An edge-computing-based distribution network distributed cooperative monitoring system

Abstract

This invention relates to the field of distributed collaborative monitoring technology and discloses a distributed collaborative monitoring system for power distribution networks based on edge computing, comprising a sensing layer, an edge collaboration layer, and a regional decision layer. The sensing layer includes a multi-source data sensing unit for real-time acquisition of electrical, mechanical, thermal, and acoustic signals from circuit breakers. The edge collaboration layer includes a fault risk entropy assessment module for calculating fault risk entropy based on multi-source signal fusion. When the fault risk entropy exceeds a preset threshold, a task migration decision is triggered. This edge computing-based distributed collaborative monitoring system for power distribution networks introduces electrical topology correlation as the core criterion for target node selection during task migration, prioritizing nodes with close electrical distance and high topological correlation as collaborative targets. This allows the post-migration collaborative diagnosis to comprehensively utilize monitoring data from electrically adjacent nodes, improving the accuracy and reliability of diagnostic conclusions.

Description

Technical Field

[0001] This invention relates to the field of distributed collaborative monitoring technology, and in particular to a distributed collaborative monitoring system for power distribution networks based on edge computing. Background Technology

[0002] With the advancement of smart grid construction, distribution network monitoring systems are evolving from traditional centralized architectures to distributed edge computing architectures. In existing technologies, edge computing nodes are typically deployed in distribution substations or transformer areas to handle real-time monitoring and fault diagnosis of critical equipment such as circuit breakers within that area. When an edge node is unable to process a diagnostic task in a timely manner due to excessive load or insufficient resources, the task is migrated to other edge nodes for collaborative processing. Current task migration strategies primarily select target nodes based on physical distance or load status, resulting in relatively fixed migration strategies and a lack of consideration for the electrical characteristics of the distribution network.

[0003] However, distribution networks have a unique electrical topology, and devices on the same feeder are electrically interconnected. When a device shows signs of failure, adjacent devices often exhibit abnormal signals as well. Selecting migration targets solely based on physical distance or load status, without considering the electrical topology relationships between nodes, prevents the use of monitoring data from neighboring electrical nodes for correlation analysis during collaborative diagnosis, thus limiting the accuracy and reference value of diagnostic conclusions. Summary of the Invention

[0004] The technical problem to be solved by this invention is that the existing technology has the disadvantage of limited reference value of monitoring data. To address this, we propose a distributed collaborative monitoring system for power distribution networks based on edge computing.

[0005] To achieve the above objectives, this application adopts the following technical solution: a distributed collaborative monitoring system for power distribution networks based on edge computing, comprising a sensing layer, an edge collaboration layer, and a regional decision-making layer;

[0006] The sensing layer includes a multi-source data sensing unit, which is used to collect electrical, mechanical, thermal and acoustic signals of the circuit breaker in real time.

[0007] The edge collaboration layer includes: a fault risk entropy assessment module, which is used to calculate the fault risk entropy based on multi-source signal fusion, and trigger a task migration decision when the fault risk entropy exceeds a preset threshold.

[0008] The electrical topology management module is used to calculate the topological correlation between nodes based on the distribution network topology and line impedance.

[0009] The node capability monitoring module is used to monitor the computing power margin, network stability, and historical migration success rate of each edge node, and to calculate the node capability matching degree.

[0010] The target selection module is used to calculate a comprehensive score based on the weighted sum of the topological correlation degree and the node capability matching degree, and select the node with the highest comprehensive score as the target migration node;

[0011] The task migration execution module is used to break down the fault diagnosis task according to the risk level and migrate it to the target migration node.

[0012] The regional decision-making layer includes a migration benefit assessment module, which is used to evaluate the effectiveness of task migration and update the historical migration success rate.

[0013] Preferably, the fault risk entropy assessment module preprocesses and extracts features from multi-source monitoring signals, calculates the probability of occurrence of each fault type based on the extracted feature vectors, and then calculates the real-time fault risk entropy. The calculated fault risk entropy is compared with multiple preset thresholds to determine the risk level.

[0014] Preferably, the electrical topology management module calculates the electrical distance between nodes based on the distribution network topology using the shortest path algorithm. The electrical distance is the sum of the impedances of each line on the shortest path between two nodes, and the topology correlation degree is the reciprocal of the electrical distance. When the switch status in the distribution network changes, the electrical topology management module updates the topology and recalculates the electrical distance and topology correlation degree.

[0015] Preferably, when calculating the comprehensive score, the target selection module gives greater weight to topology correlation than to node capability matching, so as to prioritize the selection of nodes with high electrical topology correlation as target migration nodes.

[0016] Preferably, the task migration execution module decomposes the fault diagnosis task into feature extraction subtasks, feature filtering subtasks, classification diagnosis subtasks, and result verification subtasks. The task migration execution module adopts a differentiated migration strategy based on the risk level determined by the fault risk entropy assessment module. Specifically, when the risk level is determined to be a high risk level, the feature extraction subtask, feature filtering subtask, and classification diagnosis subtask are migrated; when the risk level is determined to be an extremely high risk level, all subtasks are migrated and a dual verification mechanism is initiated.

[0017] Preferably, the migration benefit evaluation module receives task execution feedback from the target migration node, calculates the benefit value of this task migration, and determines the benefit value based on the improvement value of diagnostic accuracy, the saving value of response time, and the value of communication overhead. When the benefit value reaches the preset benefit threshold, this task migration is recorded as a successful case; otherwise, it is marked as a case to be optimized.

[0018] Preferably, the dual verification mechanism involves the source edge node and the target migration node simultaneously executing a result verification subtask to compare the consistency of the diagnostic results of the two nodes.

[0019] Preferably, the multi-source data sensing unit is deployed at the monitoring points of each circuit breaker in the distribution network. The electrical quantity signals include three-phase voltage, three-phase current, active power, reactive power and harmonic content. The mechanical quantity signals include the current waveform of the opening and closing coil, the travel displacement of the moving contact and vibration acceleration. The thermal quantity signals include the contact temperature, coil temperature and ambient temperature and humidity. The acoustic signals include the acoustic characteristics of partial discharge and the sound characteristics of electric arc.

[0020] Preferably, when the node capability monitoring module calculates the node capability matching degree, the weight of computing power margin is the highest, followed by the weight of historical migration success rate, and the weight of network stability is the lowest.

[0021] Preferably, the preset multiple thresholds include a trigger threshold, a higher risk threshold, and an extremely high risk threshold; when the fault risk entropy exceeds the trigger threshold, a task migration decision is triggered; when the fault risk entropy exceeds the higher risk threshold, it is determined to be at a higher risk level; when the fault risk entropy exceeds the extremely high risk threshold, it is determined to be at an extremely high risk level.

[0022] The technical effects and advantages of this invention are as follows:

[0023] In this invention, electrical topology correlation is introduced as the core criterion for target node selection during task migration. Nodes with close electrical distance and high topology correlation are prioritized as collaborative targets, enabling the post-migration collaborative diagnosis to comprehensively utilize monitoring data from electrically adjacent nodes, thereby improving the accuracy and reliability of diagnostic conclusions. Simultaneously, this invention employs a progressive migration strategy based on fault risk levels, dynamically adjusting the content of the migrated sub-tasks to achieve a balance between computational resources and diagnostic accuracy. Through closed-loop learning, migration decisions are continuously optimized, allowing the system to adaptively improve collaborative efficiency. Attached Figure Description

[0024] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts:

[0025] Figure 1 This is a system architecture diagram of the present invention;

[0026] Figure 2 This is a flowchart of the task migration decision process for the edge collaboration layer of the present invention. Detailed Implementation

[0027] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0028] Reference Figure 1-2 As shown, the present invention provides a technical solution: a distributed collaborative monitoring system for power distribution networks based on edge computing, comprising a sensing layer, an edge collaboration layer, and a regional decision-making layer.

[0029] The sensing layer comprises multi-source data sensing units, deployed at monitoring points of each circuit breaker in the distribution network. These units are used to collect electrical, mechanical, thermal, and acoustic signals from the circuit breakers in real time. Electrical signals include three-phase voltage, three-phase current, active power, reactive power, and harmonic content, with a sampling frequency of at least 1 kHz. Mechanical signals include the waveforms of the opening and closing coils, the displacement of the moving contact, and vibration acceleration. Thermal signals include contact temperature, coil temperature, and ambient temperature and humidity. Acoustic signals include the acoustic characteristics of partial discharge and the sound characteristics of an electric arc.

[0030] The edge collaboration layer is connected to the perception layer and includes a fault risk entropy assessment module, an electrical topology management module, a node capability monitoring module, a target selection module, and a task migration execution module.

[0031] The fault risk entropy assessment module receives multi-source monitoring signals uploaded by the multi-source data sensing unit, preprocesses and extracts features from the multi-source monitoring signals, calculates the probability of occurrence of each fault type based on the extracted feature vectors, and then calculates the real-time fault risk entropy. The formula for calculating the fault risk entropy is: ;

[0032] in, For the circuit breaker at any time The fault risk entropy, For the first The probability of occurrence of this type of failure risk. This represents the total number of fault risk types.

[0033] The fault risk entropy assessment module compares the calculated fault risk entropy with a preset trigger threshold. Higher risk threshold and extremely high risk threshold A comparison is made when the fault risk entropy exceeds the trigger threshold. The system generates task migration trigger instructions and determines the risk level at the same time.

[0034] The electrical topology management module stores and maintains the distribution network topology, which includes a node set, an edge set, and an impedance matrix. The node set represents all edge nodes in the system, and the edge set represents the impedance values ​​of each line. Based on the topology, the module uses a shortest path algorithm to calculate the electrical distance between any edge node corresponding to a circuit breaker and other edge nodes. The electrical distance is the sum of the impedances of all lines on the shortest path between the two nodes, calculated using the following formula: ;

[0035] in, For circuit breakers Corresponding edge nodes and edge nodes Electrical distance between them For the shortest path, the th The impedance value of the line segment, This represents the number of line segments. The electrical topology management module calculates topology correlation based on electrical distance. The formula for calculating topology correlation is: ;

[0036] in, For circuit breakers Corresponding edge nodes and edge nodes The electrical topology correlation between components. When the switch status changes in the distribution network, the electrical topology management module automatically updates the topology and recalculates the electrical distances and topology correlations.

[0037] Specifically, the electrical topology management module stores the distribution network topology diagram G(V,E), where V is the set of nodes, i.e., the set of all edge nodes in the system, and E is the set of edges, i.e., the impedance values ​​of each line. When it is necessary to calculate the topology correlation, the module calculates the electrical path from any edge node corresponding to a circuit breaker to other edge nodes based on the shortest path algorithm, and accumulates the impedance values ​​of each line segment along the electrical path to obtain the electrical distance. Then take the reciprocal to get the topological correlation degree. When the switch status changes in the distribution network, the module receives the switch status update message, reconstructs the topology diagram and updates the electrical distance matrix and topology correlation matrix to ensure that the topology information is consistent with the actual operating status of the distribution network.

[0038] The node capability monitoring module is used to monitor the overall status of each edge node in real time. This overall status includes real-time computing power margin, network communication stability, and historical task migration success rate. The real-time computing power margin is calculated as follows: ;

[0039] in, For edge nodes Real-time computing power margin For CPU utilization, The formulas for calculating memory utilization and network communication stability are as follows: ;

[0040] in, For edge nodes Network communication stability, The formula for calculating the historical task migration success rate is: where the variance of the recent communication delay sequence is given. ;

[0041] in For edge nodes The success rate of historical mission migration The number of times a migration task can be successfully completed. To determine the total number of migration tasks, the node capability monitoring module calculates the node capability matching degree using a weighted summation method based on real-time computing power margin, network communication stability, and historical task migration success rate. The calculation formula is as follows: ;

[0042] in, For edge nodes Compared to circuit breakers Compatibility of abilities As a weight for real-time computing power margin, The weighting of the success rate of historical mission migration. The weights for network communication stability, and satisfying .

[0043] The node capability monitoring module collects status information of each edge node at fixed intervals; the computing power margin is obtained by acquiring the node's CPU utilization and memory utilization, taking the larger of the two, subtracting 1, and obtaining a value between 0 and 1, with a larger value indicating more abundant computing power; network stability is obtained by collecting the communication delay sequence of the most recent period, calculating the variance of the delay sequence, with a smaller variance indicating less latency fluctuation and a more stable network, and taking the reciprocal of the variance as the network stability index; the historical migration success rate is obtained by statistically analyzing the proportion of successful migrations in the node's past N task migrations out of the total number of migrations; finally, the three indicators are weighted and summed according to preset weights to obtain the node capability matching degree. This value reflects the overall capability of the target node to receive and complete the migration task.

[0044] The target selection module is connected to both the electrical topology management module and the node capability monitoring module. It receives the topology correlation score from the electrical topology management module and the node capability matching score from the node capability monitoring module. The target selection module calculates the comprehensive score for each candidate edge node using a comprehensive scoring formula: ;

[0045] in, For edge nodes For circuit breakers Overall score for task migration For topological correlation, For node capability matching degree, As the first weight, It is the second weight, and the first weight Greater than the second weight The system prioritizes nodes with high electrical topology correlation. The target selection module sorts the comprehensive scores in descending order and selects the edge node with the highest comprehensive score as the target migration node. It also verifies the availability of the target migration node, including whether the computing power margin meets the threshold condition and whether the network is connected.

[0046] The target selection module first receives the topology correlation matrix output by the electrical topology management module and the node capability matching list output by the node capability monitoring module; then it filters nodes with a topology correlation greater than a minimum threshold as a candidate node set, excluding nodes with excessively distant electrical relationships; finally, it calculates a comprehensive score for each candidate node. The weight of topological association Weight greater than node capability matching degree This reflects the principle of prioritizing electrical topology; candidate nodes are sorted in descending order of comprehensive score, and the node with the highest score is selected as the target migration node; finally, the availability of the target migration node is verified, including checking whether its computing power margin meets the preset threshold and whether the network is connected. If the verification fails, the node with the second highest score is selected, until a target migration node that meets the conditions is found.

[0047] The task migration execution module responds to the task migration trigger command generated by the fault risk entropy assessment module and connects to the target selection module to obtain the target migration node. The task migration execution module decomposes the fault diagnosis task into feature extraction subtasks, feature filtering subtasks, classification diagnosis subtasks, and result verification subtasks. The feature extraction and feature filtering subtasks are lightweight tasks, while the classification diagnosis and result verification subtasks are heavyweight subtasks. The task migration execution module determines the migration strategy based on the risk level indicated by the fault risk entropy: when the risk level is high, the feature extraction, feature filtering, and classification diagnosis subtasks are migrated; when the risk level is extremely high, all subtasks are migrated and a dual verification mechanism is initiated. This dual verification mechanism involves the source edge node and the target migration node simultaneously executing the result verification subtask and comparing the results for consistency. The task migration execution module encrypts the migration data, transmits it over the network to the target migration node, receives the diagnostic results returned by the target migration node, and verifies them.

[0048] The task migration execution module breaks down the fault diagnosis task into four parts: feature extraction, feature filtering, classification diagnosis, and result verification. The feature extraction and feature filtering subtasks have lower computational complexity and data requirements, while the classification diagnosis and result verification subtasks have higher computational complexity and significantly impact diagnostic accuracy. The module executes differentiated migration strategies based on the risk level output by the fault risk entropy assessment module: when the risk level is high, the feature extraction, feature filtering, and classification diagnosis subtasks are migrated to the target migration node for execution, while only the result verification subtask is executed locally on the source edge node to balance diagnostic accuracy and communication overhead; when the risk level is extremely high, all subtasks are migrated to the target migration node for execution, while the source edge node retains a copy of the result verification subtask and initiates a dual verification mechanism. That is, the source edge node and the target migration node execute the result verification subtask simultaneously, comparing the consistency of the diagnostic results between the two nodes. If they are consistent, the diagnosis is confirmed as valid; otherwise, a manual intervention mechanism is triggered. During the migration process, the data is encrypted and transmitted to the target migration node via the communication network, and the module receives the diagnostic results returned by the target migration node for verification.

[0049] The regional decision-making layer communication connection is linked to the edge collaboration layer, including a migration benefit evaluation module. This module receives task execution feedback from the target migration node and calculates the benefit value of the current migration task. The formula for calculating the benefit value is: ;

[0050] in The value represents the efficiency of task migration. As the first benefit weight, To improve diagnostic accuracy, As the second benefit weight, To save value in response time, As the third benefit weight, The communication overhead value is [value]. The diagnostic accuracy improvement value is the increment of diagnostic accuracy before and after migration; the response time saving value is the reduction in diagnostic response time before and after migration; and the communication overhead value is the product of data transmission volume and bandwidth unit price. The migration benefit evaluation module compares the benefit value with a preset benefit threshold. When the benefit value is greater than or equal to the benefit threshold, the migration task is recorded as a successful case and stored in the case library; when the benefit value is less than the benefit threshold, it is marked as a case to be optimized. The migration benefit evaluation module updates the historical task migration success rate in the node capability monitoring module based on the evaluation results.

[0051] Example 1: This example describes an application scenario where a distributed collaborative monitoring system is used in an urban power distribution network to monitor the status of medium-voltage circuit breakers. This network covers several distribution substations, each equipped with edge computing nodes to perform real-time monitoring and fault diagnosis of circuit breakers within its substation. These edge nodes are interconnected via a fiber optic network. In actual operation, the power load varies significantly between different substations, especially during peak hours. In some substations, the computing resources of edge nodes approach saturation, while neighboring substations have computing power margins.

[0052] Taking a distribution substation in a residential community as an example, a 10kV feeder circuit breaker in this substation is equipped with a multi-source data sensing unit. The sensing unit includes electrical quantity acquisition subunits, mechanical quantity acquisition subunits, thermal quantity acquisition subunits, and acoustic quantity acquisition subunits, which are used to acquire three-phase current and voltage waveforms, opening and closing coil current and moving contact travel curves, contact temperature rise data, and vibration and acoustic characteristics during operation, respectively. All of the above monitoring signals are timestamped upon generation and uploaded in real time to the corresponding edge collaboration layer of this substation.

[0053] During the peak summer electricity consumption period, the edge node of this distribution area was already under strain due to the simultaneous processing of monitoring data from multiple devices. Under these circumstances, the fault risk entropy assessment module located within the node performed a fusion analysis of the received multi-source signals. The analysis results showed that the circuit breaker's contact temperature exhibited an unexpected rise, the frequency domain characteristics of the vibration sound pattern shifted, and the three-phase current imbalance exceeded the normal fluctuation range. Based on these characteristics, the fault risk entropy assessment module calculated the real-time fault risk entropy value of the circuit breaker according to a preset model. Upon comparison, this value exceeded the preset trigger threshold. Therefore, the circuit breaker was determined to be of a high risk level, and a trigger command to initiate task migration was immediately generated.

[0054] Furthermore, if a traditional fixed allocation model is used, even if the edge node is already under high load, the complete diagnostic process for that high-risk circuit breaker still needs to be executed. This increases the latency of the diagnostic response, potentially causing the optimal time window for fault warning to be missed. By introducing fault risk entropy, the system can proactively identify high-risk objects and prioritize scheduling resources for them.

[0055] After the task migration trigger command is generated, the electrical topology management module immediately intervenes. Based on the current topology connection relationship of the distribution network, this module evaluates the electrical distance and topology correlation index between the source node where the circuit breaker is located and other candidate edge nodes. After calculation, the edge nodes corresponding to the two adjacent substations that are closest to the source node in terms of electrical distance have a significantly higher topology correlation than other nodes that are physically close but have a higher electrical level.

[0056] Meanwhile, the node capability monitoring module collects the real-time status of each candidate node, including the node's remaining computing power, network communication latency, and the success rate of historical task migration. One edge node in an electrically adjacent substation has sufficient computing power margin and stable communication quality; while the edge node in another electrically adjacent substation is currently under high load and is not suitable for receiving new tasks.

[0057] The target selection module then integrates the two types of information mentioned above—topological relevance and node capability matching—to assign a weighted score to each candidate node. Ultimately, the system selects the adjacent edge node that possesses both high topological relevance and sufficient computing power as the target node for this migration.

[0058] When selecting target nodes, priority should be given to electrical topology correlation, rather than simply physical proximity or load status. Electrically adjacent nodes often monitor devices within the same fault range, and their monitoring data exhibit some correlation. During diagnosis, utilizing computational resources allows the diagnostic model to comprehensively consider the data characteristics of adjacent nodes, improving the accuracy of fault location. If nodes are randomly selected based solely on physical distance or load indicators, their diagnostic value will be significantly reduced if the selected nodes are on unrelated branches in the electrical topology.

[0059] After identifying the target node, the task migration execution module adopted a partial migration strategy based on the previously determined "higher risk level." Specifically, this migration packaged and migrated the feature extraction subtask, feature screening subtask, and classification diagnosis subtask to the target node for execution; while the result verification subtask remained in the local environment of the source edge node. The granularity of the migration content was dynamically adjusted according to different risk levels to seek a balance between diagnostic quality and network overhead. If a full migration were performed indiscriminately, although it would be simple to implement, the movement of a large amount of raw waveform data would inevitably consume communication bandwidth and introduce additional latency.

[0060] After receiving the task package, the target migration node utilizes its available computing resources to complete feature processing and classification diagnosis, and sends the generated diagnostic conclusion, specifically "contact overheating warning," back to the source edge node. The source node then initiates a local result verification subtask to verify the conclusion. Once confirmed to be correct, it sends a notification to the operation and maintenance system, suggesting that attention be paid to the operating status of the circuit breaker.

[0061] Finally, the migration benefit evaluation module receives the execution feedback for this migration task and quantifies the migration benefit value. Evaluation dimensions include the maintenance of diagnostic accuracy, the improvement in end-to-end response time, and the acceptability of communication overhead. If the migration benefit value meets expectations, it can be recorded as a successful migration operation. Simultaneously, the historical task success rate record of the target migration node is updated accordingly, providing a more accurate prior reference for migration decisions in subsequent high-risk events. Through this accumulation and feedback mechanism, the system possesses the ability to optimize strategies over long-term operation.

[0062] In summary, the technical solution presented in this embodiment, in actual distribution network circuit breaker monitoring scenarios, effectively overcomes the shortcomings of existing technologies in dealing with edge node computing power bottlenecks and ensuring the real-time performance of high-value monitoring objects through risk entropy-driven task triggering, electrical topology-based target selection, differentiated partial migration execution, and benefit feedback-based strategy iteration.

[0063] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. A distributed collaborative monitoring system for power distribution networks based on edge computing, characterized in that, It includes a perception layer, an edge collaboration layer, and a regional decision-making layer; The sensing layer includes a multi-source data sensing unit, which is used to collect electrical, mechanical, thermal and acoustic signals of the circuit breaker in real time. The edge collaboration layer includes: a fault risk entropy assessment module, which is used to calculate the fault risk entropy based on multi-source signal fusion, and trigger a task migration decision when the fault risk entropy exceeds a preset threshold. The electrical topology management module is used to calculate the topological correlation between nodes based on the distribution network topology and line impedance. The node capability monitoring module is used to monitor the computing power margin, network stability, and historical migration success rate of each edge node, and to calculate the node capability matching degree. The target selection module is used to calculate a comprehensive score based on the weighted sum of the topological correlation degree and the node capability matching degree, and select the node with the highest comprehensive score as the target migration node; The task migration execution module is used to break down the fault diagnosis task according to the risk level and migrate it to the target migration node. The regional decision-making layer includes a migration benefit assessment module, which is used to evaluate the effectiveness of task migration and update the historical migration success rate.

2. The distributed collaborative monitoring system for power distribution networks based on edge computing according to claim 1, characterized in that: The fault risk entropy assessment module preprocesses and extracts features from multi-source monitoring signals, calculates the probability of occurrence of each fault type based on the extracted feature vectors, and then calculates the real-time fault risk entropy. The calculated fault risk entropy is compared with multiple preset thresholds to determine the risk level.

3. The distributed collaborative monitoring system for power distribution networks based on edge computing according to claim 1, characterized in that: The electrical topology management module is based on the distribution network topology and uses the shortest path algorithm to calculate the electrical distance between nodes. The electrical distance is the sum of the impedances of each line on the shortest path between two nodes, and the topology correlation degree is the reciprocal of the electrical distance. When the switch status in the distribution network changes, the electrical topology management module updates the topology and recalculates the electrical distance and topology correlation degree.

4. The distributed collaborative monitoring system for power distribution networks based on edge computing according to claim 1, characterized in that: When calculating the comprehensive score, the target selection module prioritizes nodes with high electrical topology correlation as target migration nodes, giving them a greater weight than node capability matching.

5. The distributed collaborative monitoring system for power distribution networks based on edge computing according to claim 1, characterized in that: The task migration execution module breaks down the fault diagnosis task into feature extraction subtask, feature filtering subtask, classification diagnosis subtask and result verification subtask. The task migration execution module adopts a differentiated migration strategy based on the risk level determined by the fault risk entropy assessment module. Specifically, when the risk level is determined to be high, the feature extraction subtask, feature screening subtask, and classification diagnosis subtask are migrated; when the risk level is determined to be extremely high, all subtasks are migrated and a dual verification mechanism is initiated.

6. The distributed collaborative monitoring system for power distribution networks based on edge computing according to claim 1, characterized in that: The migration benefit evaluation module receives task execution feedback from the target migration node, calculates the benefit value of this task migration, and determines the benefit value based on the improvement value of diagnostic accuracy, the saving value of response time, and the value of communication overhead. When the benefit value reaches the preset benefit threshold, this task migration is recorded as a successful case; otherwise, it is marked as a case to be optimized.

7. The distributed collaborative monitoring system for power distribution networks based on edge computing according to claim 5, characterized in that: The dual verification mechanism compares the consistency of the diagnostic results of the two nodes by simultaneously executing the result verification subtask at the source edge node and the target migration node.

8. The distributed collaborative monitoring system for power distribution networks based on edge computing according to claim 1, characterized in that: The multi-source data sensing unit is deployed at the monitoring points of each circuit breaker in the distribution network. The electrical signals include three-phase voltage, three-phase current, active power, reactive power, and harmonic content. The mechanical signals include the current waveform of the opening and closing coils, the displacement of the moving contact stroke, and vibration acceleration. The thermal signals include contact temperature, coil temperature, and ambient temperature and humidity. The acoustic signals include partial discharge acoustic characteristics and arc sound characteristics.

9. The distributed collaborative monitoring system for power distribution networks based on edge computing according to claim 1, characterized in that: When the node capability monitoring module calculates the node capability matching degree, the weight of computing power margin is the highest, followed by the weight of historical migration success rate, and the weight of network stability is the lowest.

10. The distributed collaborative monitoring system for power distribution networks based on edge computing according to claim 2, characterized in that: The preset thresholds include a trigger threshold, a higher risk threshold, and an extremely high risk threshold; When the fault risk entropy exceeds the trigger threshold, a task migration decision is triggered. When the fault risk entropy exceeds the higher risk threshold, it is determined to be a higher risk level; When the fault risk entropy exceeds the extremely high risk threshold, it is determined to be at an extremely high risk level.

Images