeder and the branch feeders; It is determined whether the target user has a standby transfer path, and the transfer type is distinguished as feeder internal transfer or cross-feeder transfer; The position attribute and the attribute of the transfer path are cross combined to classify the target users into branch feeder users without a transfer path, branch feeder users with feeder internal transfer, branch feeder users with cross-feeder transfer, main feeder users without a transfer path, and main feeder users with a transfer path.

3. The method of claim 1, wherein the method is characterized by, The step of constructing the tree-type topology model in the S01 step comprises: Each node in the tree-type topology model is filled with a core attribute set, and the core attribute set at least includes: a feeder number for identifying a feeder to which the node belongs, a globally unique identifier of the node, a parent node ID recording a directly superior node, a depth representing a path edge number from a root node to the current node, a line hierarchical code reflecting an accurate position of the node in the topology, and a node type identifying a device type.

4. The method of claim 3, wherein the method further comprises: When the node type of the node is a transfer switch, a cross-tree reference field is defined in the core attribute set, which records a feeder number and a globally unique identifier of a corresponding connection switch, wherein the feeder is a feeder connected to a feeder to which the transfer switch itself belongs through the transfer switch.

5. The method of claim 2, wherein the method further comprises: The S03 step specifically comprises, for the branch feeder users without a transfer path: calculating a target user aggregation degree of the branch feeder users without a transfer path on the branch feeder; wherein the target user aggregation degree is calculated by a ratio of the branch feeder users without a transfer path to total target users of the branch feeder; if the target user aggregation degree is greater than a preset threshold, it is determined that the branch feeder users without a transfer path are aggregated target users, and an upstream search is performed to determine a common upstream network accessible circuit breaker closest to the aggregated target users as an operation switch. If the target user concentration is not greater than the preset threshold, it is determined that the branch feeder user of the non-transfer path is a scattered target user, and each scattered target user is independently searched to determine the respective nearest upstream network accessible breaker as an operating switch.

6. The method of claim 2, wherein the method is characterized by, The S03 step specifically includes the following steps for the target user with a transfer path: Searching and determining a first group of operating switches that cut off the main power supply path of the target user with a transfer path; On the transfer path of the target user with a transfer path, searching and determining a second group of operating switches that cut off the backup power supply path according to the principle of minimum additional loss; Combining the first group of operating switches and the second group of operating switches to obtain a set of operating switches for the target user with a transfer path.

7. The method of claim 1, wherein, The S04 step includes the joint optimization by eliminating redundant switch operations, and the rule for eliminating redundant switch operations is: When a main feeder switch in the preliminary planned power cut scheme is located upstream of a branch feeder switch, the main feeder switch is retained, and the branch feeder switch is removed from the scheme.

8. The method of claim 7, wherein the method is characterized by, The S04 step also includes the joint optimization by simplifying the switch combination structure, and the simplified switch combination structure is: Based on the parent-child node attribute between switch nodes, it is identified whether there is a higher level switch by upward traversal, the operation of the higher level switch can cover the control range of multiple switches in the current combination, and there is no other uncut power supply path downstream of the higher level switch, if there is, the multiple switches are replaced by the higher level switch.

9. The method of claim 8, wherein the method further comprises: The S04 step also performs the joint optimization according to the principle of minimizing additional loss, and the principle is specifically: When multiple feasible switch reduction or replacement schemes appear in the joint optimization process, the total additional loss of each scheme is calculated and compared, and the scheme with the minimum total additional loss is preferentially selected; Wherein, the total additional loss is the sum of the load loss borne by all non-target users due to the execution of switch operation.

10. The method of claim 1, wherein, The S05 step specifically outputs the following contents: A target switch set containing the globally unique identifier of all switches constituting the weak point; The total additional loss value associated with the optimized switch combination; And a set of coordinated switch action sequences.

Images