Fault event positioning method based on topological model tracking analysis

Abstract

The invention relates to the technical field of fault positioning, and is used for solving the problem that a fault event positioning system in the prior art has limitation and cannot adapt to dynamic changes. The invention provides a fault event positioning method based on topological model tracking analysis. The method comprises the following steps: S1, constructing a topological model based on system information and setting an edge weight; s2, tracking a fault propagation path by using time sequence data, and updating a path weight; s3, a fault propagation path is optimized through a shortest path algorithm, the shortest path algorithm adopts an A * algorithm, and a fault source is verified in combination with machine learning; and S4, outputting fault source nodes, related path information and confidence evaluation. The method is suitable for various complex network systems, is high in universality and expansibility, can adapt to the dynamic change of the topological structure and the running state of the system by dynamically updating the edge weight of the topological model, tracking the path and the machine learning model in real time, and always maintains high positioning accuracy and efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of fault location in the whole field, and specifically relates to a method for locating fault events based on topological model tracking and analysis. Background Art

[0002] With the continuous development of modern life, human beings live in various systems, such as communication systems, transportation systems, energy systems and water circulation systems. These systems bring convenience to people's lives, but once a fault occurs in the system, if it cannot be discovered and handled in time, it will bring a series of problems.

[0003] Corresponding processing solutions have also been developed in the prior art. For example, the invention patent "A method for locating fault events based on topological model tracking and analysis" with authorization announcement number CN114184881B records a method for locating fault events based on topological model tracking and analysis, including: obtaining address information provided by the user in the fault report uploaded by the user; graphically displaying the fault location according to the address information; using the topological model to analyze the operation of the power outage equipment and its upstream power supply system, assisting in judging the power outage event, and locating the fault location; the present invention first divides the fault area according to the user's fault report, and according to the number of fault problems occurring in the same area, the dispatching system respectively deploys personnel from the power supply network and the power-consuming equipment, and preliminarily divides the area of ​​the fault problem to avoid comprehensive fault analysis and improve work efficiency. Then, the topological network is used to combine the current and voltage sensors with the actual location for correlation analysis to determine the fault location. The accuracy of fault location is high, which further improves the work efficiency of maintenance personnel and achieves rapid power supply.

[0004] This technical solution is mainly used to solve the serious impact of power system failures on users, emphasizing the importance of quickly and accurately locating faults to restore power supply. It also uses the system's physical connection relationship and operating status information to build a topology model and locate the fault source by analyzing the fault propagation path.

[0005] In addition, the invention patent "Intelligent Distributed Feeder Automation Interoperability Method" with the authorization public number CN105305636B also records Step 1: Establish an interoperability topology model; Step 2, each of the feeder automation terminals checks a local fault signal and exchanges information with an adjacent feeder automation terminal; Step 3: start the feeder automation terminal according to the set timing, and the feeder automation terminal performs fault location, isolation and power supply restoration functions in non-faulty areas according to independent functional logic. In this way, fault location is performed through the topology model.

[0006] However, the technical solutions in the existing patents can only locate faults in the power grid system and cannot be applied to other related technical fields. Secondly, these technical solutions do not take into account the changes in the fault location during the fault location process, cannot adapt to dynamic changes, and cannot truly trace the source. Summary of the invention

[0007] The purpose of the present invention is to solve the problem that the system for locating fault events in the prior art has limitations and cannot adapt to dynamic changes. To this end, the present invention provides a method for locating fault events based on topological model tracking and analysis, including the following steps: S1, topological model construction: based on the physical connection relationship and operating status information of the monitored system, a topological model of the system is constructed; in the topological model, each edge is assigned a weight, and the weight can be dynamically adjusted according to the connection characteristics; S2. Dynamic event tracking: Using the time series data of fault propagation, the fault propagation path is tracked in real time in the topology model. By analyzing the state changes of each node, the cause and effect relationship of fault propagation is extracted; the dynamic update rules of the propagation path weight are defined, and the path weight is dynamically adjusted; S3. Path optimization and source tracing: Apply the shortest path algorithm to optimize the fault propagation path, eliminate noise data and invalid paths, and identify possible fault sources; combine machine learning models to further analyze the topological path and verify the accuracy of the fault source; S4. Positioning result output: Output the most likely fault source node and its related path information, and provide confidence assessment of the positioning result at the same time.

[0008] Preferably, in S1, the monitoring system includes a power grid, a communication network, an industrial control network, a transportation network, a water supply network, a natural gas pipeline network, a computer network, a logistics network, an energy network, an intelligent building system, a rail transit system, and a medical network; the connection characteristics include resistance, communication delay, bandwidth, signal attenuation, reliability, power consumption, flow, load capacity, delay jitter, security, physical distance, and environmental impact; the topological model is represented by a graph structure, in which nodes represent devices or functional units in the system, and edges represent connection relationships between nodes.

[0009] Preferably, in S2, the path weights are dynamically adjusted including signal strength, propagation speed, signal frequency characteristics, signal attenuation coefficient, noise interference level, environmental conditions, propagation medium characteristics, type of fault signal, signal directionality, node response delay to signal, node reliability or health status, and signal timeliness.

[0010] Preferably, in S3, the shortest path algorithm adopts the A* algorithm, and the specific steps of optimizing the fault propagation path are as follows: S301, define the search space of the problem: convert the topological model into a graph structure: represent the topological model of the monitored system as a weighted graph; the nodes of the graph represent the elements in the system, and the edges represent the connection relationship between the nodes; the weight of the edge is dynamically adjusted according to the connection characteristics; set the starting point and end point: according to the fault propagation time series, select the node with the latest state change as the starting point; possible fault source node; S302, define the heuristic function of the A algorithm*: Estimated cost function: The core of the A* algorithm is to combine the actual cost and the heuristic cost, and give priority to the optimal path in the path search; actual cost (g(n)): the cumulative path weight from the starting point to the current node; heuristic cost (h(n)): the estimated cost from the current node to the end point; the heuristic function should be defined according to the fault propagation characteristics, including: network distance: based on the shortest path distance between nodes in the topology model; signal delay: estimate the minimum possible value of the signal delay between nodes; fault propagation probability: estimate the possibility of node failure based on historical data or machine learning models; evaluation function (f(n)): f(n) = g(n) + h(n); give priority to searching for the path with the smallest evaluation value; S303, path search steps of algorithm A*: Initialization: put the starting point node into the open list, set the starting point's g(n)=0, f(n)=h(n); the closed list is empty; loop search: select the node with the smallest f(n) from the open list as the current node; if the current node is the end node, stop searching and return the path; otherwise, move the current node from the open list to the closed list; update neighbor nodes: traverse all neighbor nodes of the current node; if the neighbor node is in the closed list, skip it; if the neighbor node is not in the open list: add it to the open list; calculate the neighbor node's g(n), h(n) and f(n), and record the current node as its parent node; if the neighbor node is already in the open list, check whether the new path is better than the current path: if the new path's g(n) is smaller, update the neighbor node's g(n) and f(n), and set the current node as its parent node; termination condition: when the end node is added to the closed list, it means that the optimal path has been found; if the open list is empty, it means that there is no feasible path; S304, eliminate noise data and invalid paths: Dynamic adjustment of path weights: During the search process, update edge weights in real time according to the dynamic propagation characteristics of fault signals; eliminate paths with excessive weights; Path validity verification: Use machine learning models to screen and verify candidate paths, and eliminate paths that do not match historical fault patterns; Combine causal relationship analysis to eliminate paths that do not conform to the fault propagation law; S305, returning the optimal path: outputting the shortest path from the starting point to the end point, where the nodes included in the path represent the key nodes through which the fault propagates; providing the total cost of the path and the confidence evaluation of the end point node.

[0011] Preferably, in S3, the specific steps of further analyzing the topological path in combination with the machine learning model include the following: S31, data preparation: Before combining with the machine learning model, the fault propagation path and its related data are preprocessed to ensure the integrity and validity of the model input data; S311, data collection: Topological path characteristics: extract candidate paths from the output of the shortest path algorithm, including path length, total path weight, and number of nodes passed; node characteristics: node state change information, including current, voltage, signal strength, and load; historical failure rate or health score of the node; time series characteristics: time series data of fault propagation; edge characteristics: connection characteristics of each edge, including signal attenuation, delay, and reliability; external environment characteristics: the impact of external environmental conditions on signal propagation; S312, data labeling: if there is historical fault data, mark whether the path is related to the actual fault source for supervised learning; in the case of unlabeled data, unsupervised learning is performed through anomaly detection algorithms; S313, data preprocessing: Normalization: normalize the eigenvalues; Denoising: remove irrelevant or poor quality data to reduce the interference of data noise on the model; Feature engineering: reduce the dimension of complex features or construct new features; S32, select a suitable machine learning model: select a suitable machine learning model for path analysis according to data characteristics and analysis objectives; S321, Model selection: Supervised learning model: Logistic regression is used for binary classification problems; Support vector machine is used for classification of high-dimensional data to distinguish fault-related paths from irrelevant paths; Decision tree handles complex nonlinear relationships and provides feature importance analysis; Gradient boosting tree performs well in classification accuracy and speed and is suitable for processing large-scale data; Neural network is used for modeling complex nonlinear feature relationships and requires a large data scale; Unsupervised learning model: K-means clustering performs cluster analysis on paths and identifies abnormal paths; Isolation forest is used to detect abnormal paths that are different from normal propagation path patterns; S322, model training: if there is historical fault data, use supervised learning model for training: input features: path features, node features, time series features, edge features; output label: whether the path is related to the fault source; if there is a lack of label data, use unsupervised learning model: based on input features, learn the distribution pattern of the path and identify abnormal paths; S33, Model application and analysis: Further analysis and optimization of candidate paths using machine learning models; S331, path scoring: Use the trained model to score each candidate path and output the possibility of the path being related to the fault source; based on the scoring results, eliminate the paths that may be noise data; S332, feature importance analysis: the impact of weight changes on path classification, the impact of node state changes on path classification, and the optimization of topology model construction through feature importance analysis; S333, fault source verification: for paths with higher scores, check whether the end nodes in the path conform to the historical fault propagation law; combine historical data and expert knowledge to verify the rationality of the path and fault source; S334, dynamically update the model: collect new fault location data, add new fault samples to the training set, and retrain the model regularly to improve the accuracy and generalization ability of the model; S34, output analysis results: optimal path: output the optimal fault propagation path verified by the machine learning model; fault source node: locate the most likely fault source node; confidence assessment: combine the path score output by the model to provide the confidence of the fault source location result; auxiliary information: possible suboptimal paths for multi-fault source analysis; causal relationship between node state changes and path associations.

[0012] Preferably, in S4, the confidence assessment is that the confidence represents the reliability and accuracy of the current positioning result, usually expressed as a probability value or percentage; by quantifying the impact of different factors on the positioning result, a credibility reference is provided for the final fault source positioning to help decision-making.

[0013] Preferably, the factors affecting the confidence include path correlation score, path weight characteristics, historical data support, fault propagation law verification, multi-path consistency verification and model prediction reliability; according to the fault propagation path analysis, the confidence of the positioning result is evaluated, which specifically includes the following steps: S401, path relevance scoring: scoring candidate paths using a machine learning model. The higher the path score, the higher the confidence level. S402, path weight characteristics: whether the total weight of the path is within a reasonable range; the lower the weight, the more likely the path is the actual propagation path, and the higher the confidence; S403, historical data support: whether the current positioning result matches the historical fault event; if the fault source node and propagation path are consistent with the historical pattern, the confidence is increased; S404, fault propagation law verification: analyze the time series data of fault propagation and check whether the propagation path conforms to the causal relationship; if the path conforms to the propagation law, the confidence level is increased; S405, multi-path consistency verification: among multiple candidate paths, check whether each path points to the same fault source node; if multiple high-scoring paths point to the same fault source node, the confidence is increased; S406, model prediction reliability: Combine the performance indicators of the machine learning model to evaluate the applicability of the model to the current positioning task; if the model performs well in similar scenarios, the confidence level is increased.

[0014] Preferably, each influencing factor is quantified and a confidence score is assigned to each influencing factor. The specific steps include the following: path relevance score: directly take the path relevance probability output by the machine learning model, recorded as P_path; path weight characteristic score: calculate a score based on the total weight of the path: set the path weight range to [min_w, max_w], and the current path weight is w; path weight score S_weight: S_weight = 1 - (w - min_w) / (max_w - min_w); Historical data support score: Based on the frequency of similar failure events in historical data, the support score S_history is calculated: S_history = count_similar_events / total_events; Fault propagation law verification score: Verify the propagation law based on time series data. If the propagation path conforms to the law, the score is 1, otherwise it is 0; recorded as S_causal; Multipath consistency score: If multiple candidate paths point to the same fault source node, the score is 1, otherwise it is 0; denoted as S_consistency; Model prediction reliability score: The model performance index is used as the score, denoted as S_model.

[0015] Preferably, weighted average method: the weights of each factor are set as w1, w2, w3, w4, w5, w6, and the weights are adjusted according to the actual situation; Comprehensive confidence calculation formula: Confidence = w1 * P_path + w2 * S_weight + w3 * S_history + w4 * S_causal + w5 * S_consistency + w6 * S_model; Constraints: w1 + w2 + w3 + w4 + w5 +w6 = 1; Example weight distribution: path relevance score: w1 = 0.3, path weight characteristic score: w2 = 0.2, historical data support score: w3 = 0.2, propagation law verification score: w4 = 0.1, multi-path consistency score: w5 = 0.1, model prediction reliability score: w6 = 0.1.

[0016] Preferably, the specific steps of confidence verification and output include the following: verify the confidence result: verify the calculated confidence to ensure that the confidence value is associated with the actual positioning accuracy; output the confidence result: output the positioning result, including the most likely fault source node; related propagation path information; the confidence of the positioning result; maintenance personnel go to the location of the positioning result for maintenance and verify the accuracy of the positioning result at the same time; if the positioned position is far apart, relocate the position and adjust the relevant coefficients in the model according to the coordinate changes of the position correction; output the key factors affecting the confidence: provide suboptimal paths and their confidence for multi-fault source or complex scenario analysis.

[0017] The beneficial effects achieved by the present invention are as follows: compared with the prior art, the present invention improves the accuracy of fault location by constructing a dynamic topology model, constructing a topology model based on the physical connection relationship and operation status information of various monitored networks such as power grids and communication networks, and dynamically adjusting the edge weights, making use of the characteristics of different networks for adjustment, and using the fault propagation time series data to track the path in real time, analyze the node state changes and causal relationships, dynamically update the path weights, and then combine the A algorithm to optimize the path, the machine learning model analysis and verification, effectively eliminate interference data, and make the fault location accuracy of power systems, industrial control networks, water supply networks, etc. reach more than 90%, more than 95%, and more than 92%, respectively. The above is the result of the above analysis. In terms of improving the efficiency of fault location, the dynamic update and optimization of the topology model reduces redundant calculations. For example, algorithm A reasonably sets the heuristic function, comprehensively considers the actual and estimated costs, and prioritizes the search for the optimal path to avoid invalid calculations. It can quickly lock the location of the fault source through the combination of dynamic tracking, path optimization, and machine learning models. For example, in the event of network failures such as power, communication, and industrial control, it can locate and restore system operation in a timely manner. In terms of enhancing system adaptability, this method is suitable for a variety of complex network systems, with strong versatility and scalability. It can also adapt to the dynamic changes of the system topology and operating status by dynamically updating the edge weights of the topology model, real-time path tracking, and machine learning models, and always maintain high positioning accuracy and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Other features, objects and advantages of the present invention will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings. The drawings are only for the purpose of illustrating preferred embodiments and are not to be considered as limiting the present invention. Also, the same reference symbols are used throughout the drawings to represent the same parts.

[0019] Figure 1 It is a schematic diagram of the overall process of the method for locating fault events based on topology model tracing and analysis provided by an embodiment of the present invention.

[0020] Figure 2 This is a specific flow chart of the method for locating fault events based on topology model tracing and analysis in step S3 provided by an embodiment of the present invention, in which the shortest path algorithm uses the A* algorithm to optimize the fault propagation path.

[0021] Figure 3 It is a specific flowchart of the method for locating fault events based on topology model tracing and analysis in step S3 provided by an embodiment of the present invention, which further analyzes the topology path in combination with a machine learning model.

[0022] Figure 4 It is a specific flow chart for evaluating the confidence of the positioning result in step S4 of the fault event positioning method based on topology model tracing and analysis provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific implementation method described herein is only an optimal embodiment of the present invention, which is only used to explain the present invention and does not limit the scope of protection of the present invention. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0024] Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flow charts. Although the flow charts describe the operations (or steps) as sequential processes, many of the operations (or steps) therein can be implemented in parallel, concurrently, or simultaneously. In addition, the order of the operations can be rearranged. A process can be terminated when its operations are completed, but can also have additional steps not included in the accompanying drawings; a process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0025] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. It should also be understood that in various embodiments of the present invention, the size of the sequence number of each process does not mean the order of execution, and the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0026] It should be understood that, in the present invention, "multiple" means two or more. "And / or" is only a variable relationship that describes associated objects, indicating that three relationships may exist. It should be understood that, in the present invention, "B corresponding to A", "B corresponding to A", "A corresponds to B" or "B corresponds to A" means that B is associated with A, and B can be determined based on A. Determining B based on A does not mean determining B based only on A, but B can also be determined based on A and / or other information. The match between A and B means that the similarity between A and B is greater than or equal to a preset threshold.

[0027] The present invention relates to the A* algorithm, which is a heuristic search algorithm for finding the shortest path in optimization problems. It combines two key factors: actual cost and heuristic evaluation, aiming to find the most efficient path from a starting point to a target. In the present invention, the applicant optimizes the A* algorithm based on application scenarios.

[0028] See also Figure 1 The embodiment of the present invention provides a technical solution: a method for locating fault events based on topology model tracking and analysis, the method steps include the following: S1. Topology model construction: Based on the physical connection relationship and operation status information of the monitored system, a topological model of the system is constructed; In the topological model, each edge is given a weight, which can be adjusted dynamically according to the connection characteristics; S2. Dynamic event tracking: Using the time series data of fault propagation, the fault propagation path is tracked in real time in the topology model, and the cause and effect relationship of fault propagation is extracted by analyzing the state changes of each node; Define dynamic update rules for propagation path weights and dynamically adjust path weights; S3, Path Optimization and Tracing: Apply the shortest path algorithm to optimize the fault propagation path, eliminate noise data and invalid paths, and locate possible fault sources; Combine machine learning models to further analyze topology paths and verify the accuracy of fault sources; S4. Positioning result output: The most likely fault source node and its related path information are output, and the confidence assessment of the positioning result is provided.

[0029] Further improved, in step S1, the monitoring system includes a power grid, a communication network, an industrial control network, a transportation network, a water supply network, a natural gas pipeline network, a computer network, a logistics network, an energy network, an intelligent building system, a rail transit system, and a medical network; transportation network: a road network, a signal light system, etc. in an urban transportation system; water supply network: a tap water network system, including a water pumping station, pipelines, and valves; a natural gas pipeline network: a pipeline system for transporting natural gas or oil; a computer network: a data center network or a cloud computing network, including servers, switches, and routers; a logistics network: a connection system for storage and transportation nodes; an energy network: a distributed network of new energy such as wind power, photovoltaic power generation, and energy storage systems; an intelligent building system: a building automation network, including connections between sensors, controllers, and equipment; a rail transit system: including connections between trains, tracks, and signal control systems; a medical network: a networking system between medical equipment and sensors.

[0030] Connection characteristics include resistance, communication delay, bandwidth, signal attenuation, reliability, power consumption, flow, load capacity, delay jitter, security, physical distance, and environmental impact; bandwidth: the capacity of the communication link, which affects the data transmission rate; signal attenuation: the reduction in the strength of the signal during propagation; reliability: the stability of the connection, including the drop rate or failure frequency; power consumption: the energy consumption characteristics of nodes and connections; flow: the actual data or material flow (network flow, liquid flow) through the connection; load capacity: the maximum load that the connection can withstand, including current capacity or mechanical stress; delay jitter: the volatility of the delay, which is especially important in communication networks; security: the connection's anti-attack ability or the encryption of data transmission; physical distance: the actual physical distance between nodes, which may affect the propagation time; environmental impact: including the impact of temperature and humidity on connection performance. The scope of application of the topology model can be further expanded, and the accuracy and practicality of the fault event location method can be improved.

[0031] The topology model is represented by a graph structure, where nodes represent devices or functional units in the system and edges represent the connection relationships between nodes.

[0032] Further improved, in step S2, the path weights are dynamically adjusted including signal strength, propagation speed, signal frequency characteristics, signal attenuation coefficient, noise interference level, environmental conditions, propagation medium characteristics, type of fault signal, signal directionality, node response delay to signal, node reliability or health status, and signal timeliness; Frequency characteristics of the signal: Fault signals exhibit different propagation characteristics in different frequency ranges, including that the high-frequency components of some signals may decay faster, while the low-frequency components may propagate farther; Signal attenuation coefficient: The degree of attenuation of a signal due to distance or medium characteristics during propagation. Different paths may have different attenuation coefficients due to differences in media or environment. Noise interference level: the background noise or interference signal strength that may exist on the path, affecting the transmission quality and reliability of the signal; Environmental conditions: Environmental factors including temperature, humidity, and electromagnetic interference affect the propagation speed and strength of signals; Propagation medium characteristics: Path weights can be dynamically adjusted based on the characteristics of the propagation medium (including the resistance, capacitance, inductance of the conductor, or the medium type of the communication link); Type of fault signal: Different types of fault signals (including short circuit, overload, and disconnection) have different propagation modes and characteristics, which affect the calculation of path weights; Signal directionality: Fault signals have a specific propagation directionality, including that in some networks, signals preferentially propagate along low-impedance paths; Node response delay to signals: The detection or response time of nodes in the topology to fault signals may be inconsistent. This delay characteristic can affect the dynamic adjustment of weights. Node reliability or health status: The reliability of the node itself may affect the propagation of the signal, including that the propagation signal may be weak or delayed when the node is in a sub-healthy state; Signal timeliness: The timeliness of fault signals, i.e. the time sequence of signal arrival, is used to dynamically adjust weights to reflect the actual propagation of the path; The accuracy of dynamic adjustment of path weights is improved, thereby improving the performance and reliability of the fault location method.

[0033] Further improvement, such as Figure 2 As shown, in step S3, the shortest path algorithm adopts the A* algorithm, and the specific steps for optimizing the fault propagation path are as follows: S301, define the search space of the problem: Convert the topological model into a graph structure: The topological model of the monitored system is represented as a weighted graph; the nodes of the graph represent the elements (devices or subnets) in the system, and the edges represent the connection relationship between the nodes; the weights of the edges are dynamically adjusted according to the connection characteristics (signal delay, propagation strength, reliability); Set the starting point and end point: According to the fault propagation time series, select the node with the latest state change as the starting point; the possible fault source node; S302, define the heuristic function of algorithm A*: Estimated cost function: The core of the A* algorithm is to combine the actual cost and the heuristic cost, and give priority to the optimal path in the path search; actual cost (g(n)): the cumulative path weight from the starting point to the current node; heuristic cost (h(n)): the estimated cost from the current node to the end point; the heuristic function should be defined according to the fault propagation characteristics, including: network distance: based on the shortest path distance between nodes in the topology model; signal delay: estimate the minimum possible value of the signal delay between nodes; fault propagation probability: estimate the probability of node failure based on historical data or machine learning models; Evaluation function (f(n)): f(n) = g(n) + h(n); Prioritize the search for the path with the smallest evaluation value; S303, path search step of algorithm A*: Initialization: put the starting point node into the open list, set g(n)=0, f(n)=h(n) of the starting point; the closed list is empty; loop search: select the node with the smallest f(n) from the open list as the current node; if the current node is the end node, stop searching and return the path; otherwise, move the current node from the open list to the closed list; Update neighbor nodes: traverse all neighbor nodes of the current node; if the neighbor node is in the closed list, skip it; if the neighbor node is not in the open list: add it to the open list; calculate g(n), h(n) and f(n) of the neighbor node, and record the current node as its parent node; if the neighbor node is already in the open list, check whether the new path is better than the current path: if the g(n) of the new path is smaller, update the g(n) and f(n) of the neighbor node, and set the current node as its parent node; termination condition: when the end node is added to the closed list, it means that the optimal path has been found; if the open list is empty, it means that there is no feasible path; S304, remove noise data and invalid paths: Dynamic adjustment of path weights: During the search process, edge weights are updated in real time according to the dynamic propagation characteristics of fault signals; paths with excessively large weights are eliminated; Path validity verification: Use machine learning models to screen and verify candidate paths, and eliminate paths that do not match historical failure patterns; combine causal relationship analysis to eliminate paths that do not conform to the laws of fault propagation; S305, return the optimal path: Output the shortest path from the starting point to the end point. The nodes included in the path represent the key nodes through which the fault propagates. Provides the total cost of the path (propagation time or signal loss) and a confidence estimate of the endpoint node (fault source).

[0034] The key points of optimization include: Dynamic weight update: adjust edge weights based on real-time fault propagation data to improve path search accuracy; Heuristic function design: The heuristic function must be combined with the specific characteristics of the topological model to ensure the rationality of the estimated cost; Combined with machine learning analysis: verify the optimized path and further improve the accuracy of fault source location.

[0035] Through the above steps, the A* algorithm can efficiently search for fault propagation paths, eliminate interference data, and lock possible fault source nodes.

[0036] Further improvement, such as Figure 3 As shown, in step S3, the specific steps of further analyzing the topological path in combination with the machine learning model include the following: S31, data preparation: Before combining with the machine learning model, the fault propagation path and its related data are preprocessed to ensure the integrity and validity of the model input data; S311, Data Collection: Topological path features: extract candidate paths from the output of the shortest path algorithm, including path length, total path weight, and number of nodes passed; node features: node state change information, including current, voltage, signal strength, and load; historical failure rate or health score of the node; time series features: time series data of fault propagation (arrival time and propagation delay of fault signals); edge features: connection characteristics of each edge, including signal attenuation, delay, and reliability; external environment features: the impact of external environmental conditions (temperature, humidity, and electromagnetic interference) on signal propagation; S312, data label: If there is historical fault data, mark whether the path is related to the actual fault source (1 means related, 0 means irrelevant) for supervised learning; in the case of unlabeled data, unsupervised learning is performed through anomaly detection algorithms (cluster analysis); S313, data preprocessing: Normalization: normalize the feature values ​​(sum of path weights, time series data); Denoising: remove irrelevant or poor quality data to reduce the interference of data noise on the model; Feature Engineering: reduce the dimension of complex features (principal component analysis) or construct new features (weight change rate, node failure probability); S32, select a suitable machine learning model: Select appropriate machine learning models for path analysis based on data characteristics and analysis objectives; S321, Model selection: Supervised learning model: Logistic regression is used for binary classification problems (whether the path is related to the fault source); support vector machine (SVM) is used to classify high-dimensional data and distinguish between fault-related paths and irrelevant paths; decision trees handle complex nonlinear relationships and provide feature importance analysis; gradient boosting trees (XGBoost, LightGBM) perform well in classification accuracy and speed and are suitable for processing large-scale data; neural networks are used to model complex nonlinear feature relationships and require a large data scale; Unsupervised Learning Models: K-means clustering performs cluster analysis on the paths and identifies abnormal paths; Isolation Forest is used to detect abnormal paths that are different from the normal propagation path pattern; S322, model training: If there is historical fault data, use the supervised learning model for training: Input features: path features, node features, time series features, edge features; Output label: whether the path is related to the fault source; If there is a lack of labeled data, use unsupervised learning models: Based on the input features, the distribution pattern of the learning path is used to identify abnormal paths; S33, Model application and analysis: Combine machine learning models to further analyze and optimize candidate paths; S331, Path Scoring: Use the trained model to score each candidate path and output the possibility (confidence) of the path being associated with the fault source; based on the scoring results, eliminate paths that may be noise data (low-scoring paths).

[0037] S332, Feature Importance Analysis: The impact of weight changes on path classification and node status changes on path classification, and the construction of topology model is optimized through feature importance analysis; S333, fault source verification: For paths with higher scores (possible fault propagation paths), check whether the end nodes in the paths conform to the historical fault propagation rules; combine historical data and expert knowledge to verify the rationality of the paths and fault sources; S334, Dynamic Update Model: Collect new fault location data, add new fault samples to the training set, and retrain the model regularly to improve the accuracy and generalization ability of the model; S34, output analysis results: Optimal path: Outputs the optimal fault propagation path verified by the machine learning model; Fault source node: locate the most likely fault source node; Confidence evaluation: Combined with the path score output by the model, it provides the confidence level of the fault source location result; Auxiliary information: possible suboptimal paths for multi-fault source analysis; causal relationship between node state changes and path associations.

[0038] Through the above steps, combined with the machine learning model, noise data and invalid paths can be effectively eliminated, the accuracy of the fault source can be further verified, and the reliability and accuracy of the fault location method can be improved.

[0039] Further improved, in step S4, the confidence assessment is that the confidence represents the reliability and accuracy of the current positioning result, usually expressed as a probability value (such as a score between 0 and 1) or a percentage; by quantifying the impact of different factors on the positioning result, a credibility reference is provided for the final fault source positioning to help decision-making.

[0040] Further improvement, such as Figure 4 As shown, the factors affecting the confidence include path correlation score, path weight characteristics, historical data support, fault propagation law verification, multi-path consistency verification and model prediction reliability; According to the fault propagation path analysis, the confidence level of the location result evaluation includes the following steps: S401, path relevance scoring: scoring candidate paths using a machine learning model. The higher the path score, the higher the confidence level. S402, path weight characteristics: whether the total weight of the path (propagation delay, signal attenuation) is within a reasonable range; the lower the weight (indicating a lower propagation cost), the more likely the path is the actual propagation path, and the higher the confidence level; S403, historical data support: whether the current positioning result matches the historical fault event; if the fault source node and propagation path are consistent with the historical pattern, the confidence is increased; S404, fault propagation law verification: analyze the time series data of fault propagation and check whether the propagation path conforms to the causal relationship (signal propagation direction, time sequence); if the path conforms to the propagation law, the confidence level is increased; S405, multi-path consistency verification: among multiple candidate paths, check whether each path points to the same fault source node; if multiple high-scoring paths point to the same fault source node, the confidence is increased; S406, model prediction reliability: Combine the performance indicators (accuracy, precision, recall) of the machine learning model to evaluate the applicability of the model to the current positioning task; if the model performs well in similar scenarios, the confidence level is increased.

[0041] For further improvement, each influencing factor is quantified and a confidence score is assigned to each influencing factor. The specific steps include the following: Path relevance score: directly take the path relevance probability output by the machine learning model, denoted as P_path; Path weight feature scoring: Calculate a score based on the total weight of the path: Assume the path weight range is [min_w, max_w], and the current path weight is w; Path weight score S_weight: S_weight = 1 - (w - min_w) / (max_w - min_w); (The closer the weight is to the minimum value, the higher the score); Historical data supports the rating: Calculate the support score S_history based on the frequency of similar failure events in historical data: S_history = count_similar_events / total_events; (The higher the proportion of similar events, the higher the score); Fault propagation law verification score: Verify the propagation law based on time series data. If the propagation path conforms to the law, the score is 1, otherwise it is 0; recorded as S_causal; Multipath consistency score: If multiple candidate paths (e.g., the top 3 paths) point to the same fault source node, the score is 1, otherwise it is 0; denoted as S_consistency; Model prediction reliability score: The model performance indicator (historical accuracy) is used as the score, denoted as S_model.

[0042] Further improvement, the above scores are combined to calculate the overall confidence, and the weighted average method is used, which specifically includes the following steps: Weighted average method: Assume that the weights of each factor are w1, w2, w3, w4, w5, w6 respectively. The weights can be adjusted according to the actual situation. Comprehensive confidence calculation formula: Confidence = w1 * P_path + w2 * S_weight + w3 * S_history + w4 * S_causal + w5 * S_consistency + w6 * S_model; Constraints: w1 + w2 + w3 + w4 + w5 + w6 = 1; Example weight assignment: path relevance score: w1 = 0.3, path weight characteristic score: w2 = 0.2, historical data support score: w3 = 0.2, propagation law verification score: w4 = 0.1, multi-path consistency score: w5 = 0.1, model prediction reliability score: w6 = 0.1.

[0043] Specifically, the steps of confidence verification and output are as follows: Verify the confidence results: Verify the calculated confidence level to ensure that it is correlated with the actual positioning accuracy; Verify through historical data whether the positioning accuracy is significantly improved when the confidence level is higher than a certain threshold (such as 0.8); Output confidence result: Output positioning result, including: the most likely fault source node; related propagation path information; confidence of positioning result (Confidence = 0.85 or 85%); Provide auxiliary information: output key factors affecting confidence (path score, historical data support); provide suboptimal paths and their confidence for analysis of multiple fault sources or complex scenarios.

[0044] Through the above steps, confidence assessment can provide clear credibility quantification for the positioning results, assist decision-making and improve the reliability and practicality of the fault event location method.

[0045] Example 1: Fault location in power system Topology model construction: obtain real-time operation data of the power system, including the connection relationship and operation status of substations, lines, switches and other equipment; construct a topology model of the power network, where nodes represent substations and switchgear, edges represent power lines, and edge weights are dynamically adjusted according to line impedance and load conditions; Dynamic event tracking: When a short circuit occurs somewhere in the power system, the time series data of the fault signal is collected; in the topology model, the fault signal is traced layer by layer from the starting point of propagation and the propagation path is recorded; Path optimization and source tracing: Apply the shortest path algorithm to optimize the propagation path and eliminate interference paths caused by noise; combine the machine learning model trained with historical fault data to verify the path results and locate the fault source; Result output: Output the fault source node and its propagation path, with a positioning accuracy of more than 90%.

[0046] Example 2: Fault location in a communication network Topology model construction: obtain the topology information of the communication network, including the connection relationship and traffic data of routers, switches, terminal devices, etc.; build a dynamic topology model of the communication network, and dynamically adjust the edge weights according to the link delay, packet loss rate, etc.; Dynamic event tracking: When a node in the network fails, collect the propagation path data of its fault signal; track the signal propagation path in the topology model and extract key nodes; Path optimization and source tracing: Utilize the time series characteristics of fault propagation to eliminate abnormal data and optimize the propagation path; combine deep learning models to intelligently identify and verify the source of the fault; Result output: Output the most likely fault node and provide visualization results of path optimization.

[0047] Example 3: Fault location in industrial control network Topology model construction: obtain real-time data of industrial control systems, including the connection relationship and status information between PLCs (programmable logic controllers), sensors and actuators; build a topology model of the industrial control network, where nodes represent control devices, edges represent signal transmission channels, and edge weights are dynamically adjusted according to characteristics such as latency and bandwidth; Dynamic event tracking: When a control device fails, collect the time series data of the fault signal; in the topology model, trace the signal propagation path layer by layer starting from the fault node, and record the state changes and key node information; Path optimization and source tracing: Apply the shortest path algorithm to optimize the fault propagation path and eliminate invalid paths caused by environmental interference or data anomalies; verify the path with the machine learning model trained with previous fault history data to determine the source of the fault; Positioning result output: Output the most likely fault source node and its propagation path, and provide confidence assessment of the fault location result to ensure that the positioning accuracy reaches more than 95%.

[0048] Example 4: Fault location in a traffic network Topological model construction: obtain real-time data of the traffic network, including the connection relationship and flow information of roads, intersections, traffic lights, etc.; construct a topological model of the traffic network, where nodes represent intersections, edges represent roads, and edge weights are dynamically adjusted according to traffic flow, road conditions, and driving speed; Dynamic event tracking: When an accident or signal failure occurs in the traffic network, relevant time series data is collected; in the topological model, the affected area is tracked from the point where the failure occurred, and traffic flow changes and key nodes are recorded; Path optimization and tracing: Use the shortest path algorithm to optimize traffic flow in the affected area and eliminate congested paths caused by accidents; combine deep learning models to intelligently identify and verify fault sources; Positioning result output: Output the most likely fault node and its impact path, provide visualization results of traffic flow optimization, and ensure that the accuracy of the decision support system reaches more than 90%.

[0049] Example 5: Fault location in a water supply network Topological model construction: obtain real-time monitoring data of the water supply system, including the connection relationship and water flow status of water sources, pipes, valves and water meters; construct a topological model of the water supply network, where nodes represent water sources and pipe intersections, edges represent water flow paths, and edge weights are dynamically adjusted according to pipe diameter, water pressure and flow rate; Dynamic event tracking: When water leakage or abnormal water pressure occurs in the water supply system, the time series data of the fault signal is collected; in the topological model, the changes in water flow are tracked, expanding outward from the fault point, and the affected nodes and paths are recorded; Path optimization and source tracing: Apply the shortest path algorithm to optimize the water flow propagation path and eliminate interference paths caused by noise or abnormal water flow; combine historical fault data and machine learning models to verify the source of the fault; Positioning result output: Output the most likely fault source node and its related path information, and provide confidence assessment of the positioning result to ensure that the positioning accuracy reaches more than 92%.

[0050] Summary,By building a dynamically updated system topology model,combining the timing characteristics and network characteristics of the fault propagation path,,the fault source can be quickly located and the fault event can be accurately located.,It can be widely used in complex networks such as power systems,,communication networks, industrial control systems, etc.,to achieve fast and accurate location of fault events; By tracking the fault propagation path through the dynamic topology model, the fault source can be accurately located in a complex network, significantly improving the positioning accuracy; the dynamic update and optimization of the topology model reduces redundant calculations and improves the real-time performance of fault location; combining the dynamic topology model with the fault propagation path optimization algorithm realizes the adaptive tracking and optimization of the fault propagation path, which can effectively filter out noise interference.

[0051] The above specific implementation is a preferred implementation of the method for locating fault events based on topological model tracking and analysis of the present invention, and is not intended to limit the specific implementation scope of the present invention. The scope of the present invention includes but is not limited to the present specific implementation. All equivalent changes made in accordance with the shape and structure of the present invention are within the protection scope of the present invention. A person of ordinary skill in the art can understand that the implementation of all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of various embodiments of the low-voltage power quality analysis and processing method based on measured data. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

Claims

1. A method for locating fault events based on topological model tracking and analysis, used in a system with a specific physical structure, characterized in that: The method includes the following steps: S1, topology model construction: based on the physical connection relationship and operation status information of the monitored system, a topology model of the system is constructed; in the topology model, each edge is given a weight, and the weight can be dynamically adjusted according to the connection characteristics; S2. Dynamic event tracking: Using the time series data of fault propagation, the fault propagation path is tracked in real time in the topology model, and the cause and effect relationship of fault propagation is extracted by analyzing the state changes of each node; Define dynamic update rules for propagation path weights and dynamically adjust path weights; S3. Path optimization and source tracing: Apply the shortest path algorithm to optimize the fault propagation path, eliminate noise data and invalid paths, and identify possible fault sources; combine machine learning models to further analyze the topological path and verify the accuracy of the fault source; S4. Positioning result output: Output the most likely fault source node and its related path information, and provide confidence assessment of the positioning result at the same time.

2. According to the method for locating fault events based on topology model tracking and analysis according to claim 1, it is characterized in that: In S1, the monitoring systems include power grids, communication networks, industrial control networks, transportation networks, water supply networks, natural gas pipeline networks, computer networks, logistics networks, energy networks, intelligent building systems, rail transit systems, and medical networks; connection characteristics include resistance, communication delay, bandwidth, signal attenuation, reliability, power consumption, flow, load capacity, delay jitter, security, physical distance, and environmental impact; the topological model is represented by a graph structure, with nodes representing devices or functional units in the system and edges representing the connection relationships between nodes.

3. According to the method for locating fault events based on topology model tracing and analysis according to claim 1, it is characterized in that: In S2, the path weights that are dynamically adjusted include signal strength, propagation speed, signal frequency characteristics, signal attenuation coefficient, noise interference level, environmental conditions, propagation medium characteristics, type of fault signal, signal directionality, node response delay to signal, node reliability or health status, and signal timeliness.

4. According to the method for locating fault events based on topology model tracking and analysis according to claim 1, it is characterized in that: In S3, the shortest path algorithm uses the A* algorithm. The specific steps for optimizing the fault propagation path are as follows: S301, define the search space of the problem: convert the topological model into a graph structure: represent the topological model of the monitored system as a weighted graph; the nodes of the graph represent the elements in the system, and the edges represent the connection relationship between the nodes; the weight of the edge is dynamically adjusted according to the connection characteristics; set the starting point and end point: according to the fault propagation time series, select the node with the latest state change as the starting point; possible fault source node; S302, define the heuristic function of the A algorithm*: Estimated cost function: The core of the A* algorithm is to combine the actual cost and the heuristic cost, and give priority to the optimal path in the path search; actual cost (g(n)): the cumulative path weight from the starting point to the current node; heuristic cost (h(n)): the estimated cost from the current node to the end point; The heuristic function should be defined based on the fault propagation characteristics, including: network distance: based on the shortest path distance between nodes in the topology model; signal delay: estimate the minimum possible value of the signal delay between nodes; fault propagation probability: estimate the probability of node failure based on historical data or machine learning models; evaluation function (f(n)): f(n) = g(n) + h(n); prioritize the search for the path with the minimum evaluation value; S303, path search steps of algorithm A*: Initialization: put the starting point node into the open list, set the starting point's g(n)=0, f(n)=h(n); the closed list is empty; loop search: select the node with the smallest f(n) from the open list as the current node; if the current node is the end node, stop searching and return the path; otherwise, move the current node from the open list to the closed list; update neighbor nodes: traverse all neighbor nodes of the current node; if the neighbor node is in the closed list, skip it; if the neighbor node is not in the open list: add it to the open list; calculate the neighbor node's g(n), h(n) and f(n), and record the current node as its parent node; if the neighbor node is already in the open list, check whether the new path is better than the current path: if the new path's g(n) is smaller, update the neighbor node's g(n) and f(n), and set the current node as its parent node; termination condition: when the end node is added to the closed list, it means that the optimal path has been found; if the open list is empty, it means that there is no feasible path; S304, eliminate noise data and invalid paths: Dynamic adjustment of path weights: During the search process, update edge weights in real time according to the dynamic propagation characteristics of fault signals; eliminate paths with excessive weights; Path validity verification: Use machine learning models to screen and verify candidate paths, and eliminate paths that do not match historical fault patterns; Combine causal relationship analysis to eliminate paths that do not conform to the fault propagation law; S305, returning the optimal path: outputting the shortest path from the starting point to the end point, where the nodes included in the path represent the key nodes through which the fault propagates; providing the total cost of the path and the confidence evaluation of the end point node.

5. The method for locating fault events based on topology model tracing and analysis according to claim 1 is characterized in that: In S3, the specific steps of further analyzing the topological path in combination with the machine learning model include the following: S31, data preparation: Before combining with the machine learning model, the fault propagation path and its related data are preprocessed to ensure the integrity and validity of the model input data; S311, Data Collection: Topological Path Features: Extract candidate paths from the output of the shortest path algorithm, including path length, total path weights, and number of nodes passed; Node characteristics: information about node status changes, including current, voltage, signal strength, and load; historical failure rate or health score of the node; time series characteristics: time series data of fault propagation; edge characteristics: connection characteristics of each edge, including signal attenuation, delay, and reliability; external environment characteristics: the impact of external environmental conditions on signal propagation; S312, data labeling: if there is historical fault data, mark whether the path is related to the actual fault source for supervised learning; in the case of unlabeled data, unsupervised learning is performed through anomaly detection algorithms; S313, data preprocessing: Normalization: normalize the eigenvalues; Denoising: remove irrelevant or poor quality data to reduce the interference of data noise on the model; Feature engineering: reduce the dimension of complex features or construct new features; S32, select a suitable machine learning model: select a suitable machine learning model for path analysis according to data characteristics and analysis objectives; S321, Model selection: Supervised learning model: Logistic regression is used for binary classification problems; Support vector machine is used for classification of high-dimensional data to distinguish fault-related paths from irrelevant paths; Decision tree handles complex nonlinear relationships and provides feature importance analysis; Gradient boosting tree performs well in classification accuracy and speed and is suitable for processing large-scale data; Neural network is used for modeling complex nonlinear feature relationships and requires a large data scale; Unsupervised learning model: K-means clustering performs cluster analysis on paths and identifies abnormal paths; Isolation forest is used to detect abnormal paths that are different from normal propagation path patterns; S322, model training: if there is historical fault data, use supervised learning model for training: input features: path features, node features, time series features, edge features; output label: whether the path is related to the fault source; if there is a lack of label data, use unsupervised learning model: based on input features, learn the distribution pattern of the path and identify abnormal paths; S33, Model application and analysis: Further analysis and optimization of candidate paths using machine learning models; S331, path scoring: Use the trained model to score each candidate path and output the possibility of the path being related to the fault source; based on the scoring results, eliminate the paths that may be noise data; S332, feature importance analysis: the impact of weight changes on path classification, the impact of node status changes on path classification, and the optimization of topology model construction through feature importance analysis; S333, fault source verification: for paths with higher scores, check whether the end nodes in the path conform to the historical fault propagation law; combine historical data and expert knowledge to verify the rationality of the path and fault source; S334, dynamically update the model: collect new fault location data, add new fault samples to the training set, and retrain the model regularly to improve the accuracy and generalization ability of the model; S34, output analysis results: optimal path: output the optimal fault propagation path verified by the machine learning model; fault source node: locate the most likely fault source node; confidence assessment: combine the path score output by the model to provide the confidence of the fault source location result; auxiliary information: possible suboptimal paths for multi-fault source analysis; causal relationship between node state changes and path associations.

6. The method for locating fault events based on topology model tracing and analysis according to claim 1 is characterized in that: In S4, confidence assessment is the confidence that represents the reliability and accuracy of the current positioning result, usually expressed as a probability value or percentage; by quantifying the impact of different factors on the positioning result, it provides a credibility reference for the final fault source positioning and helps decision-making.

7. The method for locating fault events based on topology model tracing and analysis according to claim 6 is characterized in that: The factors affecting the confidence include path correlation score, path weight characteristics, historical data support, fault propagation law verification, multi-path consistency verification and model prediction reliability. Based on the fault propagation path analysis, the confidence of the location result is evaluated in the following steps: S401, path relevance scoring: scoring candidate paths using a machine learning model. The higher the path score, the higher the confidence level. S402, path weight characteristics: whether the total weight of the path is within a reasonable range; the lower the weight, the more likely the path is the actual propagation path, and the higher the confidence; S403, historical data support: whether the current positioning result matches the historical fault event; if the fault source node and propagation path are consistent with the historical pattern, the confidence is increased; S404, fault propagation law verification: analyze the time series data of fault propagation and check whether the propagation path conforms to the causal relationship; if the path conforms to the propagation law, the confidence level is increased; S405, multi-path consistency verification: among multiple candidate paths, check whether each path points to the same fault source node; if multiple high-scoring paths point to the same fault source node, the confidence is increased; S406, model prediction reliability: Combine the performance indicators of the machine learning model to evaluate the applicability of the model to the current positioning task; if the model performs well in similar scenarios, the confidence level is increased.

8. The method for locating fault events based on topology model tracing and analysis as claimed in claim 7 is characterized by: Quantify each influencing factor and assign a confidence score to each influencing factor. The specific steps include the following: Path relevance score: directly take the path relevance probability output by the machine learning model, recorded as P_path; Path weight characteristic score: calculate a score based on the total weight of the path: set the path weight range to [min_w, max_w], and the current path weight is w; Path weight score S_weight: S_weight = 1 - (w - min_w) / (max_w - min_w); Historical data support score: Based on the frequency of similar failure events in historical data, the support score S_history is calculated: S_history = count_similar_events / total_events; Fault propagation law verification score: Verify the propagation law based on time series data. If the propagation path conforms to the law, the score is 1, otherwise it is 0; recorded as S_causal; Multipath consistency score: If multiple candidate paths point to the same fault source node, the score is 1, otherwise it is 0; denoted as S_consistency; Model prediction reliability score: The model performance index is used as the score, denoted as S_model.

9. The method for locating fault events based on topology model tracing and analysis according to claim 8 is characterized in that: Based on the above scores, the overall confidence is calculated using the weighted average method, which includes the following steps: Weighted average method: assume that the weights of each factor are w1, w2, w3, w4, w5, and w6, and the weights are adjusted according to actual conditions; Comprehensive confidence calculation formula: Confidence = w1 * P_path + w2 * S_weight + w3 * S_history + w4 * S_causal + w5 * S_consistency + w6 * S_model; Constraints: w1 + w2 + w3 + w4 + w5 +w6 = 1; Example weight distribution: path relevance score: w1 = 0.3, path weight characteristic score: w2 = 0.2, historical data support score: w3 = 0.2, propagation law verification score: w4 = 0.1, multi-path consistency score: w5 = 0.1, model prediction reliability score: w6 = 0.

1.

10. The method for locating fault events based on topology model tracing and analysis according to any one of claims 1 to 9, characterized in that: The specific steps of confidence verification and output include the following: Verify confidence results: verify the calculated confidence to ensure that the confidence value is associated with the actual positioning accuracy; Output confidence results: output positioning results, including the most likely fault source node; Related propagation path information; confidence level of positioning results; maintenance personnel go to the location of the positioning results for maintenance and verify the accuracy of the positioning results; if the positioning position is far away, reposition the position and adjust the relevant coefficients in the model according to the coordinate changes of the position correction; Output key factors affecting confidence: Provide suboptimal paths and their confidence for analysis of multiple fault sources or complex scenarios.

Images