Power thought chain path decision method and system

Abstract

The application provides a power thinking chain path decision method and system, which comprises the following steps: analyzing preliminary fault causes of power system faults, expanding each preliminary fault cause based on a workflow engine according to fault handling logic combined with a fault propagation path of each preliminary fault cause in a power knowledge graph to obtain a target fault development path; taking each target fault development path as a fault handling process node of a workflow, arranging steps based on the correlation between the nodes to obtain an initial workflow instance; performing recovery effective verification on the state change information of the power system after the simulation execution of the processing steps of each initial workflow instance detected based on the power knowledge graph to obtain a target workflow instance; evaluating each target workflow instance based on multi-dimensional indexes to obtain an optimal workflow instance, and determining the optimal workflow instance as a power thinking chain path decision result of the power system fault. The application guarantees the safe and stable operation of the power system.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method and system for power thinking chain path decision-making. Background Technology

[0002] As a critical infrastructure for modern society, the stable and reliable operation of the power system is essential for industrial production and residents' lives. However, due to its massive scale and complex structure, various faults are inevitable during operation. If the cause of a fault is not analyzed promptly and accurately, and a reasonable handling path is not planned, it can not only lead to power outages and affect users' normal electricity use, but may also cause equipment damage, fires, and other problems.

[0003] Existing decision-making methods for power system faults typically rely on pre-defined rule bases and simple fault matching logic. This involves comparing real-time monitored fault data with standard fault patterns in the rule base to derive a fault handling path. However, these methods lack dynamic adaptability and knowledge reasoning capabilities for complex fault scenarios. When multiple fault points overlap, new fault types emerge, or fault information is incomplete or contradictory, existing methods struggle to accurately analyze the causes of the faults and rationally plan the optimal fault handling path. This results in low fault handling efficiency and may even trigger secondary faults, impacting the safe and stable operation of the power system. Summary of the Invention

[0004] The purpose of this invention is to provide a power thinking chain path decision-making method and system. By utilizing the reasoning capabilities of the power knowledge graph, it improves the logical expansion of the fault development path of each initial fault cause, and can deeply explore the potential complex fault scenarios of each initial fault cause. Then, based on the correlation between the processing steps of each target fault development path, the steps are arranged, comprehensively sorting out the fault handling ideas and making the fault handling process more standardized. Even when multiple fault points overlap, new fault types occur, or fault information is missing or contradictory, the fault cause can be accurately analyzed. This enables the rapid and accurate planning of the optimal fault handling decision path under complex fault conditions, ensuring the safe and stable operation of the power system.

[0005] In a first aspect, the present invention provides a power thinking chain path decision-making method, including: Fault analysis is performed on power system faults to obtain preliminary fault causes. Based on the workflow engine, each preliminary fault cause is expanded according to the fault handling logic and the fault propagation path in the pre-constructed power knowledge graph to obtain the target fault development path. Each target fault development path is used as a fault handling process node in the workflow. The steps are arranged based on the relationship between the processing steps of the fault handling process nodes to obtain an initial workflow instance. Based on the power knowledge graph, the processing steps of each initial workflow instance are detected and the state change information of the power system after simulation execution is obtained. The recovery validity verification of each initial workflow instance is performed to obtain the target workflow instance. Each target workflow instance is evaluated based on multi-dimensional indicators to obtain the optimal workflow instance, and the optimal workflow instance is determined as the power thinking chain path decision result for power system faults.

[0006] Optionally, the workflow engine expands each initial fault cause according to the fault handling logic and the fault propagation path of each initial fault cause in the pre-constructed power knowledge graph to obtain the target fault development path, including: Taking each initial fault cause as the starting node, the fault handling logic based on the workflow engine traverses the first downstream node directly connected to the starting node in the power knowledge graph to obtain the next layer node. Based on the connection relationships between nodes in the power knowledge graph and the fault handling logic of the workflow engine, it is determined whether there is an unexplored second downstream node for each next-level node; if so, the unexplored second downstream node is identified as an indirect layer node. Based on the power knowledge graph, the connection relationship between each indirect layer node and the other nodes except itself is determined, and the first candidate node corresponding to the connection relationship that meets the fault propagation condition is selected according to the fault handling logic of the workflow engine. The fault handling logic based on the workflow engine combines the attribute information of each first candidate node to determine whether there are abnormal nodes that affect the direction of fault propagation; if so, the first candidate node is updated based on the abnormal node to obtain the second candidate node. Based on the second candidate node, the indirect layer node, and the next layer node, each initial fault cause is expanded to obtain the target fault development path.

[0007] Optionally, the step of expanding each preliminary fault cause based on the second candidate node, indirect layer nodes, and next-layer nodes to obtain the target fault development path includes: Starting with each second candidate node, a reverse traversal is performed in the power knowledge graph to determine whether there is a first target connection edge in the power knowledge graph that directly connects each second candidate node to the next layer node or the indirect layer node; if so, the first target node is obtained by traversing along the first target connection edge in the next layer node or the indirect layer node. Based on the first target node, a reverse traversal is performed in the power knowledge graph to determine whether there is a second target connection edge in the power knowledge graph that directly connects the first target node to the starting node. If there is, the traversal is performed along the second target connection edge to obtain the first starting node, and an initial fault development path is constructed based on the nodes and connection edges traversed from each second candidate node to the first starting node. If there is no connection edge, the traversal is performed in the next layer of nodes or indirect layer nodes based on the first target node until the second starting node is obtained, and an initial fault development path is constructed based on the nodes and connection edges traversed from each second candidate node to the second starting node. The fault handling logic based on the workflow engine, combined with the functional and interrelationships of nodes on each initial fault development path determined by the power knowledge graph, determines whether there is node convergence or node fork in multiple initial fault development paths. If node convergence occurs, the initial fault development path with node convergence will be merged by the converging node to obtain the target fault development path; If a node fork occurs, the initial fault development path with the forked node will be split into the forked nodes to obtain the target fault development path.

[0008] Optionally, the fault handling process node, which uses each target fault development path as a workflow node, arranges steps based on the correlation between the processing steps of the fault handling process node to obtain an initial workflow instance, including: Based on the positional and connection relationships of each fault handling process node in the power knowledge graph, determine whether there are directly connected fault handling process nodes; if so, identify the directly connected fault handling process nodes as nodes of the same branch. The initial step arrangement is performed on each node in the same branch according to the connection order and dependency relationship between the processing steps of the nodes in the same branch, and the preliminary step arrangement result is obtained. Based on the fact that all fault handling process nodes are traversed in the same branch node, fault handling process nodes that do not belong to any same branch node are identified as cross branch nodes. Based on the initial step arrangement results and the relationship between processing steps of nodes in the same branch and nodes across branches, a secondary step arrangement is performed to obtain the initial workflow instance.

[0009] Optionally, the secondary step arrangement based on the initial step arrangement results, combined with the correlation between processing steps within the same branch node and across branch nodes, yields an initial workflow instance, including: The power knowledge graph is traversed according to the dependencies between processing steps to determine whether there is a third target node connected to the same branch node in the cross-branch node; if there is a third target node, the fourth target node connected to the same branch node is determined. Based on the edge attributes and edge functions of each third target node and each fourth target node in the power knowledge graph, the connection priority between each third target node and each fourth target node is determined. Using each third target node as an intermediate orchestration node, establish connections between fourth target nodes with the same connection priority to construct the first orchestration thinking link. Then, using any fourth target node with each connection priority as a node to be orchestrated, connect each node to be orchestrated according to the priority order of the connection priority based on the intermediate orchestration node to construct the second orchestration thinking link. Based on the node position of the fourth target node in the preliminary step arrangement result, the first and second arrangement thought links are inserted into the preliminary step arrangement result to obtain the initial workflow instance.

[0010] Optionally, the step of recovering and validating the processing steps of each initial workflow instance detected by the power knowledge graph based on the power system state change information after simulation execution, to obtain the target workflow instance, includes: After the processing steps of each initial workflow instance are simulated, the changes in the operating parameters of each system node in the power system are compared with the standard range of the operating parameters of each system node when the power system is running normally, and the comparison results of each system node are obtained. Based on the comparison results of each system node, abnormal system nodes are identified. Based on the connection relationship of abnormal system nodes of each initial workflow instance in the power knowledge graph, it is determined whether there is an abnormal region formed by multiple abnormal system nodes. If there are abnormal regions, the abnormal regions of each initial workflow instance are traversed in the power knowledge graph to obtain the boundary nodes between each abnormal region and the normal operating node region; the boundary nodes represent nodes that belong to both the abnormal region and the normal operating node region. The target workflow instance is obtained by performing recovery validity verification on each initial workflow instance based on the boundary node of each initial workflow instance.

[0011] Optionally, the step of performing effective recovery verification on each initial workflow instance based on the boundary nodes of each initial workflow instance to obtain the target workflow instance includes: Based on the difference between the changes in the operating parameters of the boundary nodes in each initial workflow instance and the standard range of the operating parameters, the severity of the fault is determined, and the total severity of the fault for each initial workflow instance is determined based on the severity of the fault for each boundary node. Candidate workflow instances are determined based on the sum of the fault severity of each initial workflow instance; Based on the power knowledge graph, the degree of coverage of the power system state changes on the fault development path nodes after the execution of each candidate workflow instance is detected, and the candidate workflow instances with a change coverage degree greater than a preset coverage degree threshold are determined as target workflow instances.

[0012] Secondly, the present invention also provides a power thinking chain path decision-making system, applied to the power thinking chain path decision-making method as described in the first aspect; the power thinking chain path decision-making system includes: The fault path expansion module is used to perform fault analysis on power system faults, obtain preliminary fault causes, and expand each preliminary fault cause according to the fault handling logic and the fault propagation path in the pre-constructed power knowledge graph based on the workflow engine, to obtain the target fault development path. The workflow instance orchestration module is used to take each target fault development path as a fault handling process node in the workflow, and to orchestrate the steps based on the relationship between the processing steps of the fault handling process node to obtain the initial workflow instance. The recovery validity verification module is used to perform recovery validity verification on each initial workflow instance based on the power system state change information after the simulation execution of the processing steps of each initial workflow instance detected by the power knowledge graph, so as to obtain the target workflow instance. The workflow instance evaluation module is used to evaluate each target workflow instance based on multi-dimensional indicators, obtain the optimal workflow instance, and determine the optimal workflow instance as the power thinking chain path decision result for power system faults.

[0013] Thirdly, the present invention also provides an electronic device, comprising: a memory for storing computer software programs; and a processor for reading and executing the computer software programs, thereby realizing the power thinking chain path decision-making method as described above.

[0014] Fourthly, the present invention also provides a non-transitory computer-readable storage medium storing a computer software program, which, when executed by a processor, implements the power thinking chain path decision-making method as described above.

[0015] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the power thinking chain path decision-making method as described above.

[0016] The power thinking chain path decision-making method provided in this invention expands each initial fault cause based on a workflow engine, combining fault handling logic with fault propagation paths in a power knowledge graph. Guided by fault handling logic, it utilizes the reasoning capabilities of the power knowledge graph to refine the logical expansion of the fault development path for each initial fault cause. Therefore, it can deeply uncover the potential complex fault scenarios of each initial fault cause. Then, based on the correlation between the processing steps of each target fault development path, it arranges the steps, comprehensively streamlining the fault handling approach and making the fault handling process more streamlined. Even when multiple fault points overlap, new fault types arise, or fault information is missing or contradictory, it can accurately analyze the fault cause. Furthermore, by simulating the execution of workflow instances, it obtains effective workflow instances capable of restoring normal system operation. Finally, based on multi-dimensional indicators, it selects the optimal workflow instance from the effective workflow instances, determining the best power thinking chain path decision-making result for handling system faults. This achieves rapid and accurate planning of the optimal fault handling decision-making path under complex fault conditions, ensuring the safe and stable operation of the power system. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the power thinking chain path decision-making method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the power thinking chain path decision-making system provided in an embodiment of the present invention; Figure 3 An embodiment diagram of the electronic device provided in this invention; Figure 4 An embodiment diagram of a computer-readable storage medium provided in accordance with the present invention. Detailed Implementation

[0018] The technical solutions of 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.

[0019] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0020] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0021] Optional, see below Figure 1 , Figure 1 This is a flowchart illustrating the power thinking chain path decision-making method provided by the present invention. In this embodiment, the executing entity of the power thinking chain path decision-making method is a fault decision-making system. Therefore, the power thinking chain path decision-making method includes: Step 10: Perform fault analysis on the power system fault to obtain preliminary fault causes. Based on the workflow engine, expand each preliminary fault cause according to the fault handling logic and the fault propagation path in the pre-constructed power knowledge graph to obtain the target fault development path.

[0022] Optionally, when a power system fault occurs, the fault decision system uses a fault diagnosis algorithm combined with real-time collected power system operating data, such as voltage, current, and power parameters, as well as equipment status information, to identify the fault type, fault location, and other preliminary causes of the fault. In one embodiment, a transmission line in the power system experiences a tripping fault. Data such as voltage drops and current surges at the substations at both ends of the line are collected. Through analysis using a fault diagnosis algorithm, the preliminary cause of the fault is determined to be a short circuit in the transmission line.

[0023] Furthermore, the fault decision-making system, based on a workflow engine, searches for the corresponding fault propagation path in a pre-built power knowledge graph for each initial fault cause. Here, the power knowledge graph in this embodiment is a structured knowledge base that stores information such as the connection relationships between devices in the power system, device characteristics, fault modes, and propagation patterns. The workflow engine, following a predetermined fault handling logic, expands the initial fault causes along the fault propagation path, and determines the complete target fault development path by analyzing the devices and areas that the fault may affect, as detailed in steps 101 to 105.

[0024] Step 20: Using each target fault development path as a fault handling process node in the workflow, arrange the steps based on the relationship between the processing steps of the fault handling process nodes to obtain an initial workflow instance.

[0025] Furthermore, the fault decision system takes each target fault development path as a fault handling process node in the workflow, analyzes the connection order and dependency relationship between each processing step in the fault handling process node. For example, some steps can only be executed after certain conditions are met, or there are sequential requirements between some steps. Based on the correlation, the processing steps are reasonably arranged, and each node and step is organized into an ordered process to obtain an initial workflow instance, specifically as in steps 201 to 204.

[0026] Step 30: Based on the power knowledge graph, the processing steps of each initial workflow instance detected are used to verify the state change information of the power system after simulation execution. The result is the target workflow instance.

[0027] Furthermore, the fault decision-making system, based on a power knowledge graph, simulates the processing steps of each initial workflow instance. During the simulation, it records the state changes of the power system after each step, such as voltage and current recovery status and changes in equipment operating status.

[0028] Furthermore, the fault decision system determines whether the initial workflow instance can effectively restore the normal operation of the power system based on the power system's state change information. If the power system can recover to a stable operating state after the simulation execution of an initial workflow instance, the initial workflow instance is determined to have passed the verification; otherwise, the initial workflow instance is adjusted or rearranged until a target workflow instance that can effectively restore system operation is obtained, as described in steps 301 to 304.

[0029] Step 40: Evaluate each target workflow instance based on multi-dimensional indicators to obtain the optimal workflow instance, and determine the optimal workflow instance as the power thinking chain path decision result for power system faults.

[0030] Furthermore, the fault decision-making system evaluates each target workflow instance based on multi-dimensional indicators. These indicators, in this embodiment, include, but are not limited to, fault handling time and resource consumption. Fault handling time refers to the time spent from the start of executing the workflow instance to the restoration of normal power system operation. Resource consumption encompasses human, material, and financial resources, such as the number of personnel involved in fault handling, and the cost of equipment and materials used. Therefore, the fault handling time and resource consumption of each target workflow instance are quantitatively analyzed. By comparing the performance of different target workflow instances in terms of fault handling time and resource consumption, the optimal workflow instance is determined. Further, the fault decision-making system uses the optimal workflow instance as the power system fault path decision result in the power system's thought chain.

[0031] In one embodiment, assume two target workflow instances for a short-circuit fault in a transmission line. Instance A has a fault handling time of 2 hours, requires 5 professional maintenance personnel, uses 2 detection devices and some repair materials, with an estimated resource consumption cost of 5000 yuan; Instance B has a fault handling time of 3 hours, requires 3 professional maintenance personnel, uses 1 detection device and fewer repair materials, with an estimated resource consumption cost of 3000 yuan. The fault decision system comprehensively evaluates the fault handling time and resource consumption of the two instances, calculating that Instance A has higher fault handling efficiency but higher resource consumption; Instance B has lower resource consumption but longer fault handling time. By setting weights (e.g., fault handling time weight 0.6, resource consumption weight 0.4), the two instances are weighted and scored. Finally, Instance A with the higher score is determined as the optimal workflow instance, and it is used as the power thinking chain path decision result for this short-circuit fault in the transmission line to guide the actual fault handling work.

[0032] This invention expands each initial fault cause based on a workflow engine, combining fault handling logic with fault propagation paths in a power knowledge graph. Guided by fault handling logic, it leverages the reasoning capabilities of the power knowledge graph to refine the logical expansion of the fault development path for each initial fault cause. This allows for in-depth exploration of potentially complex fault scenarios for each initial fault cause. Furthermore, it arranges steps according to the relationships between the processing steps of each target fault development path, comprehensively streamlining the fault handling approach and streamlining the fault handling process. Even when multiple fault points overlap, new fault types arise, or fault information is missing or contradictory, the fault cause can be accurately analyzed. By simulating the execution of workflow instances, effective workflow instances capable of restoring normal system operation are obtained. Finally, based on multi-dimensional indicators, the optimal workflow instance is selected from the effective workflow instances, determining the best power thinking chain path decision result for handling system faults. This enables rapid and accurate planning of the optimal fault handling decision path under complex fault conditions, ensuring the safe and stable operation of the power system.

[0033] In one embodiment, steps 101 to 105 include: Step 101: Starting with the initial fault cause as the starting node, the fault handling logic of the workflow engine is used to traverse the first downstream node directly connected to the starting node in the power knowledge graph to obtain the next layer node.

[0034] Optionally, the fault decision system takes each initial fault cause as the starting node and uses the fault handling logic of the workflow engine to traverse the power knowledge graph. The workflow engine searches for nodes directly connected to the starting node according to the connection relationship between nodes in the power knowledge graph according to the pre-set rules. The directly connected nodes are the first downstream nodes, and all the first downstream nodes together constitute the next layer of nodes.

[0035] Continuing with the above example of a transmission line short-circuit fault, the initial cause of the fault is a short circuit in the transmission line, which is the starting node. The fault decision system, based on the fault handling logic of the workflow engine, searches the power knowledge graph for nodes directly connected to this transmission line node. The power knowledge graph shows that the transmission line directly connects to the substation nodes at both ends. Therefore, these two substation nodes are identified as the first downstream nodes, forming the next layer of nodes, indicating that the short-circuit fault may directly affect these two substations.

[0036] Step 102: Based on the connection relationships between nodes in the power knowledge graph and the fault handling logic of the workflow engine, determine whether each next-level node has an unexplored second downstream node. If so, the unexplored second downstream node is identified as an indirect layer node.

[0037] Furthermore, based on the connections between nodes in the power knowledge graph and the fault handling logic of the workflow engine, the fault decision system checks each next-level node to determine if there are any unexplored second downstream nodes. If unexplored second downstream nodes exist, the fault decision system identifies these nodes as indirect layer nodes. Continuing with the two identified substation nodes, the fault decision system checks them based on the connections in the power knowledge graph and the fault handling logic of the workflow engine. It discovers that one of the substation nodes is connected to multiple distribution transformer nodes, and these distribution transformer nodes have not been explored in previous steps. Therefore, these distribution transformer nodes are identified as indirect layer nodes, indicating that a short-circuit fault may indirectly affect these distribution transformers through the substation.

[0038] Step 103: Based on the power knowledge graph, determine the connection relationship between each indirect layer node and the other nodes except itself, and filter out the first candidate node corresponding to the connection relationship that meets the fault propagation condition according to the fault handling logic of the workflow engine.

[0039] Furthermore, based on the power knowledge graph, the fault decision-making system determines all connections between each indirect layer node and all other nodes except itself, and filters these connections according to the fault handling logic of the workflow engine. The fault handling logic sets a series of fault propagation conditions, such as the electrical connection characteristics between devices, the compatibility between fault types and devices, etc. Only nodes corresponding to connections that meet these fault propagation conditions will be filtered out and become the first candidate nodes.

[0040] Continuing with the identified distribution transformer node (indirect layer node), the fault decision-making system examines its connections to other nodes in the power knowledge graph, discovering that the distribution transformer node is connected to multiple user-end nodes. According to the workflow engine's fault handling logic, only when the distribution transformer fails will it affect the user-end nodes directly connected to it and on the same power supply circuit. After screening, a subset of user-end nodes on the same power supply circuit were identified as the first candidate nodes; these are the nodes that are likely to be affected by the fault under the current fault condition.

[0041] Step 104: Based on the fault handling logic of the workflow engine and the attribute information of each first candidate node, determine whether there are any abnormal nodes that affect the direction of fault propagation. If so, update the first candidate nodes based on the abnormal nodes to obtain the second candidate nodes.

[0042] Furthermore, based on the workflow engine's fault handling logic and the attribute information of each first candidate node, such as equipment type, operating status, and fault threshold, the fault decision system determines whether there are any abnormal nodes affecting the direction of the fault propagation path. An abnormal node is one whose attribute information does not conform to normal fault propagation patterns, potentially altering the fault propagation path. If an abnormal node exists, the fault decision system updates the first candidate nodes accordingly, excluding nodes that do not conform to actual fault propagation and adding nodes that may be affected by the fault due to the abnormal node, thus obtaining second candidate nodes and making the fault propagation path analysis more accurate.

[0043] Continuing with the first candidate user-end nodes identified above, the fault decision system discovered that one of these nodes had a backup power supply installed, and this backup power supply was in a normal and available state. According to the workflow engine's fault handling logic, this node would not be affected by a power distribution transformer failure due to the presence of the backup power supply, thus classifying it as an abnormal node. Therefore, the fault decision system excluded this node from the first candidate nodes. Simultaneously, it was discovered that the power supply line of this backup power supply device also connected to several other user-end nodes. These nodes might be affected by the abnormality of the backup power supply line during a main power failure; therefore, these nodes were added, resulting in an updated second candidate node list.

[0044] Step 105: Based on the second candidate node, the indirect layer node, and the next layer node, expand each preliminary fault cause to obtain the target fault development path.

[0045] Furthermore, the fault decision system expands each preliminary fault cause based on the second candidate node, the indirect layer node, and the next layer node to obtain the target fault development path, as shown in steps 1051 to 1055.

[0046] Starting from the initial cause of the fault, this invention continuously explores potential influencing nodes, and the final target fault development path can accurately and comprehensively present the complete chain from the initial fault to all potentially affected equipment and areas. It clearly reveals the direction of fault propagation and the scope of impact in the power system. Therefore, it can deeply explore the potential complex fault scenarios of each initial fault cause, thereby carrying out fault repair work in a more targeted manner and reducing the impact of faults on the normal operation of the power system.

[0047] In one embodiment, steps 1051 to 1055 include: Step 1051: Starting with each second candidate node, perform a reverse traversal of the power knowledge graph to determine whether each second candidate node has a first target connection edge that directly connects to the next layer node or the indirect layer node. If it exists, traverse along the first target connection edge to obtain the first target node.

[0048] Optionally, the fault decision system uses each second candidate node as a starting node and performs a reverse traversal operation in the power knowledge graph. The reverse traversal involves searching along the node connection edges from the second candidate node towards its upstream nodes. Therefore, the fault decision system checks all connection edges of each second candidate node in the power knowledge graph to determine if there exists a connection edge that can directly connect to the next layer node or an indirect layer node, i.e., the first target connection edge.

[0049] Furthermore, if the first target connection edge is found, the fault decision system continues to traverse the next layer of nodes or the indirect layer of nodes along the first target connection edge, and the finally determined connection node is the first target node.

[0050] Continuing with the embodiment of transmission line short-circuit faults, some user-end nodes have been identified as second candidate nodes. The fault decision system uses one of these user-end nodes as the starting node and performs a reverse traversal of the power knowledge graph. It discovers that this user-end node has a connection edge directly connecting to a previously identified distribution transformer node (indirect layer node). This connection edge is the first target connection edge. Along the first target connection edge, the distribution transformer node is determined as the first target node, clarifying the direct connection relationship between the user-end node and the distribution transformer node.

[0051] Step 1052: Based on the first target node, perform a reverse traversal of the power knowledge graph to determine whether there is a second target connection edge in the power knowledge graph that directly connects the first target node to the starting node. If it exists, traverse along the second target connection edge to obtain the first starting node, and construct an initial fault development path based on the nodes and connections traversed from each second candidate node to the first starting node. If it does not exist, traverse the nodes in the next layer or indirect layer based on the first target node until the second starting node is obtained, and construct an initial fault development path based on the nodes and connections traversed from each second candidate node to the second starting node.

[0052] Furthermore, the fault decision-making system uses the first target node as a new starting point and performs a reverse traversal of the power knowledge graph again, checking all the connecting edges of the first target node to determine whether there is a connecting edge that can directly connect to the starting node (i.e., the node corresponding to the initial fault cause), i.e., a second target connecting edge. If a second target connecting edge exists, the fault decision-making system continues to traverse along the second target connecting edge until it finds the starting node, which is then recorded as the first starting node. Further, the fault decision-making system records all the nodes and connecting edges traversed from each second candidate node to the first starting node, and connects these nodes and connecting edges sequentially to construct an initial fault development path.

[0053] Furthermore, if no second target connecting edge exists, the fault decision system continues to iterate through the next layer of nodes or the indirect layer of nodes based on the first target node, continuously searching for a path that can connect to the starting node until a corresponding starting node is found, which is then recorded as the second starting node. Further, the fault decision system records all nodes and connecting edges traversed from the second candidate node to the second starting node, constructing another initial fault development path.

[0054] Continuing with the identified distribution transformer node (first target node), the fault decision system performs a reverse traversal and discovers a connecting edge that directly connects to the short-circuited transmission line node (starting node), i.e., the second target connecting edge. Following this second target connecting edge, the transmission line node is found as the first starting node. All nodes and connecting edges traversed from the user-end node (second candidate node) to the distribution transformer node (first target node) and then back to the transmission line node (first starting node) are recorded, constructing an initial fault development path: user-end node → distribution transformer node → transmission line node.

[0055] Assuming that the first target node corresponding to another second candidate node does not find a connection edge directly connected to the transmission line node during the reverse traversal, the fault decision system continues to traverse the indirect layer nodes (other distribution transformer nodes and the next layer nodes (substation nodes)). Finally, by connecting to the substation node and then connecting from the substation node to the transmission line node, the second starting node is determined to be the transmission line node, and another initial fault development path is constructed.

[0056] Step 1053: Based on the fault handling logic of the workflow engine and the functional and interrelationships of the nodes on each initial fault development path determined by the power knowledge graph, determine whether there is node convergence or node fork in multiple initial fault development paths.

[0057] Furthermore, based on the fault handling logic of the workflow engine, the fault decision system combines the functional relationships and interrelationships of the nodes on each initial fault development path determined in the power knowledge graph to perform an overall analysis of all initial fault development paths. It checks each path one by one to determine whether multiple initial fault development paths converge at a certain node, i.e., node convergence occurs; or whether an initial fault development path derives multiple branch paths at a certain node, i.e., node bifurcation occurs.

[0058] Continuing with the construction of multiple initial fault development paths, the fault decision system identified two paths: Path A is user node A → distribution transformer node A → substation node → transmission line node; Path B is user node B → distribution transformer node B → substation node → transmission line node. These two paths converge at the substation node, indicating a node convergence. Simultaneously, another path was identified: user node C → distribution transformer node C → branch node → user node D and user node E. At the branch node, two branching paths appeared, indicating a node bifurcation.

[0059] Step 1054: If node convergence occurs, the initial fault development path with node convergence is merged by the converging node to obtain the target fault development path.

[0060] Furthermore, when it is determined that there is a node convergence situation, the fault decision system merges all the initial fault development paths with the node convergence as the reference.

[0061] Optionally, during the merging process, the independent parts of each path before the intersection node are retained, and the paths after the intersection node are integrated into one path, thus obtaining an integrated target fault propagation path. Therefore, by merging paths, duplicate paths are eliminated and the fault propagation path is simplified.

[0062] Continuing with the two initial fault propagation paths (Path A and Path B) converging at the substation node, the substation node is used as the convergence point for merging. The merged target fault propagation path is: User-end node A → Distribution transformer node A → Substation node, User-end node B → Distribution transformer node B → Substation node → Transmission line node. This unifies the portions of the two originally independent paths after the substation node into a single path leading to the transmission line node, clearly demonstrating the propagation process of the fault from different user-end nodes through different distribution transformer nodes to the substation, and then affecting the transmission line.

[0063] Step 1055: If a node fork occurs, the initial fault development path with node fork is split by the fork node to obtain the target fault development path.

[0064] Furthermore, when a node fork is identified, the fault decision system splits the initial fault development path with the fork node as the baseline. During the splitting process, each branch path after the fork node is isolated, forming multiple independent paths. Each path starts from the starting node, passes through the fork node, and extends to its respective end node, resulting in multiple target fault development paths. Therefore, by splitting the paths, the propagation of the fault on different branches is clearly displayed.

[0065] Continuing with the path that branches at the branch node (User node C → Distribution transformer node C → Branch node → User node D and User node E), we further split it based on the branch node. This results in two target fault propagation paths: Path 1 is User node C → Distribution transformer node C → Branch node → User node D; Path 2 is User node C → Distribution transformer node C → Branch node → User node E. These represent two independent paths from User node C through Distribution transformer node C and the branch node, propagating to User node D and User node E respectively.

[0066] This invention constructs an initial path through reverse traversal and combines it with fault handling logic to determine node convergence and branching. It then performs targeted path merging and splitting to ultimately form the target fault development path. Therefore, it can accurately identify the complete propagation path network from the initial fault cause to the affected nodes. By integrating and splitting the paths, it eliminates redundant and duplicate paths and clearly shows the propagation path of the fault on different branches, enabling a direct understanding of the fault's impact range and propagation context.

[0067] In one embodiment, steps 201 to 204 include: Step 201: Based on the positional and connectivity relationships of each fault handling process node in the power knowledge graph, determine whether there are directly connected fault handling process nodes. If so, identify the directly connected fault handling process nodes as nodes in the same branch.

[0068] Optionally, the fault decision system takes the fault handling process nodes in each target fault development path as objects. Based on the node positional and connection relationships recorded in the power knowledge graph, it analyzes each fault handling process node one by one to determine whether there is a direct connection between any two fault handling process nodes. In this embodiment of the invention, a direct connection means that there is a direct connection edge in the power knowledge graph, representing that the two nodes are directly related in the actual structure of the power system. If a direct connection relationship exists, the fault decision system determines these two fault handling process nodes as nodes on the same branch, meaning that they are on the same path branch for fault propagation or handling in the power system.

[0069] Continuing with the transmission line short-circuit fault embodiment, the target fault development path includes two branch paths: user-end node C → distribution transformer node C → branch node → user-end node D and user-end node E, and two convergence paths: user-end node A → distribution transformer node A → substation node → transmission line node and user-end node B → distribution transformer node B → substation node → transmission line node. The fault decision-making system analyzes the location and connection relationships of these fault handling process nodes (user-end nodes, distribution transformer nodes, branch nodes, substation nodes, and transmission line nodes) in the power knowledge graph. It finds that user-end node A and distribution transformer node A are directly connected in the graph, and distribution transformer node A and substation node are also directly connected. Therefore, user-end node A, distribution transformer node A, and substation node are identified as nodes on the same branch. Similarly, user-end node B, distribution transformer node B, and substation node are also identified as nodes on the same branch. In the branch path, user-end node C, distribution transformer node C, branch node, and user-end node D are nodes on the same branch, and user-end node C, distribution transformer node C, branch node, and user-end node E are nodes on the same branch.

[0070] Step 202: Perform initial step arrangement for each node in the same branch according to the connection order and dependency relationship between the processing steps of the nodes in the same branch, and obtain the preliminary step arrangement result.

[0071] Furthermore, for each defined set of nodes in the same branch, the fault decision system arranges the processing steps of the fault handling process nodes according to the connection order and dependencies. The connection order refers to the order in which each step is executed in actual fault handling; the dependency relationship indicates that some steps can only be executed after specific conditions are met. For example, the detection step of a certain device needs to be performed after the device is powered off. Therefore, according to the connection order and dependencies, the processing steps in the nodes in the same branch are arranged sequentially to form a preliminary step arrangement result, so that the fault handling process of the nodes in the same branch has a clear execution order.

[0072] Continuing with the above embodiments, for a set of nodes in the same branch consisting of user-end node A, distribution transformer node A, and substation node, the fault handling process involves: performing status detection on the substation node (determining whether the relay protection is operating correctly, etc.); then troubleshooting the distribution transformer node A; and finally, addressing the power restoration issue for user-end node A. Based on the connection sequence and dependencies of these processing steps, the fault decision system arranges them as follows: substation node status detection → distribution transformer node A fault troubleshooting → user-end node A power restoration, thus obtaining the preliminary step arrangement result for this branch node.

[0073] Step 203: Based on the traversal of all fault handling process nodes within the same branch node, fault handling process nodes that do not belong to any same branch node are identified as cross-branch nodes.

[0074] Furthermore, the fault decision system takes all fault handling process nodes as objects and performs a traversal check within the already determined range of nodes on the same branch to determine whether each fault handling process node belongs to any of the already determined sets of nodes on the same branch. If it does not belong to any set of nodes on the same branch, the fault decision system determines the fault handling process node as a cross-branch node.

[0075] Continuing with the above embodiment, after steps 201 and 202, multiple sets of nodes on the same branch are determined. When the fault decision system traverses all fault handling process nodes, it finds that the branch node is involved in different branch paths (connecting user terminal node D and user terminal node E) and does not belong to any of the determined sets of nodes on the same branch. Therefore, the branch node is determined as a cross-branch node. In addition, although the substation node is present in the sets of nodes on the same branch of both convergence paths, it also plays the role of connecting different paths to the transmission line node, and is also determined as a cross-branch node in the broader fault handling process.

[0076] Step 204: Based on the preliminary step arrangement results and the relationship between the processing steps of nodes in the same branch and nodes across branches, perform a secondary step arrangement to obtain the initial workflow instance.

[0077] Furthermore, the fault decision system performs a secondary step arrangement based on the preliminary step arrangement results and the correlation between the processing steps of nodes in the same branch and nodes across branches, to obtain an initial workflow instance, as shown in steps 2041 to 2044.

[0078] This invention analyzes the relationships between fault handling process nodes in the power knowledge graph, first identifying nodes on the same branch and performing initial orchestration, then identifying nodes across branches, and finally completing secondary orchestration by combining the relationships between the processing steps of various nodes to obtain an initial workflow instance. Therefore, based on the actual structure of the power system and the fault handling logic, it can rationally and orderly orchestrate each processing step in the target fault development path. By distinguishing between nodes on the same branch and nodes across branches, it fully considers the independence and correlation of the fault handling process on different path branches, ensuring that the initial workflow instance not only guarantees the continuity of processing steps in each path branch but also achieves effective coordination of cross-path steps, thereby improving the accuracy of fault handling.

[0079] In one embodiment, steps 2041 to 2044 include: Step 2041: Traverse the power knowledge graph according to the dependencies between processing steps to determine whether there is a third target node connected to the same branch node among the cross-branch nodes. If a third target node exists, determine the fourth target node connected to the same branch node.

[0080] Optionally, the fault decision-making system performs a traversal operation in the power knowledge graph based on the dependencies between processing steps, using cross-branch nodes as the search object, and checks whether each cross-branch node has a direct connection with a node in the same branch. If such a connection exists, the fault decision-making system identifies the cross-branch node as the third target node, and the node in the same branch directly connected to it as the fourth target node.

[0081] Continuing with the transmission line short-circuit fault example, the branch node and substation node are known to be cross-branch nodes. Based on the processing step dependencies, the fault decision system traverses the power knowledge graph and finds that the branch node is connected to the same branch nodes of user-end node C and distribution transformer node C. Therefore, the branch node is identified as the third target node, and user-end node C and distribution transformer node C are identified as the corresponding fourth target nodes. Simultaneously, the substation node is connected to the same branch nodes of user-end node A, distribution transformer node A, user-end node B, and distribution transformer node B. Therefore, the substation node is also identified as the third target node, and the connected user-end node A, distribution transformer node A, user-end node B, and distribution transformer node B are the corresponding fourth target nodes.

[0082] Step 2042: Based on the edge attributes and edge functions of the nodes of each third target node and each fourth target node in the power knowledge graph, determine the connection priority between each third target node and each fourth target node.

[0083] Furthermore, for each pair of third and fourth target nodes, the fault decision system determines the connection priority between them based on the edge attributes and edge functions of the nodes in the power knowledge graph. Edge attributes include electrical parameters of the node connection (such as current capacity, voltage level), physical distance, etc.; edge functions involve the business associations between nodes (such as whether it is a backbone transmission path or a backup line). The fault decision system then comprehensively analyzes these attributes and functions according to preset rules, assigning priority to each pair of connections. Connections with higher priority are processed more frequently in subsequent orchestration.

[0084] Continuing with the above embodiments, taking the connection between the branch node (third target node) and user terminal node C (fourth target node) and distribution transformer node C (fourth target node) as an example, the power knowledge graph shows that the connection edge between the branch node and distribution transformer node C carries the main power transmission task, has a large current capacity and no backup line, while the connection edge with user terminal node C is a branch line with a backup path. According to preset rules, the fault decision system determines that the connection between the branch node and distribution transformer node C has a higher priority than the connection with user terminal node C. Similarly, for the connection between the substation node and other fourth target nodes, the priority ranking is determined based on the electrical parameters and business functions of each connection edge. For example, the connection between the substation node and distribution transformer node A has a higher priority than the connection with user terminal node A.

[0085] Step 2043: Using each third target node as an intermediate orchestration node, establish connections between fourth target nodes with the same connection priority to construct the first orchestration thinking link. Then, using any fourth target node with each connection priority as a node to be orchestrated, connect each node to be orchestrated according to the priority order of the connection priority based on the intermediate orchestration node to construct the second orchestration thinking link.

[0086] Furthermore, the fault decision-making system uses each third target node as an intermediate orchestration node, first connecting fourth target nodes with the same connection priority to construct the first orchestration thought chain. This chain reflects the parallel or collaborative relationship between nodes of the same priority. Next, the fault decision-making system selects any fourth target node from each connection priority as a node to be orchestrated. Based on the intermediate orchestration node, it connects the remaining nodes to be orchestrated sequentially according to their connection priority from high to low, constructing the second orchestration thought chain. This chain clarifies the execution order and dependencies between nodes of different priorities. These two chains together constitute the complete orchestration logic framework.

[0087] Continuing with the above embodiments, for branch nodes (intermediate orchestration nodes), the fourth target node with higher priority is the distribution transformer node C, and the one with lower priority is the user terminal node C. The fault decision system first connects two distribution transformer nodes C with the same priority (assuming there are multiple nodes with the same priority) to construct the first orchestration thought link; then, taking one of the distribution transformer nodes C as the node to be orchestrated, it connects the user terminal node C according to priority order to construct the second orchestration thought link, forming the orchestration logic of "distribution transformer node C → branch node → user terminal node C". For substation nodes (intermediate orchestration nodes), distribution transformer nodes A and B with the same priority are connected to construct the first orchestration thought link, and then, taking distribution transformer node A as the node to be orchestrated, it connects the user terminal node A with lower priority in sequence to construct the complete second orchestration thought link.

[0088] Step 2044: Based on the node position of the fourth target node in the preliminary step arrangement result, insert the first arrangement thinking link and the second arrangement thinking link into the preliminary step arrangement result to obtain the initial workflow instance.

[0089] Furthermore, based on the node position of the fourth target node in the preliminary step arrangement results, the fault decision system precisely inserts the first and second arrangement thought links into the preliminary step arrangement results. During this process, the original step order is adjusted according to the order and dependency of the nodes in the link, so that the entire fault handling process integrates the processing steps of cross-branch nodes and same-branch nodes, forming a coherent, orderly whole that conforms to the actual fault handling logic, and finally obtaining the initial workflow instance.

[0090] Continuing with the above embodiments, in the preliminary arrangement results, the nodes on the same branch where user node C and distribution transformer node C are located already have a preliminary arrangement order. The fault decision system inserts the constructed first and second arrangement thought links about branch nodes into it, adjusting it to "branch node status check → distribution transformer node C fault investigation → branch node and user node C connection detection → user node C power supply restoration". For substation node-related links, they are inserted into the preliminary arrangement results involving user nodes A and B and distribution transformer nodes A and B, and the order is adjusted according to priority and dependency, finally forming an initial workflow instance that covers all nodes and steps and has a clear execution order, such as "substation node status detection → distribution transformer node A fault investigation → distribution transformer node B fault investigation → branch node status check → user node A power supply restoration → user node B power supply restoration → user node C power supply restoration → user node D power supply restoration → user node E power supply restoration".

[0091] This invention locates key nodes by traversing dependencies, determines connection priorities based on edge attributes and functions, constructs an orchestration thought chain, and inserts preliminary orchestration results to generate an initial workflow instance. Therefore, it can deeply mine the logical relationships between nodes in the power knowledge graph and construct an execution framework for fault handling steps in a priority-driven manner. By distinguishing between the parallel relationships of nodes with the same priority and the sequential relationships of nodes with different priorities, it ensures the priority execution of critical paths during fault handling and coordinates the collaborative relationships between nodes. This enables the initial workflow instance to accurately match the actual structure of the power system and the fault handling logic, improving the operability of the fault handling solution.

[0092] In one embodiment, steps 301 to 304 include: Step 301: After the processing steps of each initial workflow instance are simulated and executed, the changes in the operating parameters of each system node in the power system are compared with the standard range of the operating parameters of each system node when the power system is running normally, and the comparison results of each system node are obtained.

[0093] Optionally, after simulating the processing steps of each initial workflow instance, the fault decision system collects the operating parameters of each system node in the power system. These operating parameters include key indicators such as voltage, current, power, and frequency. The fault decision system compares the operating parameters of each system node after the simulation with the standard range of operating parameters for that node during normal power system operation. If the node's operating parameters are within the standard range, it indicates that the node is operating normally; if they exceed the standard range, the node is determined to be operating abnormally, thus judging the operating status of the system nodes.

[0094] Continuing with the above embodiments, after the initial workflow instance simulation of transmission line short-circuit fault handling is executed, the fault decision system acquires the operating parameters of each node in the power system. For example, the voltage of the substation node is 330kV after the simulation, while the standard voltage range for this substation node during normal power system operation is 350kV-360kV. Therefore, the voltage parameter of this substation node exceeds the standard range. The current of distribution transformer node C is 800A, while its standard current range during normal operation is 500A-700A. The current parameter of distribution transformer node C also exceeds the standard range. The power of user-end node A is 100kW, which is within its normal operating power standard range of 80kW-120kW, indicating that the power parameter of this node is normal. After comparing the operating parameters of all system nodes, the fault decision system obtains the comparison results for each system node and records whether the operating status of each node is normal.

[0095] Step 302: Based on the comparison results of each system node, identify the abnormal system nodes. Based on the connection relationship of the abnormal system nodes of each initial workflow instance in the power knowledge graph, determine whether there is an abnormal region formed by multiple abnormal system nodes.

[0096] Furthermore, based on the comparison results of each system node, the fault decision system filters out system nodes with abnormal operating status, i.e., abnormal system nodes.

[0097] Furthermore, the fault decision-making system analyzes these abnormal system nodes based on the connection relationships recorded in the power knowledge graph, checking whether multiple abnormal system nodes are interconnected in the power knowledge graph to form a relatively concentrated area. If such an area composed of multiple interconnected abnormal system nodes exists, it is determined to be an abnormal area; if there are no multiple interconnected abnormal system nodes, that is, the abnormal system nodes are relatively scattered, then there is no abnormal area.

[0098] Continuing with the above embodiment, the fault decision system identifies the substation node and distribution transformer node C as anomalous system nodes. A review of the power knowledge graph reveals a connection between the substation node and distribution transformer node C, forming a connected region. Therefore, the fault decision system determines that an anomalous region exists formed by these two anomalous system nodes. Other anomalous system nodes (if they exist) that do not form a connected region with these two nodes or with each other are not included in this anomalous region.

[0099] Step 303: If abnormal regions exist, the abnormal regions of each initial workflow instance are traversed in the power knowledge graph to obtain the boundary nodes between each abnormal region and the normal operating node region. Boundary nodes represent nodes that belong to both abnormal regions and have connections to the normal operating node region.

[0100] Furthermore, when an abnormal region is identified, the fault decision system takes the abnormal region of each initial workflow instance as the object and performs a traversal operation in the power knowledge graph to find nodes that belong to the abnormal region and are connected to the normally operating node region in the power system, i.e., boundary nodes. The boundary nodes serve as the connection hub between the abnormal region and the normally operating node region, and their status is crucial for judging the fault handling effect and the effectiveness of the workflow instance.

[0101] Continuing with the aforementioned abnormal region comprised of the substation node and distribution transformer node C, the fault decision system traversed the power knowledge graph and discovered that, in addition to being connected to the substation node (within the abnormal region), distribution transformer node C is also connected to user-end node C. Furthermore, user-end node C operates normally after the simulation, belonging to the normal operating node region. Therefore, distribution transformer node C is identified as a boundary node. Similarly, the substation node is connected to other transmission line nodes that have not yet shown any abnormalities, and this substation node is also identified as a boundary node. These boundary nodes clearly define the boundary between the abnormal region and the normal operating node region.

[0102] Step 304: Perform recovery validity verification on each initial workflow instance based on the boundary nodes of each initial workflow instance to obtain the target workflow instance.

[0103] Furthermore, the fault decision system performs recovery validity verification on each initial workflow instance based on the boundary nodes of each initial workflow instance to obtain the target workflow instance, as detailed in steps 3041 to 3043.

[0104] This invention, through comparing system node operating parameters, locating abnormal areas, determining boundary nodes, and completing verification, effectively verifies the recovery of the initial workflow instance. Therefore, it can systematically start from the operating status of power system nodes, accurately focus on key aspects of fault handling through the identification of abnormal areas and the location of boundary nodes. Using boundary nodes as the core of verification, it can effectively determine whether the initial workflow instance has truly achieved fault isolation and system recovery, ensuring that the target workflow instance meets the actual recovery needs of the power system, improving the effectiveness of power system fault handling, and guaranteeing the rapid restoration of the power system to normal operation.

[0105] In one embodiment, steps 3041 to 3043 include: Step 3041: Determine the severity of the fault based on the difference between the changes in the operating parameters of the boundary nodes in each initial workflow instance and the standard range of the operating parameters, and determine the sum of the fault severity of each initial workflow instance based on the fault severity of each boundary node.

[0106] Optionally, for each boundary node in the initial workflow instance, the fault decision system calculates the difference between its operating parameter changes and the standard range of operating parameters. These operating parameters include voltage, current, and power, and the difference is measured by the difference between the actual operating parameters and the upper and lower limits of the standard range. Therefore, the fault decision system maps the difference value to a corresponding fault severity level according to a preset fault severity assessment rule; the larger the difference value, the higher the fault severity level. Further, after determining the fault severity of each boundary node, the fault decision system summarizes the fault severity of all boundary nodes in the same initial workflow instance to obtain the total fault severity of that initial workflow instance, reflecting the overall severity of the fault impact on the power system after the execution of that workflow instance.

[0107] Continuing with the initial workflow embodiment for handling short-circuit faults in transmission lines, distribution transformer node C and substation node have been identified as boundary nodes. The simulated current at distribution transformer node C is 800A, while the standard current range during normal operation is 500A-700A, resulting in a current difference of 800-700=100A. According to preset rules, this corresponds to a fault severity level of 3 (assuming a total of 5 levels, with higher levels indicating more severe faults). The simulated voltage at substation node is 330kV, while the standard voltage range during normal operation is 350kV-360kV, resulting in a voltage difference of 350-330=20kV, corresponding to a fault severity level of 2.

[0108] Furthermore, the fault decision system adds the fault severity of these two boundary nodes, resulting in a total fault severity of 3 + 2 = 5 for the initial workflow instance. If other boundary nodes exist, the fault severity is calculated and accumulated in the same way.

[0109] Step 3042: Determine candidate workflow instances based on the sum of the fault severity of each initial workflow instance.

[0110] Furthermore, the fault decision system filters all initial workflow instances based on the sum of fault severity of each initial workflow instance. In one embodiment, a fault severity threshold is set, and the fault decision system selects initial workflow instances whose sum of fault severity is less than or equal to the threshold, determining them as candidate workflow instances. Candidate workflow instances, to a certain extent, indicate that the fault state of the power system can be effectively controlled and have the potential to be further verified as target workflow instances.

[0111] Continuing with the above embodiment, assume the fault severity threshold set by the fault decision system is 8. After calculating the sum of fault severity for multiple initial workflow instances, it is found that the initial workflow instance with a sum of fault severity of 5 meets the threshold condition. Additionally, two other initial workflow instances with sums of fault severity of 6 and 7 respectively also meet the threshold condition. Therefore, the fault decision system identifies these three initial workflow instances as candidate workflow instances, while the initial workflow instances with a sum of fault severity greater than 8 are excluded and not further verified.

[0112] Step 3043: Based on the power knowledge graph, detect the degree of coverage of the power system state changes on the nodes on the fault development path after the execution of each candidate workflow instance, and determine the candidate workflow instance with the degree of change coverage greater than the preset coverage threshold as the target workflow instance.

[0113] Furthermore, the fault decision system, based on the power knowledge graph, detects the state changes of the power system after the execution of each candidate workflow instance. This embodiment of the invention focuses on the changes of nodes on the fault development path and calculates the overlap ratio between the actual changes and the expected changes of nodes on the fault development path after the execution of the candidate workflow instance, using this as the degree of change coverage.

[0114] Furthermore, a coverage threshold is pre-set. The fault decision system identifies candidate workflow instances with a change coverage greater than the threshold as target workflow instances. This indicates that after execution, the workflow instance can better meet the fault handling expectations and effectively cover the key node changes on the fault development path, thereby achieving effective handling and recovery of power system faults.

[0115] Continuing with the three identified candidate workflow instances, the fault decision system detects changes in the power system's state after execution based on the power knowledge graph. For the first candidate workflow instance, there are 10 key nodes in the fault development path, and 8 nodes actually undergo the expected changes after execution, resulting in a change coverage of 8 / 10*100%=80%. For the second candidate workflow instance, there are 8 key nodes in the fault development path, and 6 nodes actually undergo the expected changes, resulting in a change coverage of 6 / 8*100%=75%. For the third candidate workflow instance, there are 12 key nodes in the fault development path, and 10 nodes actually undergo the expected changes, resulting in a change coverage of 10 / 12*100%≈83.3%. Assuming a preset coverage threshold of 70%, the fault decision system identifies all three candidate workflow instances with a change coverage greater than 70% as target workflow instances, effectively handling the fault and promoting the power system's recovery to a normal state.

[0116] This invention provides an in-depth evaluation of the initial workflow instance from two aspects: fault severity and handling effect. First, it eliminates instances with poor handling effect by quantifying the fault severity, thus narrowing the verification scope. Then, it uses the coverage of fault development path node changes as the core standard to ensure that the final target workflow instance can accurately match the fault handling requirements, effectively cover the fault impact range, and achieve targeted handling and efficient recovery of power system faults, thereby ensuring the safe and stable operation of the power system.

[0117] Furthermore, the power thinking chain path decision system provided by the present invention will be described below. The power thinking chain path decision system described below can be referred to in correspondence with the power thinking chain path decision method described above.

[0118] Optional, refer to Figure 2 , Figure 2This is a schematic diagram of the power thinking chain path decision-making system provided by the present invention. The power thinking chain path decision-making system includes: The fault path expansion module 210 is used to perform fault analysis on power system faults, obtain preliminary fault causes, and expand each preliminary fault cause according to the fault handling logic and the fault propagation path in the pre-constructed power knowledge graph based on the workflow engine, so as to obtain the target fault development path. The workflow instance orchestration module 220 is used to orchestrate steps based on the relationship between the processing steps of each target fault development path as a fault handling process node in the workflow, and to obtain an initial workflow instance. The recovery validity verification module 230 is used to perform recovery validity verification on each initial workflow instance based on the power system state change information of the processing steps of each initial workflow instance detected by the power knowledge graph after simulation execution, and obtain the target workflow instance. The workflow instance evaluation module 240 is used to evaluate each target workflow instance based on multi-dimensional indicators, obtain the optimal workflow instance, and determine the optimal workflow instance as the power thinking chain path decision result for power system faults.

[0119] This invention expands each initial fault cause based on a workflow engine, combining fault handling logic with fault propagation paths in a power knowledge graph. Guided by fault handling logic, it leverages the reasoning capabilities of the power knowledge graph to refine the logical expansion of the fault development path for each initial fault cause. This allows for in-depth exploration of potentially complex fault scenarios for each initial fault cause. Furthermore, it arranges steps according to the relationships between the processing steps of each target fault development path, comprehensively streamlining the fault handling approach and streamlining the fault handling process. Even when multiple fault points overlap, new fault types arise, or fault information is missing or contradictory, the fault cause can be accurately analyzed. By simulating the execution of workflow instances, effective workflow instances capable of restoring normal system operation are obtained. Finally, based on multi-dimensional indicators, the optimal workflow instance is selected from the effective workflow instances, determining the best power thinking chain path decision result for handling system faults. This enables rapid and accurate planning of the optimal fault handling decision path under complex fault conditions, ensuring the safe and stable operation of the power system.

[0120] Please see Figure 3 , Figure 3 An embodiment diagram of an electronic device provided in accordance with the present invention. For example... Figure 3As shown, this embodiment of the invention provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor 320. When the processor 320 executes the computer program 311, it performs the following steps: Fault analysis is performed on power system faults to obtain preliminary fault causes. Based on the workflow engine, each preliminary fault cause is expanded according to the fault handling logic and the fault propagation path in the pre-constructed power knowledge graph to obtain the target fault development path. Each target fault development path is used as a fault handling process node in the workflow. The steps are arranged based on the relationship between the processing steps of the fault handling process nodes to obtain an initial workflow instance. Based on the power knowledge graph, the processing steps of each initial workflow instance are detected and the state change information of the power system after simulation execution is used to restore the validity of each initial workflow instance and obtain the target workflow instance. Each target workflow instance is evaluated based on multi-dimensional indicators to obtain the optimal workflow instance, which is then determined as the power thinking chain path decision result for power system faults.

[0121] Please see Figure 4 , Figure 4 An embodiment diagram of a computer-readable storage medium provided in accordance with an embodiment of the present invention is shown. Figure 4 As shown, this embodiment provides a computer-readable storage medium 400 on which a computer program 311 is stored. When the computer program 311 is executed by a processor, it performs the following steps: Fault analysis is performed on power system faults to obtain preliminary fault causes. Based on the workflow engine, each preliminary fault cause is expanded according to the fault handling logic and the fault propagation path in the pre-constructed power knowledge graph to obtain the target fault development path. Each target fault development path is used as a fault handling process node in the workflow. The steps are arranged based on the relationship between the processing steps of the fault handling process nodes to obtain an initial workflow instance. Based on the power knowledge graph, the processing steps of each initial workflow instance are detected and the state change information of the power system after simulation execution is used to restore the validity of each initial workflow instance and obtain the target workflow instance. Each target workflow instance is evaluated based on multi-dimensional indicators to obtain the optimal workflow instance, which is then determined as the power thinking chain path decision result for power system faults.

[0122] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the power thinking chain path decision-making method provided by the above methods, the method including: Fault analysis is performed on power system faults to obtain preliminary fault causes. Based on the workflow engine, each preliminary fault cause is expanded according to the fault handling logic and the fault propagation path in the pre-constructed power knowledge graph to obtain the target fault development path. Each target fault development path is used as a fault handling process node in the workflow. The steps are arranged based on the relationship between the processing steps of the fault handling process nodes to obtain an initial workflow instance. Based on the power knowledge graph, the processing steps of each initial workflow instance are detected and the state change information of the power system after simulation execution is used to restore the validity of each initial workflow instance and obtain the target workflow instance. Each target workflow instance is evaluated based on multi-dimensional indicators to obtain the optimal workflow instance, which is then determined as the power thinking chain path decision result for power system faults.

[0123] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0124] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0125] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A power thinking chain path decision-making method, characterized in that, include: The power system faults are analyzed to obtain preliminary fault causes. Based on the workflow engine, each preliminary fault cause is expanded according to the fault handling logic and the fault propagation path in the pre-constructed power knowledge graph to obtain the target fault development path. Taking each fault handling process node in the target fault development path as an object, the steps are arranged based on the relationship between the processing steps of the fault handling process node to obtain an initial workflow instance. Based on the processing steps of each initial workflow instance, simulate the state change information of the power system after execution, and perform effective recovery verification on each initial workflow instance to obtain the target workflow instance; Each target workflow instance is evaluated based on multi-dimensional indicators to obtain the optimal workflow instance, and the optimal workflow instance is determined as the power thinking chain path decision result for power system faults. The steps involved in determining the target fault development path include: Taking each initial fault cause as the starting node, the fault handling logic based on the workflow engine traverses the first downstream node directly connected to the starting node in the power knowledge graph to obtain the next layer node. Based on the connection relationships between nodes in the power knowledge graph and the fault handling logic of the workflow engine, it is determined whether there is an unexplored second downstream node for each next-level node; if so, the unexplored second downstream node is identified as an indirect layer node. Based on the power knowledge graph, the connection relationship between each indirect layer node and the other nodes except itself is determined, and the first candidate node corresponding to the connection relationship that meets the fault propagation condition is selected according to the fault handling logic of the workflow engine. The fault handling logic based on the workflow engine combines the attribute information of each first candidate node to determine whether there are abnormal nodes that affect the direction of the fault propagation path; if so, the first candidate nodes are updated based on the abnormal nodes to obtain second candidate nodes; wherein, the abnormal nodes represent nodes that cause changes in the fault propagation path. Based on the second candidate node, the indirect layer node, and the next layer node, each initial fault cause is expanded to obtain the target fault development path.

2. The power thinking chain path decision-making method according to claim 1, characterized in that, The process of expanding each initial fault cause based on the second candidate node, indirect layer nodes, and next-layer nodes to obtain the target fault development path includes: Starting with each second candidate node, a reverse traversal is performed in the power knowledge graph to determine whether there is a first target connection edge in the power knowledge graph that directly connects each second candidate node to the next layer node or the indirect layer node; if so, the first target node is obtained by traversing along the first target connection edge in the next layer node or the indirect layer node. Based on the first target node, a reverse traversal is performed in the power knowledge graph to determine whether there is a second target connection edge in the power knowledge graph that directly connects the first target node to the starting node. If there is, the traversal is performed along the second target connection edge to obtain the first starting node, and an initial fault development path is constructed based on the nodes and connection edges traversed from each second candidate node to the first starting node. If there is no connection edge, the traversal is performed in the next layer node or indirect layer node based on the first target node until the second starting node is obtained, and an initial fault development path is constructed based on the nodes and connection edges traversed from each second candidate node to the second starting node. The fault handling logic based on the workflow engine, combined with the functional and interrelationships of nodes on each initial fault development path determined by the power knowledge graph, determines whether there is node convergence or node fork in multiple initial fault development paths. If node convergence occurs, the initial fault development path with node convergence will be merged by the converging node to obtain the target fault development path; If a node fork occurs, the initial fault development path with the forked node will be split into the forked nodes to obtain the target fault development path.

3. The power thinking chain path decision-making method according to claim 1, characterized in that, The process involves taking each fault handling workflow node in the target fault development path as an object, and arranging the steps based on the relationships between the processing steps of the fault handling workflow nodes to obtain an initial workflow instance, including: Based on the positional and connection relationships of each fault handling process node in the power knowledge graph, determine whether there are directly connected fault handling process nodes; if so, identify the directly connected fault handling process nodes as nodes of the same branch. The initial step arrangement is performed on each node in the same branch according to the connection order and dependency relationship between the processing steps of the nodes in the same branch, and the preliminary step arrangement result is obtained. Based on the fact that all fault handling process nodes are traversed in the same branch node, fault handling process nodes that do not belong to any same branch node are identified as cross branch nodes. Based on the initial step arrangement results and the relationship between processing steps of nodes in the same branch and nodes across branches, a secondary step arrangement is performed to obtain the initial workflow instance.

4. The power thinking chain path decision-making method according to claim 3, characterized in that, The initial workflow instance is obtained by combining the preliminary step arrangement results with the relationships between processing steps of nodes within the same branch and nodes across branches, and then performing a secondary step arrangement, including: The power knowledge graph is traversed according to the dependencies between processing steps to determine whether there is a third target node connected to the same branch node in the cross-branch node; if there is a third target node, the fourth target node connected to the same branch node is determined. Based on the edge attributes and edge functions of each third target node and each fourth target node in the power knowledge graph, the connection priority between each third target node and each fourth target node is determined. Using each third target node as an intermediate orchestration node, establish connections between fourth target nodes with the same connection priority to construct the first orchestration thinking link. Then, using any fourth target node with each connection priority as a node to be orchestrated, connect each node to be orchestrated according to the priority order of the connection priority based on the intermediate orchestration node to construct the second orchestration thinking link. Based on the node position of the fourth target node in the preliminary step arrangement result, the first and second arrangement thought links are inserted into the preliminary step arrangement result to obtain the initial workflow instance.

5. The power thinking chain path decision-making method according to claim 1, characterized in that, The process involves simulating the state change information of the power system after the processing steps of each initial workflow instance are executed, and valid recovery verification is performed on each initial workflow instance to obtain the target workflow instance, including: After simulating the execution of the processing steps of each initial workflow instance, the comparison results for each system node are obtained by comparing the changes in the operating parameters of each system node in the power system with the standard range of the operating parameters of each system node during normal operation. Based on the comparison results of each system node, abnormal system nodes are identified. Based on the connection relationship of abnormal system nodes of each initial workflow instance in the power knowledge graph, it is determined whether there is an abnormal region formed by multiple abnormal system nodes. If there are abnormal regions, the abnormal regions of each initial workflow instance are traversed in the power knowledge graph to obtain the boundary nodes between each abnormal region and the normal operating node region; the boundary nodes represent nodes that belong to both the abnormal region and the normal operating node region. The target workflow instance is obtained by performing recovery validity verification on each initial workflow instance based on the boundary node of each initial workflow instance.

6. The power thinking chain path decision-making method according to claim 5, characterized in that, The process of validly restoring each initial workflow instance based on its boundary nodes to obtain the target workflow instance includes: Based on the difference between the changes in the operating parameters of the boundary nodes in each initial workflow instance and the standard range of the operating parameters, the severity of the fault is determined, and the total severity of the fault for each initial workflow instance is determined based on the severity of the fault for each boundary node. Candidate workflow instances are determined based on the sum of the fault severity of each initial workflow instance; Based on the power knowledge graph, the degree of coverage of the power system state changes on the fault development path nodes after the execution of each candidate workflow instance is detected, and the candidate workflow instances with a change coverage degree greater than a preset coverage degree threshold are determined as target workflow instances.

7. A power thinking chain path decision-making system, characterized in that, Applied to the power thinking chain path decision-making method as described in any one of claims 1 to 6; The power thinking chain path decision-making system includes: The fault path expansion module is used to perform fault analysis on power system faults, obtain preliminary fault causes, and expand each preliminary fault cause according to the fault handling logic and the fault propagation path in the pre-constructed power knowledge graph based on the workflow engine, to obtain the target fault development path. The workflow instance orchestration module is used to orchestrate steps based on the relationship between the processing steps of each fault handling process node in the target fault development path, and obtain an initial workflow instance. The recovery validity verification module is used to simulate the state change information of the power system after the processing steps of each initial workflow instance, and to perform recovery validity verification on each initial workflow instance to obtain the target workflow instance. The workflow instance evaluation module is used to evaluate each target workflow instance based on multi-dimensional indicators, obtain the optimal workflow instance, and determine the optimal workflow instance as the power thinking chain path decision result for power system faults. The steps involved in determining the target fault development path include: Taking each initial fault cause as the starting node, the fault handling logic based on the workflow engine traverses the first downstream node directly connected to the starting node in the power knowledge graph to obtain the next layer node. Based on the connection relationships between nodes in the power knowledge graph and the fault handling logic of the workflow engine, it is determined whether there is an unexplored second downstream node for each next-level node; if so, the unexplored second downstream node is identified as an indirect layer node. Based on the power knowledge graph, the connection relationship between each indirect layer node and the other nodes except itself is determined, and the first candidate node corresponding to the connection relationship that meets the fault propagation condition is selected according to the fault handling logic of the workflow engine. The fault handling logic based on the workflow engine combines the attribute information of each first candidate node to determine whether there are abnormal nodes that affect the direction of the fault propagation path; if so, the first candidate nodes are updated based on the abnormal nodes to obtain second candidate nodes; wherein, the abnormal nodes represent nodes that cause changes in the fault propagation path. Based on the second candidate node, the indirect layer node, and the next layer node, each initial fault cause is expanded to obtain the target fault development path.

8. An electronic device, comprising: Memory, used to store computer software programs; A processor for reading and executing the computer software program, characterized in that, when the processor executes the computer software program, it implements the power thinking chain path decision method as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium, wherein a computer software program is stored therein, characterized in that, When the computer software program is executed by the processor, it implements the power thinking chain path decision method as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the power thinking chain path decision method as described in any one of claims 1 to 6.

Images