Edge fusion diagnosis system and method for multi-source heterogeneous power transmission monitoring data

Abstract

The invention provides an edge fusion diagnosis system and method for multi-source heterogeneous power transmission monitoring data, and belongs to the technical field of power transmission line monitoring. The multi-source heterogeneous power transmission monitoring data edge fusion diagnosis system comprises a multi-source heterogeneous data access module, a data preprocessing and standardization module, an edge computing node, an edge node collaborative diagnosis module, a data compression and feature extraction module and a cloud fusion analysis platform. The method comprises the following steps: accessing multi-source heterogeneous data, preprocessing and standardizing, locally diagnosing by an edge computing node, triggering multi-node collaborative diagnosis when the credibility is insufficient, and uploading key feature data to a cloud for global analysis and model optimization. According to the method, comprehensive integration and efficient processing of multi-source data are realized, the real-time performance, accuracy and comprehensiveness of power transmission monitoring diagnosis are improved, and the efficient monitoring requirement of a complex power transmission network can be met.

Description

Technical Field

[0001] The present invention relates to the technical field of power transmission line monitoring, and in particular to an edge fusion diagnosis system and method for multi-source heterogeneous power transmission monitoring data. Background Art

[0002] Current transmission line monitoring faces challenges such as difficulty integrating heterogeneous data from multiple sources, insufficient real-time diagnostics, and poor accuracy. Existing technologies often rely on single-type data access, making it difficult to fully cover all the information required for transmission line monitoring. Data processing often relies on centralized cloud-based processing, resulting in large data transmission volumes and high latency, impacting timely diagnosis. Diagnostic models lack specificity and have limited processing capabilities for diverse data types. Single-node diagnostics are susceptible to data limitations and lack collaboration, making it difficult to achieve comprehensive and accurate fault diagnosis and trend prediction, thus failing to meet the demands of efficient monitoring of complex transmission networks.

[0003] Therefore, an edge fusion diagnosis system and method for multi-source heterogeneous power transmission monitoring data are proposed. Summary of the Invention

[0004] The present invention aims to solve the problems raised in the background technology and provides an edge fusion diagnosis system and method for multi-source heterogeneous power transmission monitoring data.

[0005] The specific technical solutions are as follows:

[0006] An edge fusion diagnosis system for multi-source heterogeneous power transmission monitoring data, comprising:

[0007] A multi-source heterogeneous data access module is used to access multi-source heterogeneous data generated during the transmission line monitoring process. The multi-source heterogeneous data includes real-time monitoring data collected by sensors, transmission line image data obtained by image acquisition equipment, transmission equipment status data, and environmental monitoring data;

[0008] A data preprocessing and standardization module, connected to the multi-source heterogeneous data access module, is used to clean, denoise and standardize the multi-source heterogeneous data, convert data in different formats into a unified data format, and identify and correct outliers in the data based on a deep learning algorithm;

[0009] An edge computing node, connected to the data preprocessing and standardization module, is deployed near the power transmission monitoring data source and is used to perform local diagnostic analysis on the preprocessed standardized data. The edge computing node has built-in intelligent diagnostic models for different power transmission monitoring objects, including a power transmission line image fault recognition model based on a convolutional neural network, a sensor time series data anomaly diagnosis model based on a recurrent neural network, and an equipment status assessment model based on a random forest.

[0010] An edge node collaborative diagnosis module is connected to the plurality of edge computing nodes, and is used to trigger the plurality of edge computing nodes to perform collaborative diagnosis when the credibility of the local diagnosis result of a single edge computing node is lower than a preset threshold, thereby improving the accuracy of the diagnosis result through data sharing and joint analysis;

[0011] A data compression and feature extraction module, connected to the edge computing node, is used to compress the diagnostic result data of the edge computing node and extract key feature data;

[0012] The cloud-based fusion analysis platform is connected to the data compression and feature extraction module, and is used to receive key feature data uploaded by the edge computing node, perform global fusion analysis in combination with historical data, realize trend prediction and tracing of transmission line faults, and feed back model optimization parameters to the edge computing node.

[0013] The above-mentioned edge fusion diagnostic system for multi-source heterogeneous power transmission monitoring data, wherein the sensor types supported by the multi-source heterogeneous data access module include temperature sensors, humidity sensors, wind speed sensors, tilt sensors, tension sensors and partial discharge sensors; the image acquisition equipment includes high-definition cameras, infrared thermal imagers and image acquisition devices carried by drones.

[0014] In the above-mentioned edge fusion diagnostic system for multi-source heterogeneous power transmission monitoring data, the deep learning algorithm adopted by the data preprocessing and standardization module is an outlier detection algorithm based on an autoencoder. Normal data is learned by constructing an autoencoder model, and outliers are identified by reconstructing errors, and outliers are corrected based on the distribution characteristics of normal data.

[0015] The above-mentioned edge fusion diagnosis system for multi-source heterogeneous power transmission monitoring data, wherein the edge node collaborative diagnosis module adopts a trusted data interaction mechanism based on blockchain. Each edge computing node acts as a node in the blockchain network. When sharing data, it records data interaction information through the distributed ledger of the blockchain to ensure the data's immutability and traceability. At the same time, smart contracts are used to realize the automatic allocation of collaborative diagnosis tasks and result verification.

[0016] The above-mentioned edge fusion diagnosis system for multi-source heterogeneous transmission monitoring data, wherein the cloud-based fusion analysis platform includes a data storage unit, a global diagnosis model unit and a model optimization unit; the data storage unit is used to store key feature data and historical monitoring data uploaded by the edge computing node; the global diagnosis model unit adopts a fusion diagnosis model based on a graph neural network to integrate the topological structure of the transmission line into the diagnosis and analysis process, and realize the correlation analysis of cross-regional and multi-type faults; the model optimization unit generates the optimization parameters of the built-in intelligent diagnosis model of the edge computing node through a reinforcement learning algorithm based on the deviation between the global diagnosis result and the local diagnosis result of the edge computing node.

[0017] The present invention also provides an edge fusion diagnosis method for multi-source heterogeneous power transmission monitoring data, comprising the following steps:

[0018] S1. Multi-source heterogeneous data access: The multi-source heterogeneous data access module collects multi-source heterogeneous data during the transmission line monitoring process. The multi-source heterogeneous data includes real-time monitoring data collected by sensors, transmission line image data obtained by image acquisition equipment, transmission equipment status data, and environmental monitoring data.

[0019] S2. Data Preprocessing and Standardization: The data preprocessing and standardization module processes the multi-source heterogeneous data received in step S1. First, it cleans and denoises the data to remove redundant data and noise interference. Then, it converts the data in different formats into a unified format. Finally, it uses a deep learning algorithm to identify and correct outliers in the data.

[0020] S3. Local diagnosis at the edge node: The standardized data preprocessed in step S2 is transmitted to the edge computing node, which calls the built-in intelligent diagnostic model for different transmission monitoring objects to perform local diagnostic analysis and obtain preliminary diagnostic results;

[0021] S4. Collaborative diagnosis at the edge node: The edge node collaborative diagnosis module determines the credibility of the preliminary diagnosis result obtained in step S3. If the credibility is lower than the preset threshold, multiple edge computing nodes are triggered to perform collaborative diagnosis, and more accurate diagnosis results are obtained through data sharing and joint analysis.

[0022] S5 data compression and upload: using the data compression and feature extraction module to compress the diagnostic data obtained in step S3 or step S4, extract key feature data, and upload the key feature data to the cloud fusion analysis platform;

[0023] S6. Cloud-based fusion analysis and model optimization: The cloud-based fusion analysis platform receives the key feature data uploaded in step S5 and performs a global fusion analysis based on historical data to predict and trace the trend of transmission line faults. Based on the analysis results, it generates optimization parameters for the built-in intelligent diagnosis model of the edge computing node and feeds them back to the edge computing node to optimize its diagnostic performance.

[0024] In the above-mentioned edge fusion diagnostic method for multi-source heterogeneous power transmission monitoring data, the specific process of identifying and correcting outliers in the data based on the deep learning algorithm in step S2 is as follows: constructing an outlier detection model based on an autoencoder, inputting normal multi-source heterogeneous data into the autoencoder for training, so that the autoencoder learns the distribution characteristics of normal data; inputting the data to be processed into the trained autoencoder, calculating the reconstruction error of the data, and when the reconstruction error is greater than a preset threshold, determining that the data is an outlier; and correcting the outlier using an interpolation method based on the distribution characteristics of the normal data.

[0025] In the above-mentioned edge fusion diagnosis method for multi-source heterogeneous transmission monitoring data, the working process of the intelligent diagnosis model called by the edge computing node in step S3 includes: for transmission line image data, a transmission line image fault recognition model based on a convolutional neural network is adopted to extract image features through the convolution layer and perform fault classification and identification through the fully connected layer; for sensor time series data, a sensor time series data anomaly diagnosis model based on a recurrent neural network is adopted to capture the time series dependency of the data and identify abnormal fluctuations in the data; for equipment status data, an equipment status assessment model based on a random forest is adopted to assess the health status of the equipment through multi-dimensional feature analysis.

[0026] In the above-mentioned edge fusion diagnosis method for multi-source heterogeneous transmission monitoring data, the specific process of edge node collaborative diagnosis in step S4 is as follows: when the credibility of the preliminary diagnosis result of a single edge computing node is lower than a preset threshold, the edge computing node sends a collaborative diagnosis request to the edge node collaborative diagnosis module; the edge node collaborative diagnosis module selects other edge computing nodes associated with the monitoring area of ​​the requesting node to form a collaborative diagnosis group based on the topological structure of the transmission line and the monitoring range of each edge computing node; each edge computing node in the collaborative diagnosis group shares relevant monitoring data and diagnostic intermediate results through a trusted data interaction mechanism based on blockchain; each edge computing node performs joint analysis based on the shared data and determines the final collaborative diagnosis result through a voting mechanism.

[0027] The above-mentioned edge fusion diagnosis method for multi-source heterogeneous transmission monitoring data, wherein the specific process of the global fusion analysis performed by the cloud-based fusion analysis platform in step S6 is as follows: the global diagnosis model unit of the cloud-based fusion analysis platform adopts a fusion diagnosis model based on a graph neural network, takes the towers, conductors, insulators and other equipment of the transmission line as graph nodes, and takes the connection relationship between the equipment as the edge to construct a transmission line topology map; takes the key feature data uploaded by each edge computing node as the attribute feature of the graph node, and performs feature learning and fusion on the topology map through the graph neural network to realize the correlation analysis of cross-regional and multi-type faults and obtain a global diagnosis result; the model optimization unit generates the optimization parameters of the built-in intelligent diagnosis model of the edge computing node through the reinforcement learning algorithm according to the deviation between the global diagnosis result and the local diagnosis result of the edge computing node, and sends the optimization parameters to the corresponding edge computing node to dynamically optimize the diagnostic performance of the edge computing node.

[0028] The present invention has the following beneficial effects:

[0029] The edge fusion diagnosis system and method for multi-source heterogeneous transmission monitoring data provided by the present invention solves the problem of data fragmentation through comprehensive access and standardized processing of multi-source heterogeneous data, provides a complete and reliable data foundation for diagnosis, and improves the comprehensiveness of data utilization; local diagnosis of edge computing nodes reduces data transmission delays, and combined with targeted intelligent diagnosis models, improves the real-time and accuracy of fault diagnosis; the edge node collaborative diagnosis mechanism compensates for the limitations of single-node diagnosis, uses blockchain to ensure data security and reliability, and improves the reliability and collaborative efficiency of diagnostic results; cloud-based global analysis realizes cross-regional fault correlation analysis and trend prediction, while dynamically optimizing edge models to form a continuously evolving diagnostic system. The overall solution comprehensively improves the real-time, accuracy, comprehensiveness, and reliability of transmission monitoring and diagnosis through the "edge-cloud" collaborative architecture, effectively meeting the needs of efficient monitoring of complex transmission networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 A schematic diagram of the composition of an edge fusion diagnostic system for multi-source heterogeneous power transmission monitoring data provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The technical solution of the present invention will be further described below with reference to the accompanying drawings and through specific implementation methods.

[0032] Among them, the drawings are only used for illustrative purposes and represent only schematic diagrams rather than actual pictures, and should not be understood as limiting this patent; in order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of the actual product; for those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0033] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "inside", "outside" and the like indicate an orientation or position relationship based on the orientation or position relationship shown in the drawings, it is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting this patent. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0034] In the description of the present invention, unless otherwise expressly specified or limited, when the term "connection" or the like appears to indicate a connection relationship between components, such term should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be internal communication between two components or an interaction between two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood in specific circumstances.

[0035] Example

[0036] This embodiment provides an edge fusion diagnosis system for multi-source heterogeneous power transmission monitoring data, such as Figure 1 As shown, it includes: multi-source heterogeneous data access module, data preprocessing and standardization module, edge computing node, edge node collaborative diagnosis module, data compression and feature extraction module and cloud fusion analysis platform, among which:

[0037] The multi-source heterogeneous data access module is used to access the multi-source heterogeneous data generated during the transmission line monitoring process. The multi-source heterogeneous data includes real-time monitoring data collected by sensors, transmission line image data obtained by image acquisition equipment, transmission equipment status data and environmental monitoring data;

[0038] The data preprocessing and standardization module is connected to the multi-source heterogeneous data access module to clean, denoise, and standardize multi-source heterogeneous data, convert data in different formats into a unified data format, and identify and correct outliers in the data based on deep learning algorithms;

[0039] The edge computing node is connected to the data preprocessing and standardization module and is deployed close to the transmission monitoring data source. It is used to perform local diagnostic analysis on the preprocessed and standardized data. The edge computing node has built-in intelligent diagnostic models for different transmission monitoring objects. The intelligent diagnostic models include a transmission line image fault recognition model based on a convolutional neural network, a sensor time series data anomaly diagnosis model based on a recurrent neural network, and an equipment status assessment model based on a random forest.

[0040] The edge node collaborative diagnosis module is connected to multiple edge computing nodes respectively. When the credibility of the local diagnosis result of a single edge computing node is lower than the preset threshold, it triggers multiple edge computing nodes to perform collaborative diagnosis, thereby improving the accuracy of the diagnosis result through data sharing and joint analysis.

[0041] The data compression and feature extraction module is connected to the edge computing node and is used to compress the diagnostic result data of the edge computing node and extract key feature data;

[0042] The cloud-based fusion analysis platform is connected to the data compression and feature extraction module to receive key feature data uploaded by the edge computing nodes, perform global fusion analysis in combination with historical data, realize trend prediction and tracing of transmission line faults, and feed back model optimization parameters to the edge computing nodes.

[0043] The edge fusion diagnosis system of multi-source heterogeneous transmission monitoring data adopting the above technical solution realizes comprehensive access to various types of transmission monitoring data by setting up a multi-source heterogeneous data access module, ensuring the integrity of the data; the data preprocessing and standardization module cleans, denoises, standardizes and corrects outliers on the data, improves the data quality, and provides a reliable data basis for subsequent diagnosis; the edge computing node performs diagnostic analysis locally, reduces the amount of data transmission and delay, and improves the real-time performance of diagnosis; the edge node collaborative diagnosis module triggers collaborative diagnosis when the reliability of single-node diagnosis is low, and improves the accuracy of diagnosis results; the data compression and feature extraction module reduces the amount of uploaded data, reduces transmission costs and pressure; the cloud-based fusion analysis platform combines historical data for global analysis, realizes fault trend prediction and tracing, and can optimize the edge node model, which improves the comprehensiveness, real-timeness and accuracy of transmission monitoring diagnosis as a whole.

[0044] Specifically, in this embodiment, the sensor types supported by the multi-source heterogeneous data access module include temperature sensors, humidity sensors, wind speed sensors, tilt sensors, tension sensors and partial discharge sensors; image acquisition devices include high-definition cameras, infrared thermal imagers and image acquisition devices carried by drones.

[0045] The above technical solution clarifies the specific sensor types and image acquisition devices supported by the multi-source heterogeneous data access module, making data access more targeted and practical, and enabling more comprehensive and accurate acquisition of various types of monitoring data related to transmission lines. It provides a richer and more practical data source for subsequent data processing and diagnostic analysis, further ensuring the reliability of diagnostic results.

[0046] Specifically, in this embodiment, the deep learning algorithm used in the data preprocessing and standardization module is an outlier detection algorithm based on an autoencoder. Normal data is learned by constructing an autoencoder model, and then outliers are identified by reconstructing errors, and outliers are corrected based on the distribution characteristics of normal data.

[0047] The above technical solution is adopted to identify and correct data outliers using an outlier detection algorithm based on an autoencoder. By learning the normal data distribution characteristics, the autoencoder can identify outliers more accurately, which improves the accuracy of outlier identification compared with traditional methods. The outliers are corrected based on the normal data distribution characteristics, which ensures the rationality of the corrected data, effectively improves the integrity and reliability of the data, and provides higher-quality data support for subsequent diagnostic analysis.

[0048] Specifically, in this embodiment, the edge node collaborative diagnosis module adopts a trusted data interaction mechanism based on blockchain. Each edge computing node acts as a node in the blockchain network. When sharing data, it records data interaction information through the distributed ledger of the blockchain to ensure the data's immutability and traceability. At the same time, smart contracts are used to realize the automatic allocation of collaborative diagnosis tasks and result verification.

[0049] Using the above technical solution, the edge node collaborative diagnosis module adopts a trusted data interaction mechanism based on blockchain. The distributed ledger of blockchain ensures the immutability and traceability of data interaction, and improves the security and credibility of data sharing; smart contracts realize the automatic allocation and result verification of collaborative diagnosis tasks, reduce manual intervention, improve the efficiency and standardization of collaborative diagnosis, and thus ensure the reliability of collaborative diagnosis results.

[0050] Specifically, in this embodiment, the cloud-based fusion analysis platform includes a data storage unit, a global diagnosis model unit and a model optimization unit; the data storage unit is used to store key feature data and historical monitoring data uploaded by the edge computing node; the global diagnosis model unit adopts a fusion diagnosis model based on a graph neural network to integrate the topological structure of the transmission line into the diagnosis and analysis process, and realize the correlation analysis of cross-regional and multi-type faults; the model optimization unit generates the optimization parameters of the built-in intelligent diagnosis model of the edge computing node through a reinforcement learning algorithm based on the deviation between the global diagnosis results and the local diagnosis results of the edge computing node.

[0051] By adopting the above technical solution, the data storage unit of the cloud-based fusion analysis platform realizes the effective storage of key feature data and historical data, providing data support for global analysis; the global diagnosis model unit adopts a fusion diagnosis model based on graph neural network, integrating the transmission line topology structure, which can more comprehensively analyze the correlation between cross-regional and multi-type faults, and improve the accuracy and comprehensiveness of global diagnosis; the model optimization unit generates optimization parameters through reinforcement learning algorithm and feeds them back to the edge node, realizing the dynamic optimization of the edge node intelligent diagnosis model and continuously improving the diagnostic performance of the edge node.

[0052] This embodiment also provides an edge fusion diagnosis method for multi-source heterogeneous power transmission monitoring data, including the following steps:

[0053] S1. Multi-source heterogeneous data access: The multi-source heterogeneous data access module collects multi-source heterogeneous data during the transmission line monitoring process. The multi-source heterogeneous data includes real-time monitoring data collected by sensors, transmission line image data obtained by image acquisition equipment, transmission equipment status data, and environmental monitoring data.

[0054] S2. Data Preprocessing and Standardization: The data preprocessing and standardization module processes the multi-source heterogeneous data received in step S1. First, it cleans and denoises the data to remove redundant data and noise interference. Then, it converts the data in different formats into a unified format. Finally, it uses a deep learning algorithm to identify and correct outliers in the data.

[0055] S3. Local diagnosis at the edge node: The standardized data preprocessed in step S2 is transmitted to the edge computing node, which calls the built-in intelligent diagnostic model for different transmission monitoring objects to perform local diagnostic analysis and obtain preliminary diagnostic results;

[0056] S4. Collaborative diagnosis at the edge node: The edge node collaborative diagnosis module determines the credibility of the preliminary diagnosis result obtained in step S3. If the credibility is lower than the preset threshold, multiple edge computing nodes are triggered to perform collaborative diagnosis, and more accurate diagnosis results are obtained through data sharing and joint analysis.

[0057] S5 data compression and upload: using the data compression and feature extraction module to compress the diagnostic data obtained in step S3 or step S4, extract key feature data, and upload the key feature data to the cloud fusion analysis platform;

[0058] S6. Cloud-based fusion analysis and model optimization: The cloud-based fusion analysis platform receives the key feature data uploaded in step S5 and performs a global fusion analysis based on historical data to predict and trace the trend of transmission line faults. Based on the analysis results, it generates optimization parameters for the built-in intelligent diagnosis model of the edge computing node and feeds them back to the edge computing node to optimize its diagnostic performance.

[0059] With the above technical solution, the multi-source heterogeneous data access step comprehensively collects various monitoring data, ensuring the comprehensiveness of the data; the data preprocessing and standardization step cleans, denoises, unifies the format, and corrects outliers on the data, improving data quality; the edge node local diagnosis step realizes local rapid processing of data, reduces transmission delay, and improves the timeliness of diagnosis; the edge node collaborative diagnosis step improves the reliability of the diagnosis results through multi-node joint analysis when necessary; the data compression and upload step reduces the amount of data transmitted and reduces the transmission cost; the cloud-based fusion analysis and model optimization step realizes global analysis and model optimization, improves the fault prediction and tracing capabilities and edge node diagnostic performance, and the overall process improves the efficiency and accuracy of transmission monitoring and diagnosis.

[0060] Specifically, in this embodiment, the specific process of identifying and correcting outliers in the data based on the deep learning algorithm in step S2 is: constructing an outlier detection model based on an autoencoder, inputting normal multi-source heterogeneous data into the autoencoder for training, so that the autoencoder learns the distribution characteristics of normal data; inputting the data to be processed into the trained autoencoder, calculating the reconstruction error of the data, and when the reconstruction error is greater than a preset threshold, determining that the data is an outlier; and correcting the outlier using the interpolation method based on the distribution characteristics of the normal data.

[0061] The above technical solution clarifies the specific process of outlier identification and correction based on the autoencoder. After training, the autoencoder can accurately capture the characteristics of normal data, judge outliers by reconstructing errors, and improve the accuracy of outlier identification. The interpolation method is used to correct outliers based on the distribution characteristics of normal data, so that the corrected data is more in line with the actual data rules, ensuring the validity of the data, providing more reliable data for subsequent diagnostic analysis, and further improving the accuracy of the diagnostic results.

[0062] Specifically, in this embodiment, the working process of the intelligent diagnosis model called by the edge computing node in step S3 includes: for transmission line image data, a transmission line image fault recognition model based on a convolutional neural network is adopted to extract image features through the convolution layer, and fault classification and identification is performed through the fully connected layer; for sensor time series data, a sensor time series data anomaly diagnosis model based on a recurrent neural network is adopted to capture the time series dependency of the data and identify abnormal fluctuations in the data; for equipment status data, an equipment status assessment model based on a random forest is adopted to assess the health status of the equipment through multi-dimensional feature analysis.

[0063] By adopting the above technical solution, corresponding intelligent diagnosis models are called for different types of data. The image fault recognition model based on convolutional neural network can effectively extract image features, improving the accuracy of transmission line image fault recognition; the sensor time series data anomaly diagnosis model based on recurrent neural network can capture the data time series dependency, improving the accuracy of time series data anomaly diagnosis; the equipment status assessment model based on random forest improves the reliability of equipment status assessment through multi-dimensional analysis. The targeted application of various models improves the overall effect of local diagnosis of edge nodes.

[0064] Specifically, in this embodiment, the specific process of edge node collaborative diagnosis in step S4 is: when the credibility of the preliminary diagnosis result of a single edge computing node is lower than the preset threshold, the edge computing node sends a collaborative diagnosis request to the edge node collaborative diagnosis module; the edge node collaborative diagnosis module selects other edge computing nodes associated with the monitoring area of ​​the requesting node to form a collaborative diagnosis group based on the topological structure of the transmission line and the monitoring range of each edge computing node; each edge computing node in the collaborative diagnosis group shares relevant monitoring data and diagnostic intermediate results through a trusted data interaction mechanism based on blockchain; each edge computing node conducts joint analysis based on the shared data and determines the final collaborative diagnosis result through a voting mechanism.

[0065] The above technical solution clarifies the specific process of collaborative diagnosis of edge nodes, triggers collaborative diagnosis by judging credibility, and ensures the necessity of collaborative diagnosis. The selection of related nodes to form collaborative groups based on the transmission line topology and monitoring range makes collaborative diagnosis more targeted. The blockchain mechanism ensures the security and reliability of data sharing. The voting mechanism determines the final result, which improves the fairness and accuracy of the collaborative diagnosis results and overall enhances the effectiveness and reliability of collaborative diagnosis.

[0066] Specifically, in this embodiment, the specific process of the global fusion analysis performed by the cloud-based fusion analysis platform in step S6 is as follows: the global diagnosis model unit of the cloud-based fusion analysis platform adopts a fusion diagnosis model based on a graph neural network, takes the poles, conductors, insulators and other equipment of the transmission line as graph nodes, and takes the connection relationship between the equipment as the edge to construct a transmission line topology map; the key feature data uploaded by each edge computing node is used as the attribute feature of the graph node, and the topology map is subjected to feature learning and fusion through the graph neural network to realize the correlation analysis of cross-regional and multi-type faults and obtain a global diagnosis result; the model optimization unit generates the optimization parameters of the built-in intelligent diagnosis model of the edge computing node through the reinforcement learning algorithm according to the deviation between the global diagnosis result and the local diagnosis result of the edge computing node, and sends the optimization parameters to the corresponding edge computing node to dynamically optimize the diagnostic performance of the edge computing node.

[0067] By adopting the above technical solution, the specific process of global fusion analysis in the cloud is clarified, a transmission line topology map is constructed, and key feature data is used as node attributes, so that the graph neural network can better learn the relationship between devices and improve the accuracy of cross-regional and multi-type fault correlation analysis; through the reinforcement learning algorithm, optimization parameters are generated according to the deviation of the diagnostic results, and dynamic optimization of the edge node model is achieved, so that the edge node can adapt to different monitoring scenarios, continuously improve the diagnostic performance, and thus improve the diagnostic accuracy and adaptability of the entire system.

[0068] In summary, the edge fusion diagnosis system and method for multi-source heterogeneous power transmission monitoring data provided by this embodiment has the following advantages:

[0069] Through comprehensive access and standardized processing of multi-source heterogeneous data, data fragmentation is resolved, providing a complete and reliable data foundation for diagnosis and enhancing the comprehensiveness of data utilization. Local diagnosis at edge computing nodes reduces data transmission latency and, combined with targeted intelligent diagnostic models, improves the real-time and accuracy of fault diagnosis. The collaborative diagnosis mechanism at the edge node overcomes the limitations of single-node diagnosis, leveraging blockchain to ensure data security and reliability, and enhances the reliability and collaborative efficiency of diagnostic results. Global cloud-based analysis enables cross-regional fault correlation analysis and trend prediction, while dynamically optimizing edge models to form a continuously evolving diagnostic system. Through this collaborative "edge-cloud" architecture, the overall solution comprehensively enhances the real-time, accuracy, comprehensiveness, and reliability of transmission monitoring and diagnosis, effectively meeting the needs for efficient monitoring of complex transmission networks.

[0070] How it works

[0071] This system and method realizes the fusion diagnosis of multi-source heterogeneous power transmission monitoring data based on the "edge-cloud" collaborative architecture. The multi-source heterogeneous data access module aggregates various types of information such as real-time sensor data and image data. The data preprocessing and standardization module unifies the format and corrects anomalies to provide high-quality data for diagnosis. The edge computing node uses the built-in targeted intelligent diagnosis model to quickly analyze data locally to improve real-time performance; when the credibility is insufficient, the edge node collaborative diagnosis module realizes multi-node data sharing and joint analysis through the blockchain mechanism to enhance diagnostic accuracy. The data compression and feature extraction module reduces the amount of uploaded data. The cloud-based fusion analysis platform combines historical data and uses models such as graph neural networks to perform global analysis to achieve fault tracing and trend prediction. It also optimizes the edge node model through reinforcement learning to form a closed-loop working mechanism of "data access-preprocessing-edge diagnosis-collaborative optimization-cloud-based global analysis".

[0072] How to use

[0073] When in use, the corresponding sensors and image acquisition equipment are first deployed through the multi-source heterogeneous data access module to continuously access various monitoring data of the transmission line; the data preprocessing and standardization module is started to automatically complete data cleaning, denoising, format unification and outlier correction; the edge computing node automatically calls the corresponding intelligent diagnosis model to perform local diagnosis on the processed data and generate preliminary results; when the credibility of the preliminary results is insufficient, the system automatically triggers collaborative diagnosis of the edge nodes, and each node shares data through the blockchain and conducts joint analysis; the processed diagnostic results are compressed and feature extracted and uploaded to the cloud, where the cloud fusion analysis platform performs global analysis, generates fault trend predictions and model optimization parameters, and feeds back to the edge nodes to optimize subsequent diagnosis. The entire process does not require manual intervention in key diagnostic links, only regular maintenance of equipment and updating of system parameters are required.

[0074] The above are only preferred embodiments of the present invention and do not limit the implementation mode and protection scope of the present invention. For those skilled in the art, it should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the description and illustrations of the present invention should be included in the protection scope of the present invention.

Claims

1. An edge fusion diagnosis system for multi-source heterogeneous power transmission monitoring data, characterized in that: include: A multi-source heterogeneous data access module is used to access multi-source heterogeneous data generated during the transmission line monitoring process. The multi-source heterogeneous data includes real-time monitoring data collected by sensors, transmission line image data obtained by image acquisition equipment, transmission equipment status data, and environmental monitoring data; A data preprocessing and standardization module, connected to the multi-source heterogeneous data access module, is used to clean, denoise and standardize the multi-source heterogeneous data, convert data in different formats into a unified data format, and identify and correct outliers in the data based on a deep learning algorithm; An edge computing node, connected to the data preprocessing and standardization module, is deployed near the power transmission monitoring data source and is used to perform local diagnostic analysis on the preprocessed standardized data. The edge computing node has built-in intelligent diagnostic models for different power transmission monitoring objects, including a power transmission line image fault recognition model based on a convolutional neural network, a sensor time series data anomaly diagnosis model based on a recurrent neural network, and an equipment status assessment model based on a random forest. An edge node collaborative diagnosis module is connected to the plurality of edge computing nodes, and is used to trigger the plurality of edge computing nodes to perform collaborative diagnosis when the credibility of the local diagnosis result of a single edge computing node is lower than a preset threshold, thereby improving the accuracy of the diagnosis result through data sharing and joint analysis; A data compression and feature extraction module, connected to the edge computing node, is used to compress the diagnostic result data of the edge computing node and extract key feature data; The cloud-based fusion analysis platform is connected to the data compression and feature extraction module, and is used to receive key feature data uploaded by the edge computing node, perform global fusion analysis in combination with historical data, realize trend prediction and tracing of transmission line faults, and feed back model optimization parameters to the edge computing node.

2. The edge fusion diagnosis system for multi-source heterogeneous power transmission monitoring data according to claim 1 is characterized in that: The sensor types supported by the multi-source heterogeneous data access module include temperature sensors, humidity sensors, wind speed sensors, tilt sensors, tension sensors and partial discharge sensors; the image acquisition equipment includes high-definition cameras, infrared thermal imagers and image acquisition devices carried by drones.

3. The edge fusion diagnosis system for multi-source heterogeneous power transmission monitoring data according to claim 1 is characterized in that: The data preprocessing and standardization module uses a deep learning algorithm that is an outlier detection algorithm based on an autoencoder. It learns normal data by building an autoencoder model, identifies outliers through reconstruction errors, and corrects outliers based on the distribution characteristics of normal data.

4. The edge fusion diagnosis system for multi-source heterogeneous power transmission monitoring data according to claim 1 is characterized in that: The edge node collaborative diagnosis module adopts a trusted data interaction mechanism based on blockchain. Each edge computing node acts as a node in the blockchain network. When sharing data, it records data interaction information through the distributed ledger of the blockchain to ensure the data's immutability and traceability. At the same time, smart contracts are used to realize the automatic allocation of collaborative diagnosis tasks and result verification.

5. The edge fusion diagnosis system for multi-source heterogeneous power transmission monitoring data according to claim 1 is characterized in that: The cloud-based fusion analysis platform includes a data storage unit, a global diagnostic model unit and a model optimization unit; the data storage unit is used to store key feature data and historical monitoring data uploaded by edge computing nodes; the global diagnostic model unit adopts a fusion diagnostic model based on graph neural networks to integrate the topological structure of the transmission line into the diagnostic analysis process, thereby realizing correlation analysis of cross-regional and multi-type faults; the model optimization unit generates optimization parameters of the built-in intelligent diagnostic model of the edge computing node through a reinforcement learning algorithm based on the deviation between the global diagnostic results and the local diagnostic results of the edge computing node.

6. A method for edge fusion diagnosis of multi-source heterogeneous power transmission monitoring data based on the edge fusion diagnosis system of multi-source heterogeneous power transmission monitoring data according to any one of claims 1 to 5, characterized in that: The following steps are involved: S1. Multi-source heterogeneous data access: The multi-source heterogeneous data access module collects multi-source heterogeneous data during the transmission line monitoring process. The multi-source heterogeneous data includes real-time monitoring data collected by sensors, transmission line image data obtained by image acquisition equipment, transmission equipment status data, and environmental monitoring data. S2. Data Preprocessing and Standardization: The data preprocessing and standardization module processes the multi-source heterogeneous data received in step S1. First, it cleans and denoises the data to remove redundant data and noise interference. Then, it converts the data in different formats into a unified format. Finally, it uses a deep learning algorithm to identify and correct outliers in the data. S3. Local diagnosis at the edge node: The standardized data preprocessed in step S2 is transmitted to the edge computing node, which calls the built-in intelligent diagnostic model for different transmission monitoring objects to perform local diagnostic analysis and obtain preliminary diagnostic results; S4. Collaborative diagnosis at the edge node: The edge node collaborative diagnosis module determines the credibility of the preliminary diagnosis result obtained in step S3. If the credibility is lower than the preset threshold, multiple edge computing nodes are triggered to perform collaborative diagnosis, and more accurate diagnosis results are obtained through data sharing and joint analysis. S5 data compression and upload: using the data compression and feature extraction module to compress the diagnostic data obtained in step S3 or step S4, extract key feature data, and upload the key feature data to the cloud fusion analysis platform; S6. Cloud-based fusion analysis and model optimization: The cloud-based fusion analysis platform receives the key feature data uploaded in step S5 and performs a global fusion analysis based on historical data to predict and trace the trend of transmission line faults. Based on the analysis results, it generates optimization parameters for the built-in intelligent diagnosis model of the edge computing node and feeds them back to the edge computing node to optimize its diagnostic performance.

7. The edge fusion diagnosis method for multi-source heterogeneous power transmission monitoring data according to claim 6, characterized in that: The specific process of identifying and correcting outliers in the data based on the deep learning algorithm in step S2 is as follows: constructing an outlier detection model based on an autoencoder, inputting normal multi-source heterogeneous data into the autoencoder for training, and allowing the autoencoder to learn the distribution characteristics of normal data; inputting the data to be processed into the trained autoencoder, calculating the reconstruction error of the data, and when the reconstruction error is greater than a preset threshold, determining that the data is an outlier; and correcting the outlier using the interpolation method based on the distribution characteristics of normal data.

8. The edge fusion diagnosis method for multi-source heterogeneous power transmission monitoring data according to claim 6, characterized in that: The working process of the intelligent diagnosis model called by the edge computing node in step S3 includes: for transmission line image data, a transmission line image fault recognition model based on a convolutional neural network is adopted to extract image features through the convolution layer and perform fault classification and identification through the fully connected layer; for sensor time series data, a sensor time series data anomaly diagnosis model based on a recurrent neural network is adopted to capture the time series dependency of the data and identify abnormal fluctuations in the data; for equipment status data, an equipment status assessment model based on a random forest is adopted to assess the health status of the equipment through multi-dimensional feature analysis.

9. The edge fusion diagnosis method for multi-source heterogeneous power transmission monitoring data according to claim 6, characterized in that: The specific process of edge node collaborative diagnosis in step S4 is as follows: when the credibility of the preliminary diagnosis result of a single edge computing node is lower than the preset threshold, the edge computing node sends a collaborative diagnosis request to the edge node collaborative diagnosis module; the edge node collaborative diagnosis module selects other edge computing nodes associated with the monitoring area of ​​the requesting node to form a collaborative diagnosis group based on the topological structure of the transmission line and the monitoring range of each edge computing node; each edge computing node in the collaborative diagnosis group shares relevant monitoring data and diagnostic intermediate results through a trusted data interaction mechanism based on blockchain; each edge computing node conducts joint analysis based on the shared data and determines the final collaborative diagnosis result through a voting mechanism.

10. The edge fusion diagnosis method for multi-source heterogeneous power transmission monitoring data according to claim 6, characterized in that: The specific process of the global fusion analysis performed by the cloud-based fusion analysis platform in step S6 is as follows: the global diagnosis model unit of the cloud-based fusion analysis platform adopts a fusion diagnosis model based on a graph neural network, takes the transmission line towers, conductors, insulators and other equipment as graph nodes, and the connection relationships between the equipment as edges to construct a transmission line topology map; the key feature data uploaded by each edge computing node is used as the attribute features of the graph node, and the topology map is learned and integrated through the graph neural network to realize the correlation analysis of cross-regional and multi-type faults and obtain a global diagnosis result; Based on the deviation between the global diagnostic results and the local diagnostic results of the edge computing node, the model optimization unit generates the optimization parameters of the built-in intelligent diagnostic model of the edge computing node through the reinforcement learning algorithm, and sends the optimization parameters to the corresponding edge computing node to dynamically optimize the diagnostic performance of the edge computing node.

Images