Cable accessory intelligent monitoring system based on multi-sensor fusion

Abstract

The invention relates to the technical field of electric power, in particular to a cable accessory intelligent monitoring system based on multi-sensor fusion. In the scheme of the application, the preprocessing module performs preliminary data processing at the edge, so that the data transmission burden is reduced; the regional processing module can establish a dynamic communication group to realize efficient data interaction between nodes. And the dynamic communication group dynamically adjusts the communication link according to the actual operation condition of the cable accessory and the network state. When a new cable accessory is accessed or an existing accessory has a fault, the system can quickly complete the reconfiguration of a monitoring strategy and a communication link, ensures that the monitoring work is continuously and efficiently carried out, and effectively improves the adaptability and flexibility of the system in a dynamic change network environment. The regional processing module can perform cross-regional data fusion, considers cable operation characteristic differences in different regions, and integrates and analyzes data in a targeted manner. Therefore, the system can carry out fine management on a large-scale complex cable network and accurately identify associated faults among regions.

Description

Technical Field

[0001] The present invention relates to the field of power technology, and particularly to an intelligent monitoring system for cable accessories based on multi-sensor fusion. Background Art

[0002] As a key carrier for power transmission, the safe and stable operation of cables is of utmost importance. Cable accessories, being the weak links in the cable line, such as cable terminals and joints, are prone to failures due to various factors like electricity, mechanics, and environment, thus threatening the reliability of the entire power system. According to incomplete statistics, approximately 30% - 40% of power system failures are related to cable accessories. Therefore, accurate monitoring of the operating status of cable accessories is of great practical significance. Traditional monitoring methods for cable accessories mostly rely on a single type of sensor. For example, only temperature sensors are used to monitor the temperature of cable joints. Such methods have a single monitoring dimension and are difficult to comprehensively reflect the complex operating status of cable accessories, often resulting in the failure to detect potential faults in a timely manner. Although some improved technologies have increased the types of sensors, each sensor works independently, and the data lacks effective fusion and collaborative analysis, making it impossible to grasp the health status of cable accessories from an overall perspective. In recent years, although some intelligent monitoring systems have attempted to apply advanced algorithms and communication technologies, there are obvious shortcomings in the data processing architecture. Most systems adopt a centralized data processing mode, where all raw data needs to be uploaded to the central server for processing. This not only incurs high communication costs and transmission delays but also, when faced with a large-scale cable network, the central server is extremely prone to processing bottlenecks due to excessive data volume, seriously affecting the real-time performance and accuracy of monitoring. Summary of the Invention

[0003] To solve the above problems, the present invention discloses an intelligent monitoring system for cable accessories based on multi-sensor fusion.

[0004] In one aspect, the present application discloses an intelligent monitoring system for cable accessories based on multi-sensor fusion, including: A sensor group, including a variety of sensor modules distributed at key positions of cable accessories, for collecting multi-dimensional physical quantity data reflecting the operating status of the cable; A preprocessing module, connected to the sensor group, for performing real-time preprocessing and feature extraction on the raw data; A regional processing module, arranged above the preprocessing module, for aggregating the processing results of multiple preprocessing modules, performing cross-regional data fusion, and establishing a dynamic communication group in the preprocessing module; A cloud processing module, respectively connected to the regional processing module and the dynamic communication group, for constructing a prediction model based on global data and realizing iterative updates of the monitoring strategy.

[0005] The sensor group at least includes a temperature sensing module, a discharge sensing module, a mechanical state sensing module, and an environmental parameter sensing module; Each module in the sensor group includes an original data transmission module, and the original data transmission module is used to establish a wireless communication link connection with a regional processing module or a cloud processing module according to a communication strategy.

[0006] The preprocessing module is provided with a high-speed data acquisition unit, and the high-speed data acquisition unit supports a multi-channel synchronous acquisition function; A lightweight model is set in the preprocessing module for real-time identification of key monitoring indicators; The preprocessing module is also configured with a local data cache unit for retaining historical data within a set time window.

[0007] The preprocessing module constructs a dynamic communication group according to a distributed hash table, and maps cable accessories to a ring network according to geographical locations through a consistent hashing algorithm, so that adjacent nodes automatically establish two-way subscription and publication channels.

[0008] The preprocessing module constructs a dynamic communication group through a consistent hashing algorithm, which specifically includes: Input the geographical coordinates of the cable accessory into the SHA-1 hash function to generate a 128-bit hash value as the node identifier, and all node identifiers form a circular address space; Maintain a routing table for each node, including the IP addresses, geographical coordinates, and communication keys of the predecessor and successor nodes on the ring; When a new node joins the network, query the nearest node on the ring through a communication protocol and recursively update the whole network routing table to realize the automatic discovery of geographically adjacent nodes.

[0009] The fault self-recovery method of the dynamic communication group includes: The node periodically sends heartbeat packets. If the responses from the predecessor / successor nodes are not received continuously for multiple times, route repair is triggered; Broadcast a route query request to other nodes on the ring to obtain the nearest surviving node; Recalculate the hash ring, and remove the faulty node from the routing table; Adjust the subscription relationships of adjacent nodes to ensure that each node is always connected to two neighbors.

[0010] On the other hand, the present application also discloses an intelligent monitoring method for cable accessories, which specifically includes the following steps: Collect multi-dimensional physical quantity data at key positions of cable accessories through a sensor group deployed in a distributed manner. The sensor group includes a temperature field reconstruction sensor array, a partial discharge ultrasonic sensor, and an interface stress sensor; In the preprocessing module, real-time preprocessing and feature extraction are performed on the collected raw data, including noise reduction, outlier removal, and eigenvalue calculation; In the region processing module, the processing results of multiple preprocessing modules are aggregated, cross-region data fusion is performed, and a dynamic communication group is established among the preprocessing modules; In the cloud processing module, a cable accessory health status prediction model is constructed based on global data, and the monitoring strategy is iteratively updated through machine learning algorithms; The updated monitoring strategy is sent to the region processing module and / or through the dynamic communication group.

[0011] The method further includes: The basic model is sent to the region processing module by the cloud processing module, and the region processing module coordinates the preprocessing modules for local training; The global model is updated through encrypted gradient aggregation, and the model update is periodically fed back to each hierarchical node.

[0012] In the solution of this application, the preprocessing module performs preliminary data processing at the edge, reducing the data transmission burden; the region processing module can establish a dynamic communication group to achieve efficient data interaction between nodes. The dynamic communication group dynamically adjusts the communication link according to the actual operation of the cable accessories and the network status. When a new cable accessory is connected or an existing accessory fails, the system can quickly complete the reconfiguration of the monitoring strategy and the communication link, ensuring that the monitoring work is uninterrupted and efficient, and effectively improving the adaptability and flexibility of the system in a dynamically changing network environment. The region processing module can perform cross-region data fusion, considering the differences in cable operation characteristics in different regions, and integrally analyzing the data in a targeted manner. Thus, the system can perform refined management on a large-scale complex cable network and accurately identify associated faults between regions. Brief Description of the Drawings

[0013] Figure 1 It is the architecture diagram of the cable accessory intelligent monitoring system based on multi-sensor fusion in the embodiment of this application; Figure 2 It is the structural schematic diagram of a sensor group in the embodiment of this application; Figure 3 It is the flowchart of the cable accessory intelligent monitoring method in the embodiment of this application; Figure 4 It is the method flowchart for constructing a dynamic communication group through the consistent hashing algorithm in the embodiment of this application. Detailed Embodiment

[0014] In order to make the object, features, and advantages of the present invention more obvious and understandable, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the embodiments described below are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention. The principles and features of the present invention will be described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0015] The term "including" in the specification or claims of the present invention and other similar expressions mean covering non-exclusive inclusion. For example, a process, method, system, or device that includes a series of steps or units is not limited to the listed steps or units.

[0016] Example 1: As Figure 1-2 shown, an intelligent monitoring system for cable accessories based on multi-sensor fusion includes: A sensor group, which contains a variety of sensor modules distributed at key positions of the cable accessories and is used to collect multi-dimensional physical quantity data reflecting the operating state of the cable. The sensor group can be composed of a variety of different types of sensor modules. For example, a temperature sensor is installed at the cable joint to monitor the temperature change caused by the current heating effect at the joint part; a partial discharge sensor is installed at the cable terminal to detect the partial discharge phenomenon caused by possible insulation defects. Through the above targeted distribution, various key information during the cable operation can be captured more comprehensively.

[0017] A preprocessing module, connected to the sensor group, is used to perform real-time preprocessing and feature extraction on the original data; perform real-time preprocessing and feature extraction on the original data. During the real-time preprocessing process, noise reduction processing is performed on the original data to remove useless signals generated due to environmental interference, sensor self-noise, and other factors, and improve the quality of the data. For example, a filtering algorithm is used to filter out the common 50Hz power frequency interference in the power system. At the same time, normalization processing is performed on the data to unify the data with different dimensions and value ranges collected by different sensors to a standard scale for subsequent analysis and comparison. During feature extraction, a preset algorithm is used to extract key features that can characterize the operating state of the cable from the preprocessed data. For example, for temperature data, features such as temperature change rate and temperature peak value are extracted; for partial discharge data, features such as the amplitude, frequency, and phase of the discharge pulse are extracted. These features can more concisely and effectively reflect the operating state of the cable.

[0018] The regional processing module is set above the preprocessing module and is used to aggregate the processing results of multiple preprocessing modules, perform cross-regional data fusion, and establish dynamic communication groups in the preprocessing module; The regional processing module is used to aggregate the processing results of multiple preprocessing modules. It collects and integrates the preprocessed data and features from different geographical locations corresponding to different cable accessories. For example, in the power grid of a city, the data related to cable accessories in each substation and transmission line section are gathered together to comprehensively analyze the overall operation status of the cables at the regional level and discover some systematic problems that may be overlooked in the monitoring of individual cable accessories. In addition to aggregating the data within the region, the regional processing module also performs cross-regional data fusion. In a large power system, there may be mutual influences and correlations between the cable operation states in different regions. Through cross-regional data fusion, information within a wider range can be comprehensively considered to improve the accuracy and comprehensiveness of the assessment of the cable operation state. For example, when the power consumption load in one region suddenly increases, it may affect the cable operation in adjacent regions. Through cross-regional data fusion, such correlations can be discovered in a timely manner and corresponding decisions can be made. At the same time, the regional processing module can establish dynamic communication groups in the preprocessing module. The dynamic communication groups dynamically organize the communication connections between the preprocessing modules according to factors such as the geographical location and operation state of the cable accessories. For example, when an abnormality occurs in the cables in a certain region, the regional processing module can quickly establish a communication group with the preprocessing modules in that region and related regions, enabling the relevant modules to share data and information in a timely manner and cooperate in fault diagnosis and analysis. In addition, for tasks that require quick response, the dynamic communication groups can also be directly connected to the cloud processing module, directly send data to the cloud processing module, and the cloud processing module directly issues control strategies to each module within the dynamic communication group to further improve the system response speed.

[0019] The cloud processing module is connected to the regional processing module and the dynamic communication group respectively and is used to build a prediction model based on global data and realize the iterative update of the monitoring strategy.

[0020] The cloud processing module is connected to the regional processing module and the dynamic communication group respectively. The connection with the regional processing module enables it to obtain the comprehensive data from each region and achieve global control of the cable operation state of the entire power system; the connection with the dynamic communication group can ensure that it can obtain the latest and detailed cable operation information in real time and update the monitoring strategy in a timely manner. In one embodiment, the cloud processing module can perform time series analysis on parameters such as the temperature and partial discharge amount of the cable by establishing a prediction model based on a neural network, predict the change trends of these parameters in a future period of time, and thus determine whether the cable is likely to fail. If it is found that the cables in a certain area frequently fail, but the original monitoring strategy does not pay enough attention to this area, the cloud processing module can adjust the monitoring strategy, increase the monitoring frequency and key parameters of the cables in this area, and improve the monitoring effect and reliability of the entire system.

[0021] Embodiment 2: The sensor group at least includes a temperature sensing module, a discharge sensing module, a mechanical state sensing module, and an environmental parameter sensing module; Each module in the sensor group includes an original data transmission module, and the original data transmission module is used to establish a wireless communication link connection with the area processing module or the cloud processing module according to the communication strategy.

[0022] Among them, the mechanical state sensing module can monitor the mechanical vibration generated by the cable due to the action of current through devices such as vibration sensors to detect mechanical fault hazards such as loose internal connections of the cable. The environmental parameter sensing module can cover sensors such as temperature and humidity, air pressure, and harmful gas concentration, and is used to monitor the influence of the surrounding environmental factors of the cable on its operating state. For example, excessive humidity may cause the insulation performance of the cable to decline, and monitoring the environmental humidity helps to predict risks in advance.

[0023] In terms of data transmission, each sensor module is equipped with an original data transmission module, and a wireless communication link connection with the area processing module or the cloud processing module is constructed according to the communication strategy. When an emergency fault indication appears in the cable, such as a sudden large increase in the partial discharge amount, the original data transmission module of the discharge sensing module can quickly establish a low-latency and high-bandwidth wireless communication link with the cloud processing module according to the high-priority communication strategy to ensure that key data can be uploaded in time for quick decision-making and fault handling. In the normal operating state, each module can establish a stable wireless link with the area processing module according to the conventional communication strategy, regularly upload data for area-level summary analysis, optimize the utilization of data transmission resources, and improve the operating efficiency of the entire monitoring system.

[0024] Embodiment 3: The preprocessing module is provided with a high-speed data acquisition unit, and the high-speed data acquisition unit supports multi-channel synchronous acquisition function; in the intelligent monitoring system of cable accessories, the sensor group includes a variety of sensor modules, such as temperature perception module, discharge perception module, mechanical state perception module, and environmental parameter perception module, etc. Each module may correspond to one or more data channels. The temperature perception module may arrange multiple temperature sensors at different positions of the cable accessory, and each sensor corresponds to a data channel. The high-speed data acquisition unit can collect data from these different channels at the same time to ensure data synchronization.

[0025] A lightweight model is set in the preprocessing module for real-time identification of key monitoring indicators; the lightweight model set in the preprocessing module is mainly used for real-time identification of key monitoring indicators. The key monitoring indicators include but are not limited to the temperature of the cable, partial discharge amount, mechanical vibration parameters, etc.

[0026] The preprocessing module is also configured with a local data cache unit for retaining historical data within a set time window. The local data cache unit configured by the preprocessing module is used to retain historical data within a set time window. The set time window can be adjusted according to actual needs, for example, it can be set to 1 hour, 1 day, etc. By analyzing the temperature historical data over a period of time, the change law of the cable temperature can be understood, and the future temperature change trend can be predicted. Secondly, in case of abnormal situations, the historical data can help technicians conduct fault troubleshooting and analysis.

[0027] Embodiment 4: The preprocessing module constructs a dynamic communication group according to a distributed hash table, and maps the cable accessories into a ring network by the consistent hashing algorithm, so that adjacent nodes automatically establish a two-way subscription and publication channel.

[0028] As Figure 4 shown, the preprocessing module constructs a dynamic communication group by the consistent hashing algorithm, specifically including: S201: Input the geographical coordinates of the cable accessories into the SHA-1 hash function to generate a 128-bit hash value as the node identifier, and all node identifiers form a circular address space; S202: Maintain a routing table for each node, including the IP addresses, geographical coordinates, and communication keys of the predecessor and successor nodes on the ring; S203: When a new node joins the network, query the nearest node on the ring through the communication protocol, and recursively update the whole network routing table to achieve automatic discovery of geographically adjacent nodes.

[0029] Exemplarily, the specific implementation steps include: Taking the geographical coordinates of the cable accessory as input, it is processed using the SHA-1 hash function to generate a 128-bit hash value, which serves as the node identifier. For example, for a certain cable accessory with geographical coordinates (longitude x, latitude y), after inputting this set of coordinates into the SHA-1 function, a 128-bit hash value in the form of 0x123456789abcdef... will be obtained.

[0030] The identifiers of all nodes together form a circular address space. The circular space is similar to a ring that is connected end to end, and the identifier of each node has a corresponding position on the ring.

[0031] For each node, a routing table is maintained. The routing table contains information about the predecessor and successor nodes on the ring, specifically including the IP address, geographical coordinates, and communication key. For example, in the routing table of node A, it will record the IP address 192.168.1.100, geographical coordinates (longitude m, latitude n), and communication key key123 of its predecessor node B, as well as the corresponding information of its successor node C.

[0032] With the help of the routing table, the node can clearly know the situation of the nodes adjacent to itself.

[0033] When a new node joins the network, it queries the node on the ring that is closest to itself through a communication protocol (such as the Kademlia protocol). For example, after the new node D joins the network, it will send a query request to find the existing node E that is closest to its hash value.

[0034] Then, the new node D will recursively update the routing tables of the entire network. That is, after finding the closest node E, E will assist D in notifying its own information to other relevant nodes, so that the routing tables of all nodes can be updated in a timely manner, thereby realizing the automatic discovery of geographically adjacent nodes.

[0035] After constructing the dynamic communication group and the circular network, adjacent nodes can automatically establish a two-way subscription-publishing channel. For example, between node A and its adjacent node B, A can subscribe to the data about the operating status of the cable accessory published by B, and at the same time A can also publish the data it has collected to B. The two-way subscription-publishing mechanism makes the data interaction between nodes more efficient and flexible, can share key information in a timely manner, and helps to quickly discover and handle possible problems with the cable accessory.

[0036] In the method of this embodiment, due to the adoption of the distributed hash table and the consistent hashing algorithm, the network can dynamically adapt to the joining and leaving of nodes. When a new cable accessory is connected to the system or a certain node fails, the network can automatically adjust the routing table to ensure the normal progress of communication.

[0037] At the same time, the cable accessories are mapped into a ring network according to their geographical location, so that geographically adjacent nodes can communicate and share data more conveniently. The monitoring of cable accessories is particularly important because the operating status of adjacent cable accessories may be correlated to a certain extent.

[0038] In addition, the bidirectional subscription-publish channel improves the data transmission efficiency between nodes, reduces the data transmission delay, and can deliver monitoring data to the required nodes in a timely manner.

[0039] The fault self-recovery method of the dynamic communication group includes: The node periodically sends heartbeat packets. If no response is received from the predecessor or successor node for multiple consecutive times, a routing repair is triggered. A routing query request is broadcast to other nodes on the ring to obtain the nearest surviving node. When a node (such as node A) detects a neighboring node failure and triggers routing repair, it broadcasts a routing query request to all other nodes on the ring. The purpose of the request is to obtain information about the surviving node closest to the failed node (such as the failed predecessor node B). Through broadcasting, other nodes on the ring receive this request, and if they know of any surviving nodes that meet the requirements, they feed this information back to node A, which initiated the request. For example, after receiving a routing query request from node A, nodes D, E, etc. determine the surviving node F closest to the failed node B based on the routing table information they maintain, and send relevant information about F (such as its IP address and node identifier) to node A.

[0040] The hash ring is recalculated to remove the faulty node from the routing table. After obtaining information about the nearest surviving node, the relevant nodes (such as Node A and other affected nodes) recalculate the hash ring. The hash ring is a logical ring structure composed of node identifiers (generated using a consistent hashing algorithm). The purpose of recalculating the hash ring is to redefine the position and adjacency of each node on the ring under the new circumstances. During this process, the faulty node (such as Node B) is removed from the routing tables of all nodes. Because the faulty node is no longer able to communicate normally, retaining it in the routing table will result in incorrect routing information and communication failures. For example, Node A will remove Node B from its routing table and update its predecessor or successor node information to the newly determined surviving node (such as Node F).

[0041] Adjust the subscription relationships of adjacent nodes to ensure that each node is always connected to two neighbors. After recalculating the hash ring and removing the faulty node, it is necessary to adjust the subscription relationships of adjacent nodes. Since the adjacent relationships of the nodes have changed (the faulty node is removed and the new adjacent nodes are determined), the subscription objects of each node also need to be changed accordingly. The goal of the adjustment is to ensure that each node is always connected to two neighbor nodes (the predecessor node and the successor node) and can correctly subscribe to and receive the information published by these two neighbor nodes. For example, originally node A subscribed to the information of faulty node B and successor node C. After B fails and the routing is repaired, A will readjust the subscription relationship and change it to subscribing to the information of the new predecessor node F and successor node C. This can ensure that the data interaction and communication between nodes can continue to proceed normally and maintain the stable operation of the dynamic communication group.

[0042] Through the above-mentioned fault self-recovery method, the dynamic communication group can automatically detect and repair faults when problems such as node failures or communication link interruptions occur, reconstruct a stable communication network, and ensure that the data transmission and collaborative work between each preprocessing module in the cable accessory intelligent monitoring system are not affected too much.

[0043] Embodiment 5: As Figure 3 shown, a cable accessory intelligent monitoring method specifically includes the following steps: S101: Collect multi-dimensional physical quantity data of key positions of cable accessories through a distributed sensor group. The sensor group includes a temperature field reconstruction sensor array, a partial discharge ultrasonic sensor, and an interface stress sensor; S102: Perform real-time preprocessing and feature extraction on the collected original data in the preprocessing module, including noise reduction processing, outlier removal, and eigenvalue calculation; S103: Aggregate the processing results of multiple preprocessing modules in the regional processing module, perform cross-regional data fusion, and establish a dynamic communication group between preprocessing modules; S104: Build a cable accessory health status prediction model based on global data in the cloud processing module, and realize the iterative update of the monitoring strategy through machine learning algorithms; S105: Send the updated monitoring strategy to the regional processing module and / or through the dynamic communication group.

[0044] In this embodiment, the updated monitoring strategy can be quickly sent to the regional processing module and conveyed through the dynamic communication group, ensuring that the entire monitoring system can operate in a timely manner according to the latest strategy, adapt to the complex and changeable cable operating environment, and enhance the timeliness and flexibility of the system response. The method further includes: S111: The cloud processing module distributes the basic model to the regional processing module, and the regional processing module coordinates the preprocessing module to perform local training. The cloud processing module serves as the core decision-making layer of the entire system and has abundant computing resources. Through it, a basic machine learning or deep learning model is constructed. This model can be an initial health status prediction model for cable accessories and has a certain generalization ability. Then, the cloud distributes this basic model to each regional processing module. The regional processing module is usually responsible for managing and coordinating the preprocessing modules within a certain range and has a certain computing and storage capacity.

[0045] After receiving the basic model, since the regional processing module is directly connected to the sensor group, it has the raw data collected locally. The raw data reflects the operating characteristics of cable accessories in a specific area and has certain regional characteristics and local laws. By performing local training in the preprocessing module, the characteristics of these local data can be fully utilized to fine-tune the basic model and make it more adaptable to the local cable operating environment.

[0046] S112: Update the global model through the encrypted gradient aggregation method and periodically feedback the model update to each hierarchical node. To protect the privacy and security of local data, the preprocessing module encrypts the calculated gradients. After receiving the aggregated encrypted gradients, the cloud processing module decrypts and further processes them, and then uses this update amount to update the global model. In this way, the global model can absorb the local data characteristics of each region, continuously optimize its own performance, and improve the accuracy of predicting the health status of cable accessories.

[0047] The technical means disclosed in the solution of the present invention are not limited to the technical means disclosed in the above embodiments, but also include technical solutions composed of any combination of the above technical features. It should be noted that for those of ordinary skill in the art of this technology, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements are also regarded as the protection scope of the present invention.

Claims

1. An intelligent monitoring system for cable accessories based on multi-sensor fusion, characterized in that, Comprising: A sensor group, including a variety of sensor modules distributed at key positions of cable accessories, for collecting multi-dimensional physical quantity data reflecting the operating state of the cable; A preprocessing module, connected to the sensor group, for performing real-time preprocessing and feature extraction on the original data; A regional processing module, arranged on the upper layer of the preprocessing module, for aggregating the processing results of multiple preprocessing modules, performing cross-regional data fusion, and establishing a dynamic communication group in the preprocessing module; A cloud processing module, connected to the regional processing module and the dynamic communication group respectively, for constructing a prediction model based on global data and realizing iterative update of the monitoring strategy.

2. The system according to claim 1, wherein: The sensor group at least includes a temperature sensing module, a discharge sensing module, a mechanical state sensing module, and an environmental parameter sensing module; Each module in the sensor group includes an original data transmission module, and the original data transmission module is used to establish a wireless communication link connection with the regional processing module or the cloud processing module according to the communication strategy.

3. The system according to claim 1, wherein: The preprocessing module is provided with a high-speed data acquisition unit, and the high-speed data acquisition unit supports multi-channel synchronous acquisition function; A lightweight model is set in the preprocessing module for real-time identification of key monitoring indicators; The preprocessing module is also configured with a local data cache unit for retaining historical data within a set time window.

4. The system according to claim 1, wherein: The preprocessing module constructs a dynamic dynamic communication group according to a distributed hash table, and maps the cable accessories into a ring network according to the geographical location through a consistent hashing algorithm, so that adjacent nodes automatically establish a two-way subscription and publication channel.

5. The system according to claim 4, characterized in that: The preprocessing module constructs a dynamic communication group through a consistent hashing algorithm, specifically including: Inputting the geographical coordinates of the cable accessories into the SHA-1 hash function to generate a 128-bit hash value as the node identifier, and all node identifiers constitute a circular address space; Maintaining a routing table for each node, including the IP addresses, geographical coordinates, and communication keys of the predecessor and successor nodes on the ring; When a new node joins the network, query the nearest node on the ring through the communication protocol, and recursively update the whole network routing table to realize automatic discovery of geographically adjacent nodes.

6. The system according to claim 1, wherein the fault self-recovery method of the dynamic communication group includes: The node periodically sends a heartbeat packet. If the response from the predecessor / successor node is not received continuously for multiple times, route repair is triggered; Broadcast a route query request to other nodes on the ring to obtain the nearest surviving node; Recalculate the hash ring and remove the faulty node from the routing table; Adjust the subscription relationship of adjacent nodes to ensure that each node is always connected to two neighbors.

7. An intelligent monitoring method for cable accessories based on the system according to any one of claims 1-6, characterized in that, Including the following steps: Collecting multi-dimensional physical quantity data at key positions of cable accessories through a sensor group deployed in a distributed manner, the sensor group including a temperature field reconstruction sensor array, a partial discharge ultrasonic sensor, and an interface stress sensor; Performing real-time preprocessing and feature extraction on the collected original data in the preprocessing module, including noise reduction processing, outlier removal, and eigenvalue calculation; Aggregating the processing results of multiple preprocessing modules in the regional processing module, performing cross-regional data fusion, and establishing a dynamic communication group between the preprocessing modules; Build a cable accessory health status prediction model based on global data in the cloud processing module, and implement iterative updates of the monitoring strategy through machine learning algorithms; Send the updated monitoring strategy to the regional processing module and / or through a dynamic communication group.

8. The method according to claim 6, wherein: The method further includes: Send the basic model to the regional processing module by the cloud processing module, and the regional processing module coordinates the preprocessing module for local training; Update the global model through encrypted gradient aggregation, and periodically feedback model updates to each level of nodes.

Images