Fault remote diagnosis method and system for power grid operation and maintenance

Abstract

The invention provides a remote fault diagnosis method and system for operation and maintenance of a power grid. The method belongs to the technical field of power system operation and maintenance, and comprises the following steps: collecting operation state data of power grid equipment in real time through a sensor network, and carrying out real-time processing and preliminary analysis on the operation state data through an edge computing node; uploading the preliminary diagnosis result of the edge computing node and the key sensor data to a block chain platform, and storing and verifying the preliminary diagnosis result and the key sensor data through a distributed account book of a block chain; the operation state data of the power grid equipment are collected in real time, and the edge computing nodes are used for preliminary analysis, so that potential fault signals can be quickly found.

Description

Technical Field

[0001] The present invention provides a remote fault diagnosis method and system for power grid operation and inspection, belonging to the technical field of power system operation and maintenance. Background Art

[0002] With the accelerated pace of industrialization and informatization, the safe, reliable, and efficient operation of power grids, the infrastructure that supports the national economic lifeline and social stability, has become increasingly crucial. However, as power grids continue to expand in scale and complexity, they face increasing challenges, including but not limited to aging equipment, harsh environmental impacts, load fluctuations, and sudden failures. These challenges place higher demands on power grid operation and maintenance, particularly in fault detection, diagnosis, and emergency response.

[0003] Traditional power grid operation and maintenance models rely primarily on manual inspections and post-event processing, resulting in slow response times, inaccurate fault location, and high maintenance costs. Furthermore, due to the widespread distribution of power grid equipment and the vast amount of operational data, traditional centralized processing not only increases the burden on data transmission and processing but also hinders real-time monitoring and early warning. Therefore, improving the intelligence of power grid operation and maintenance, and enabling remote diagnosis and rapid fault resolution, has become a pressing issue. Summary of the Invention

[0004] The present invention provides a method and system for remote fault diagnosis for power grid operation and inspection, which is used to solve the problems mentioned in the above background technology:

[0005] The present invention proposes a remote fault diagnosis method for power grid operation and inspection, the method comprising:

[0006] S1. Collect the operating status data of power grid equipment in real time through the sensor network, and perform real-time processing and preliminary analysis of the operating status data through the edge computing node;

[0007] S2. Upload the preliminary diagnostic results and key sensor data of the edge computing node to the blockchain platform, and store and verify the preliminary diagnostic results and key sensor data through the blockchain's distributed ledger;

[0008] S3: Build a data set, use it to train deep learning models, interact with the blockchain platform, and receive data from edge nodes for deeper analysis and diagnosis.

[0009] S4. Information is transmitted through a real-time data communication mechanism. When an edge computing node detects an abnormal signal or potential fault, the relevant information is uploaded to the blockchain platform for verification and sharing, and a deep learning model is triggered for further analysis and diagnosis.

[0010] S5. Based on the diagnostic results of the deep learning model, remote decision support is obtained, and fault handling is performed based on the decision support information. After the fault handling is completed, the processing results are recorded and uploaded to the blockchain platform to form a complete fault handling file.

[0011] Furthermore, the S1 includes:

[0012] S11. Determine the deployment location of edge computing nodes through an optimization algorithm based on the grid topology, equipment density, and fault history data.

[0013] S12. Integrate multiple types of sensors to form a comprehensive monitoring network, and monitor the status of power grid equipment through the comprehensive monitoring network;

[0014] S13. Based on the Internet of Things, sensor data is collected in real time and transmitted to the edge computing node;

[0015] S14, the edge computing node performs real-time processing and preliminary analysis on the received sensor data;

[0016] S15. Based on a preset threshold, perform preliminary fault diagnosis on the data after preliminary analysis to identify possible abnormalities or fault signs.

[0017] Furthermore, the S15 includes:

[0018] Utilize historical fault data and current operating data, combined with machine learning algorithms, to dynamically adjust the thresholds of various monitoring parameters;

[0019] Through the threshold adaptive learning mechanism, the threshold adjustment effect is regularly evaluated, and the algorithm parameters are adjusted based on the evaluation results to form a closed-loop optimization system;

[0020] Integrate data fusion algorithms on edge computing nodes to fuse multi-source data from different sensors and extract comprehensive feature vectors that reflect the health status of the equipment;

[0021] Use the fused feature vectors for pattern recognition to identify the characteristic patterns of the equipment under different operating conditions;

[0022] Based on the anomaly detection algorithm, the fused data is subjected to real-time anomaly detection to identify data points that deviate from the normal operating mode as potential fault signals;

[0023] For the detected abnormal patterns, cluster analysis or case-based reasoning methods are used to match the abnormal patterns with known fault types to preliminarily predict the possible fault types and their severity.

[0024] Furthermore, the threshold adaptive learning mechanism is used to regularly evaluate the effect of threshold adjustment, and the algorithm parameters are adjusted based on the evaluation results to form a closed-loop optimization system, including:

[0025] Establish a data feedback channel to transmit the actual thresholds and diagnostic results used by edge computing nodes during preliminary fault diagnosis to the central control system in real time or periodically;

[0026] In the central control system, the threshold usage effect data from each edge computing node is processed through the data processing and analysis module, and a data set for threshold optimization is constructed;

[0027] Using reinforcement learning algorithms, combined with the operating characteristics and failure modes of power grid equipment, a threshold optimization strategy is designed;

[0028] Utilize historical fault data and real-time feedback data to train and iterate the reinforcement learning model, continuously optimizing the parameters of the threshold adjustment strategy;

[0029] Regularly evaluate the effectiveness of the optimized thresholds, and automatically adjust the parameters of the reinforcement learning model based on the evaluation results to form a closed-loop system for continuous optimization; any threshold adjustment strategies that perform poorly should be corrected or replaced in a timely manner.

[0030] Furthermore, the S2 includes:

[0031] S21. Encrypt the preliminary diagnostic results and key sensor data using an encryption algorithm and transmit the encrypted data to the blockchain platform.

[0032] S22, based on smart contracts, is automatically triggered when edge computing nodes upload data to execute data verification and storage processes;

[0033] S23. Through the decentralized nature of blockchain, preliminary diagnostic results and key data are stored in multiple nodes, and the uploaded data is verified through the consensus mechanism of blockchain.

[0034] Furthermore, the S23 includes:

[0035] Evaluate the sensitivity of preliminary diagnostic results and key sensor data;

[0036] The preliminary diagnostic results and key sensor data are compressed using compression algorithms; hash functions are used to intelligently segment data based on data size, type, and access frequency;

[0037] The compressed data is fragmented and stored on multiple nodes in the blockchain network. Through an automated data verification process, when new data is uploaded to the blockchain, the integrity and authenticity of the data are verified through a consensus mechanism and recorded on the blockchain to form an unalterable chain of evidence.

[0038] Regularly conduct security audits on blockchain networks to check for potential security vulnerabilities and threats, and build data backup mechanisms.

[0039] Furthermore, the S3 includes:

[0040] S31. Collect data through a comprehensive data integration platform and use streaming data processing technology to process real-time data streams;

[0041] S32. Establish a data warehouse, use a distributed storage system to store historical data, and perform a pre-processing module on the stored data;

[0042] S33. Conduct preliminary exploratory analysis of data using statistical analysis and data mining techniques to discover hidden patterns and anomalies;

[0043] S34. Build and train a machine learning prediction model based on grid operation data. Integrate a real-time data feedback mechanism to continuously monitor model performance and update and optimize the model based on new data.

[0044] S35. Use the intelligent diagnostic engine, combined with the output of the machine learning prediction model and the expert knowledge base, to diagnose and locate faults.

[0045] S36. Automatically generate a treatment plan based on the diagnosis results through a decision support system.

[0046] Furthermore, the S4 includes:

[0047] S41. Edge computing nodes continuously monitor the operating status of power grid equipment, capture sensor data in real time, and perform preliminary processing. When an abnormal signal or potential fault is detected by the edge computing node, it is immediately marked and prepared to upload the relevant data.

[0048] S42. Upload preliminary diagnostic results and key sensor data to the blockchain platform through a secure data transmission protocol;

[0049] S43. The blockchain platform verifies the received data, stores it in the distributed ledger, and shares the verified data with all authorized nodes.

[0050] S44. When the blockchain platform detects newly uploaded abnormal data, it automatically triggers the deep learning model in the data center for further analysis;

[0051] S45. The deep learning model obtains data from the blockchain platform, performs preprocessing, and uses the trained deep learning model to conduct in-depth analysis of the processed data;

[0052] S46. Generate a detailed diagnostic report based on the analysis results. The deep learning model returns the diagnostic results to the edge computing node through a secure channel and displays them to the operation and maintenance personnel on the user interface.

[0053] Furthermore, the S5 includes:

[0054] S51. Operations and maintenance personnel view the deep learning model's diagnostic results and decision support information through the user interface and evaluate the feasibility of the fault handling solution and the rationality of the resource allocation suggestion.

[0055] S52. Execute fault handling operations based on the decision support information and adjust the power grid operation strategy based on the impact of the fault;

[0056] S53. During the fault handling process, continuously monitor the operating status of the power grid equipment and provide real-time feedback on the handling effect;

[0057] S54. Record the entire process of troubleshooting, organize the results into documents or reports, upload them through the blockchain platform, and form a complete troubleshooting file together with the preliminary diagnosis results and diagnostic report.

[0058] The present invention proposes a remote fault diagnosis system for power grid operation and inspection, comprising a memory, a processor, and a computer program stored in and executable on the memory. The processor executes the program to implement any of the above-described remote fault diagnosis methods for power grid operation and inspection.

[0059] Beneficial effects of the present invention: By collecting the operating status data of power grid equipment in real time and performing preliminary analysis using edge computing nodes, potential fault signals can be quickly discovered. At the same time, by combining deep learning models for more in-depth analysis and diagnosis, the accuracy and efficiency of fault diagnosis are further improved; by uploading preliminary diagnostic results and key sensor data to the blockchain platform, and using the distributed ledger of the blockchain to store and verify the data, the security, reliability and non-tamperability of the data are ensured. In addition, the data verification and storage process is automatically triggered based on smart contracts, which improves the degree of automation of data processing; based on the diagnostic results of the deep learning model, remote decision support is obtained, and fault processing is performed based on the decision support information. At the same time, information is transmitted through a real-time data communication mechanism, so that operation and maintenance personnel can understand the fault situation in a timely manner and take corresponding measures; by performing real-time monitoring and fault diagnosis of power grid equipment, potential faults are discovered and handled in a timely manner, the equipment failure rate is reduced, and the reliability and stability of power grid equipment are improved; through remote fault diagnosis and decision support systems, the frequency and cost of on-site inspections and maintenance are reduced, and the operation and maintenance efficiency is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is a step diagram of the method of the present invention. DETAILED DESCRIPTION

[0061] In order to more clearly understand the above-mentioned objects, features and advantages of the present invention, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, in the absence of conflict, the embodiments of the present application and the features therein may be combined with each other.

[0062] The following description sets forth numerous specific details to facilitate a thorough understanding of the present invention. The embodiments described are merely a portion of the embodiments of the present invention, not all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort are intended to fall within the scope of protection of the present invention.

[0063] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0064] One embodiment of the present invention, as Figure 1 As shown, a method for remote fault diagnosis for power grid operation and inspection includes:

[0065] S1. Deploy edge computing nodes at key nodes of the power grid (such as substations and transmission lines), connect the edge computing nodes to the sensor network, collect the operating status data of power grid equipment in real time through the sensor network, and perform real-time processing and preliminary analysis of the sensor data through the edge computing nodes;

[0066] S2. Upload the preliminary diagnostic results and key sensor data of the edge computing node to the blockchain platform, and store and verify the preliminary diagnostic results and key sensor data through the blockchain's distributed ledger;

[0067] S3. Build a dataset based on historical fault data, normal operation data, and preliminary diagnostic results from edge computing nodes. Use the dataset to train a deep learning model. The trained deep learning model is deployed to the data center, interacting with the blockchain platform and receiving data from edge nodes for deeper analysis and diagnosis.

[0068] S4. Data, diagnostic results, and decision-making instructions are transmitted between edge computing nodes, blockchain platforms, and deep learning models through a real-time data communication mechanism. When an edge computing node detects an abnormal signal or potential fault, it immediately uploads the relevant information to the blockchain platform for verification and sharing, and triggers the deep learning model for further analysis. The deep learning model performs intelligent diagnosis based on the received data and returns the diagnosis results to the edge computing node and operation and maintenance personnel.

[0069] S5. Based on the diagnostic results of the deep learning model, operation and maintenance personnel obtain remote decision support through the user interface, including fault handling solutions, resource allocation suggestions, etc.; operation and maintenance personnel perform fault handling or adjust the power grid operation strategy according to the decision support information to prevent the fault from expanding; after the fault handling is completed, the processing results are recorded and uploaded to the blockchain platform to form a complete fault handling file.

[0070] The working principle of the above technical solution is as follows: Edge computing nodes are deployed at key nodes in the power grid (such as substations and transmission lines). These nodes serve as the front end for data processing and are directly connected to the sensor network. The sensor network collects real-time operating status data of power grid equipment, such as key parameters such as current, voltage, temperature, and vibration. Upon receiving this data, the edge computing nodes immediately perform preliminary processing and analysis, such as data cleansing, outlier detection, and simple fault diagnosis. This step aims to reduce data transmission volume, improve data processing efficiency, and preliminarily screen out potential faults or abnormal signals. The edge computing nodes upload the preliminary diagnostic results and key sensor data to the blockchain platform. The blockchain platform utilizes its distributed ledger technology to ensure data immutability and traceability. Each data block contains information such as a timestamp, data content, and a hash value. Consensus is reached within the blockchain network through a consensus mechanism (such as Proof-of-Work or Proof-of-Stake), enabling data storage and verification. This step enhances data security and credibility, providing a reliable data source for subsequent data analysis and fault diagnosis. A comprehensive dataset is constructed based on historical fault data, normal operation data, and the preliminary diagnostic results of the edge computing nodes. This dataset contains operational data from power grid equipment under various conditions, providing a rich set of samples for training deep learning models. Deep learning algorithms (such as convolutional neural networks and recurrent neural networks) are used to train the dataset, building an intelligent diagnostic model capable of accurately identifying power grid equipment faults. The trained model is deployed in the data center, interacting with the blockchain platform in real time. Upon receiving data from edge nodes, the deep learning model can rapidly perform more in-depth analysis and diagnosis, improving the accuracy and efficiency of fault diagnosis. Real-time data communication mechanisms (such as MQTT and WebSocket) enable rapid transmission of data, diagnostic results, and decision-making instructions between the edge computing nodes, the blockchain platform, and the deep learning model. When an edge computing node detects an abnormal signal or potential fault, it immediately uploads the relevant information to the blockchain platform for verification and sharing. After ensuring the authenticity and integrity of the data, the blockchain platform triggers the deep learning model for further analysis. Based on the received data, the deep learning model leverages its powerful learning and reasoning capabilities to perform intelligent diagnosis and generate detailed diagnostic reports and decision recommendations. These diagnostic results and decision-making instructions are returned in real time to the edge computing nodes and operations personnel, enabling them to take timely countermeasures. Operations and maintenance personnel access deep learning model diagnostic results and remote decision-making support information through user interfaces (such as web interfaces and mobile apps). This information includes the specific location, nature, and severity of the fault, as well as corresponding treatment plans and resource allocation recommendations. Based on this information, operations and maintenance personnel can quickly develop a fault handling plan and conduct fault resolution remotely or through on-site operations. Furthermore, they can adjust grid operation strategies based on the diagnostic results to prevent the fault from escalating or recurring.After troubleshooting is complete, operations personnel record and upload the results to the blockchain platform, creating a complete troubleshooting archive. This archive not only provides valuable data support for subsequent fault analysis and prevention, but also enhances the transparency and traceability of power grid operations and management.

[0071] The above technical solution achieves the following: By deploying edge computing nodes at key grid nodes, real-time data collection and processing of grid equipment operating status data can be collected and processed, significantly improving fault diagnosis response speed. Preliminary analysis by edge computing nodes reduces data transmission and improves data processing efficiency. Leveraging the blockchain platform's distributed ledger technology, preliminary diagnostic results and key sensor data are stored and verified, ensuring data immutability and traceability. This enhances data security and reliability, providing a reliable data source for subsequent deep learning and intelligent diagnosis. Intelligent diagnostic methods based on deep learning models can be trained using large amounts of historical fault data and normal operation data, thereby improving diagnostic accuracy and precision. Deep learning models can handle complex nonlinear relationships, capturing hidden fault characteristics in the data and providing more accurate diagnostic results for operations and maintenance personnel. Through a user interface, operations and maintenance personnel can access the deep learning model's diagnostic results and remote decision-making support information, including fault resolution plans and resource allocation recommendations, in real time. This enables operations and maintenance personnel to respond to faults quickly, reducing troubleshooting time and minimizing the impact of the fault on grid operations. After fault resolution is completed, the results are recorded and uploaded to the blockchain platform, forming a complete fault resolution archive. These archives not only provide valuable data support for subsequent fault analysis and prevention, but also enhance the transparency and traceability of power grid operations and maintenance management. Operations and management can easily review historical fault handling records, summarize lessons learned, and continuously optimize operations and maintenance strategies. Real-time monitoring and intelligent diagnosis enable early detection of potential faults and proactive action to address them, preventing further losses from escalating. Furthermore, resource allocation based on recommended resource allocations can reduce unnecessary resource waste and lower operations and maintenance costs. This technical solution, integrating multiple technologies such as edge computing, blockchain, and deep learning, improves the robustness and reliability of the power grid operation and maintenance system. Even if some nodes or devices fail, redundant backups and data recovery mechanisms in other nodes or devices ensure system operation.

[0072] In one embodiment of the present invention, the S1 includes:

[0073] S11. Determine the optimal deployment location of edge computing nodes using an optimization algorithm (such as a genetic algorithm) based on the grid topology, device density, and fault history data.

[0074] Based on the data processing requirements of each node, customized high-performance, low-power edge computing hardware is selected, and necessary storage and communication modules are configured to ensure the real-time and reliability of data processing.

[0075] S12, integrates various types of sensors such as temperature, vibration, current, and voltage to form a comprehensive monitoring network, and monitors the status of power grid equipment through the comprehensive monitoring network;

[0076] S13. Based on the Internet of Things, sensor data is collected in real time and transmitted to the edge computing node;

[0077] S14. The edge computing node performs real-time processing and preliminary analysis of the received sensor data, including feature extraction and anomaly detection, to reduce the burden of data transmission.

[0078] S15. Based on a preset threshold, perform preliminary fault diagnosis on the data after preliminary analysis to identify possible abnormalities or fault signs.

[0079] The working principle of the above technical solution is as follows: a detailed analysis of the power grid topology is conducted to understand the grid's composition, the connections between nodes, and the paths of power flow. Based on the density of equipment in the grid and historical fault data, the fault detection and diagnosis needs of each region are assessed. An optimization algorithm (such as a genetic algorithm) comprehensively considers these factors to calculate the optimal deployment location of edge computing nodes to ensure efficient coverage of the entire grid and minimize data transmission latency. Based on the data processing requirements of each node, high-performance, low-power edge computing hardware is selected and equipped with the necessary storage and communication modules. This customized design aims to ensure real-time and reliable data processing while reducing energy consumption. A variety of sensors (such as temperature, vibration, current, and voltage) are integrated to form a comprehensive monitoring network. These sensors can comprehensively monitor the operating status of power grid equipment and capture various physical quantity changes that may indicate faults. Through the comprehensive monitoring network, real-time status monitoring of power grid equipment is carried out to ensure that any anomalies or fault signs are promptly detected. Based on the Internet of Things (IoT), the sensor network is tightly integrated with the edge computing nodes. IoT technology enables real-time collection of sensor data and immediate transmission to the edge computing nodes for processing. This real-time performance is crucial for quickly responding to power grid failures. After receiving sensor data, the edge computing node immediately performs real-time processing. This processing method reduces the bandwidth and time required to transmit data to the data center, improving the system's response speed. The processing process includes steps such as feature extraction and anomaly detection. Feature extraction helps extract information useful for fault diagnosis from the raw data. Anomaly detection is used to identify outliers or abnormal patterns in the data, which may indicate a failure of power grid equipment. Based on preset thresholds (which may be determined based on historical data, equipment specifications, or expert experience), the data after preliminary analysis is judged. If the data exceeds the preset threshold or meets other preset conditions, it is preliminarily diagnosed as a possible anomaly or failure. At this point, the edge computing node marks the relevant information as urgent or important and prepares to upload it to the blockchain platform for further verification and sharing.

[0080] The effect of the above technical solution is: by analyzing the grid topology, equipment density and fault history data, and using optimization algorithms (such as genetic algorithms) to determine the optimal deployment location of edge computing nodes, it can ensure the efficient use of edge computing resources, reduce unnecessary redundant deployment, and improve the coverage and response speed of fault detection. Customized selection of high-performance, low-power edge computing hardware and configuration of necessary storage and communication modules not only meet the data processing needs of each node, but also reduce energy consumption, extend the service life of the equipment, and further improve the economy and sustainability of the system; integrate various types of sensors (such as temperature, vibration, current, voltage, etc.) to form a comprehensive monitoring network, which can comprehensively and accurately monitor the operating status of power grid equipment and capture various possible precursors to faults; based on the real-time data acquisition and transmission mechanism of the Internet of Things, it ensures that sensor data can be quickly and accurately transmitted to the edge computing node for processing, greatly improving the real-time nature of fault detection; the edge computing node processes and responds to the received sensor data in real time. Initial analysis, including feature extraction and anomaly detection, effectively reduces the bandwidth and time required to transmit data to the data center, thus reducing the data transmission burden. The results of this initial analysis provide an important basis for subsequent in-depth diagnosis and provide operators with preliminary fault warning information, enabling them to take timely measures to prevent the occurrence or escalation of faults. Preliminary fault diagnosis based on preset thresholds on the analyzed data can quickly identify possible anomalies or fault signs, providing operators with timely fault information. Once an anomaly or fault sign is detected, the edge computing node can immediately upload the relevant information to the blockchain platform for verification and sharing, triggering a deep learning model for further analysis. This rapid response mechanism helps shorten fault handling time and reduce the impact of faults on grid operations. Through the implementation of these steps, the intelligent level of grid operation and maintenance has been significantly improved, allowing operators to more efficiently monitor and manage grid equipment, reducing the frequency and cost of manual inspections. Furthermore, the ability to promptly detect and address potential faults avoids further losses and repair costs caused by fault escalation, further reducing the overall cost of grid operation and maintenance.

[0081] In one embodiment of the present invention, the S15 includes:

[0082] By using historical fault data and current operation data, combined with machine learning algorithms (such as long short-term memory network LSTM), the thresholds of various monitoring parameters are dynamically adjusted.

[0083] Through the threshold adaptive learning mechanism, the threshold adjustment effect is regularly evaluated, and the algorithm parameters are adjusted based on the evaluation results to form a closed-loop optimization system;

[0084] Integrate data fusion algorithms (such as principal component analysis (PCA) or deep neural networks (DNN)) on edge computing nodes to fuse multi-source data such as temperature, vibration, current, and voltage from different sensors to extract comprehensive feature vectors that reflect the health status of the equipment.

[0085] The fused feature vectors are used for pattern recognition to identify the characteristic patterns of the equipment under different operating conditions, providing a richer information basis for subsequent fault diagnosis.

[0086] Based on anomaly detection algorithms (such as Isolation Forest or Support Vector Machine (SVM)), real-time anomaly detection is performed on the fused data to identify data points that deviate from the normal operating mode as potential fault signals;

[0087] For the detected abnormal patterns, cluster analysis or case-based reasoning methods are used to match the abnormal patterns with known fault types to preliminarily predict the possible fault types and their severity.

[0088] The working principle of the above technical solution is as follows: It utilizes historical fault data and real-time operating data of current power grid equipment, which captures the performance of the equipment under different environmental conditions and operating states. It also uses machine learning algorithms such as long short-term memory (LSTM) networks to dynamically adjust the thresholds of monitoring parameters. LSTMs can capture long-term dependencies in time series data, thereby adapting to performance fluctuations of power grid equipment caused by factors such as environmental changes and aging. The effectiveness of threshold adjustments is regularly evaluated, and algorithm parameters are adjusted based on the feedback from the evaluation results, forming a closed-loop optimization system. This allows the thresholds to continuously adapt to changes in equipment performance, improving the accuracy and sensitivity of fault diagnosis. Data fusion algorithms (such as principal component analysis (PCA) or deep neural networks (DNNs)) are integrated on edge computing nodes to fuse multi-source data such as temperature, vibration, current, and voltage from different sensors. This sensor data reflects different aspects of the equipment's health. Through the data fusion algorithm, comprehensive feature vectors reflecting the overall health of the equipment are extracted. These feature vectors more comprehensively and accurately describe the equipment's operating status than single sensor data. Pattern recognition is performed using the fused feature vectors to identify characteristic patterns of the equipment under different operating states. These modes include normal operating modes, abnormal operating modes, and potential failure modes. Real-time anomaly detection is performed on the fused data based on anomaly detection algorithms such as Isolation Forest or Support Vector Machine (SVM). These algorithms can identify data points that deviate from normal operating modes as potential fault signals. For detected abnormal patterns, cluster analysis or case-based reasoning methods are used to match the abnormal patterns with known fault types. This method can preliminarily predict the possible fault type and its severity. Based on the results of the fault type prediction, remote decision support is provided to operation and maintenance personnel, including fault handling solutions and resource allocation recommendations. Based on this information, operation and maintenance personnel can quickly respond to faults, prevent the faults from escalating, and ensure the safe and stable operation of the power grid.

[0089] The above technical solution achieves the following: By combining historical fault data with current operating data and utilizing machine learning algorithms (such as LSTM) to dynamically adjust the thresholds of monitoring parameters, this solution can accurately capture performance fluctuations of power grid equipment caused by environmental changes, aging, and other factors, thereby significantly improving the accuracy and sensitivity of fault diagnosis. This dynamic adjustment mechanism ensures the real-time and adaptability of the thresholds, avoiding the false positives and false negatives that can result from fixed thresholds. The introduction of a threshold adaptive learning mechanism enables the system to regularly evaluate the effectiveness of threshold adjustments and adjust algorithm parameters based on the feedback from the evaluation results. This closed-loop optimization system not only enhances the intelligence level of the fault diagnosis system but also ensures continuous optimization and improvement to adapt to the ever-changing performance of power grid equipment. Data fusion algorithms (such as PCA or DNN) are integrated on edge computing nodes to fuse multi-source data from different sensors and extract a comprehensive feature vector that reflects the health of the equipment. This approach fully utilizes the complementary nature of multi-source data, improving data utilization efficiency and information comprehensiveness, and providing a richer and more accurate information foundation for subsequent fault diagnosis. Pattern recognition using the fused feature vectors can accurately identify characteristic patterns of equipment under different operating conditions. These characteristic patterns not only represent the normal operating status of the equipment, but also include abnormal operating status and potential failure modes. This capability enables the system to detect potential failure signs earlier, providing operation and maintenance personnel with more time and space for decision-making; real-time anomaly detection is performed on the fused data based on anomaly detection algorithms (such as IsolationForest or SVM), which can quickly identify data points that deviate from the normal operating mode as potential failure signals. This real-time detection mechanism ensures that the system can respond at the early stage of the failure, effectively avoiding the further development and expansion of the failure; for the detected abnormal patterns, cluster analysis or case-based reasoning methods are used to match them with known fault types to preliminarily predict the possible fault type and its severity. This predictive capability provides operation and maintenance personnel with targeted fault handling suggestions and resource allocation plans, which helps to improve the efficiency and effectiveness of fault handling.

[0090] In one embodiment of the present invention, the threshold adaptive learning mechanism is used to periodically evaluate the effect of threshold adjustment, and the algorithm parameters are adjusted based on the evaluation results to form a closed-loop optimization system, including:

[0091] Establish a data feedback channel to transmit the actual thresholds and diagnostic results (including correct diagnoses, false positives, and missed negatives) used by edge computing nodes during preliminary fault diagnosis to the central control system in real time or regularly.

[0092] In the central control system, the threshold usage effect data from each edge computing node is processed through the data processing and analysis module, and a data set for threshold optimization is constructed;

[0093] A reinforcement learning algorithm (such as Q-learning or deep reinforcement learning (DRL)) is used to design a threshold optimization strategy based on the operating characteristics and failure modes of power grid equipment. The algorithm aims to minimize the false alarm rate and missed alarm rate and maximize the diagnostic accuracy.

[0094] The reinforcement learning model is trained and iterated using historical fault data and real-time feedback data, continuously optimizing the parameters of the threshold adjustment strategy, such as the learning rate and reward function, so that the model can more accurately adapt to the dynamic changes of power grid equipment.

[0095] Regularly evaluate the effectiveness of the optimized thresholds, and evaluate the effectiveness of the threshold adjustment strategy by comparing indicators such as false alarm rate, missed alarm rate, and diagnostic accuracy before and after optimization. Based on the evaluation results, automatically adjust the parameters of the reinforcement learning model to form a continuously optimized closed-loop system. For threshold adjustment strategies that perform poorly, make corrections or replacements in a timely manner.

[0096] The working principle of the above technical solution is to establish a data feedback channel to ensure that the actual thresholds used by edge computing nodes during the initial fault diagnosis process and the corresponding diagnostic results (including correct diagnoses, false positives, and missed negatives) are transmitted back to the central control system in real time or periodically. This step is the starting point of the closed-loop optimization system and provides basic data for subsequent data processing and analysis. In the central control system, the data processing and analysis module centrally processes the threshold usage data from each edge computing node. This step includes data cleaning, organization, and classification to construct a dataset for threshold optimization. The dataset contains threshold usage and corresponding diagnostic results under different time periods and conditions, providing rich data support for subsequent optimization algorithms. Reinforcement learning algorithms (such as Q-learning or deep reinforcement learning (DRL)) are used to combine the operating characteristics and failure modes of power grid equipment to design threshold optimization strategies. Reinforcement learning algorithms simulate the interaction between the environment and the agent to learn how to select the optimal action based on the current state to maximize cumulative rewards. In this scenario, the intelligent agent represents the threshold adjustment strategy. The state space includes operational data of power grid equipment and historical fault records, while the action space represents the direction of threshold adjustment. The reward function is designed based on changes in false alarm rate, false negative rate, and diagnostic accuracy. A reinforcement learning model is trained and iterated using historical fault data and real-time feedback. During training, the model continuously tries different threshold adjustment strategies and evaluates the effectiveness of each strategy (i.e., the reward value) based on actual diagnostic feedback. Through multiple iterations of training, the model gradually learns how to adjust the threshold based on the dynamic changes of power grid equipment to minimize false alarm rate and false negative rate and maximize diagnostic accuracy. The optimized threshold is regularly evaluated. The effectiveness of the threshold adjustment strategy is assessed by comparing metrics such as false alarm rate, false negative rate, and diagnostic accuracy before and after optimization. If the evaluation results show that the optimized strategy effectively improves diagnostic performance, the strategy is retained. If the performance is poor, the reinforcement learning model parameters (such as the learning rate and reward function) are automatically adjusted based on the evaluation results to find a more optimal threshold adjustment strategy. This process forms a continuously optimizing closed-loop system, ensuring that the threshold adjustment strategy can continuously adapt to the dynamic changes of power grid equipment. The system can promptly correct or replace underperforming threshold adjustment strategies. This can be achieved by redesigning the optimization strategy, adjusting model parameters, or introducing new algorithms. Through continuous trial and error, the system can gradually approach the optimal threshold adjustment strategy, providing more accurate and reliable support for power grid equipment fault diagnosis.

[0097] The above technical solution achieves the following: By feeding back the actual thresholds and diagnostic results of edge computing nodes to the central control system in real time or periodically, the system can promptly adjust thresholds to address dynamic changes in power grid equipment. This mechanism effectively reduces false positives and missed negatives caused by improper fixed threshold settings, improving the accuracy of fault diagnosis. The application of a reinforcement learning algorithm enables the system to continuously learn from historical fault data and real-time feedback to optimize the threshold adjustment strategy. As data accumulates and the learning process deepens, the system's understanding of the operating characteristics of power grid equipment deepens, enhancing its adaptive capabilities and enabling it to better cope with various complex situations. The establishment of a closed-loop optimization system ensures the continuous optimization of the threshold adjustment strategy. By regularly evaluating the effectiveness of the optimized thresholds and automatically adjusting the parameters of the reinforcement learning model based on the evaluation results, the system maintains its stability and reliability. Even in the face of sudden failures or abnormal changes in power grid equipment, the system can respond quickly, ensuring accurate and timely fault diagnosis. By improving the accuracy and timeliness of fault diagnosis, the system can reduce unnecessary inspections and maintenance work, thereby reducing operation and maintenance costs. Furthermore, thanks to the system's adaptive capabilities, operators and maintenance personnel no longer need to frequently adjust thresholds or update algorithm parameters, further reducing both labor and time costs. This technical solution is a crucial component of smart grid development. By enabling intelligent and automated fault diagnosis of power grid equipment, the system can improve the overall operational efficiency and safety of the power grid. This will help promote the rapid development of smart grids and provide strong support for building a safer, more reliable, and more efficient power network. By constructing a dataset for threshold optimization, the system provides operators with rich data support. This data not only reflects the operating status and failure modes of power grid equipment, but also reveals performance trends and potential risks. Based on this data, operators and maintenance personnel can make more informed and rational decisions, optimizing resource allocation and operation and maintenance strategies.

[0098] In one embodiment of the present invention, the S2 includes:

[0099] S21. Encrypt the preliminary diagnostic results and key sensor data using an encryption algorithm and transmit the encrypted data to the blockchain platform.

[0100] S22, based on smart contracts, is automatically triggered when edge computing nodes upload data to execute data verification and storage processes;

[0101] S23. Through the decentralized nature of blockchain, preliminary diagnostic results and key data are stored in multiple nodes, and the uploaded data is verified through the blockchain's consensus mechanism (such as PoW, PoS or DPoS).

[0102] The working principle of the above technical solution is as follows: Before sending preliminary diagnostic results and key sensor data to the blockchain platform, the data is first encrypted using an encryption algorithm. This step ensures data security and privacy during transmission, preventing it from being intercepted or tampered with by unauthorized third parties. The encrypted data is then transmitted to the blockchain platform via a secure channel (such as HTTPS or VPN). This transmission method further enhances data security, ensuring that the data is not stolen or leaked during transmission. When the edge computing node successfully uploads data, it triggers a smart contract pre-deployed on the blockchain platform. Smart contracts are self-executing contracts that automatically execute corresponding actions when specific conditions are met without human intervention. When the smart contract executes, it first verifies the uploaded data. The verification process may include checking the data's integrity, authenticity, and compliance with pre-set data formats and specifications. Only verified data is further processed or stored. If the data passes verification, the smart contract executes the data storage process. This includes recording the data in the corresponding block on the blockchain and ensuring its immutability and traceability. The decentralized nature of the blockchain enables the distributed storage of preliminary diagnostic results and key data across multiple nodes. This storage method not only improves data reliability and availability (even if some nodes fail, data will not be lost) but also enhances the security and stability of the system (because no centralized node can control or tamper with the data). When data is uploaded to the blockchain, the blockchain's consensus mechanism (such as PoW, PoS, or DPoS) is triggered. Consensus is the process by which nodes in a blockchain network reach agreement on the validity of block information. In PoW (Proof of Work), nodes compete for the right to create blocks by solving complex mathematical problems; in PoS (Proof of Stake) and DPoS (Delegated Proof of Stake), nodes determine their right to participate in consensus based on the number of tokens they hold or the number of votes they receive. Through these mechanisms, the blockchain network ensures that uploaded data is correctly verified and recorded, and that all nodes can reach consensus, guaranteeing the authenticity and consistency of the data.

[0103] The above technical solution achieves the following benefits: Preliminary diagnostic results and key sensor data are encrypted using an encryption algorithm, ensuring data security during transmission and preventing the risk of data leakage and tampering. This is particularly important in industrial or medical fields involving sensitive information, as it protects user privacy and trade secrets. The data verification process based on smart contracts can be automated without manual intervention, significantly improving the efficiency and accuracy of data verification. The transparency and immutability of smart contracts also ensure the fairness and credibility of the verification process. Leveraging the decentralized nature of blockchain, preliminary diagnostic results and key data are stored across multiple nodes, avoiding the risks of single points of failure and data concentration. This distributed storage approach improves data reliability and availability. Even if some nodes fail, data is not lost and access can be quickly restored. The blockchain's consensus mechanism (such as Proof-of-Work, Proof-of-Stake, or Delegated Proof-of-Stake) ensures that uploaded data is correctly verified and recorded. Through the consensus process, each node in the blockchain network can reach consensus on the authenticity and consistency of the data, thereby preventing malicious tampering or falsification of the data. The open and transparent nature of blockchain ensures that all transactions and data records in the system are viewable and verifiable by all participants. This transparency not only enhances the system's credibility but also allows data sources, processing, and results to be traced and audited, helping to identify potential issues and vulnerabilities. On a blockchain platform, different edge computing nodes and stakeholders can share and collaborate based on consensus-based data. This data sharing approach reduces the risk of data silos, promotes information flow, and optimizes resource allocation. Furthermore, blockchain's smart contract functionality supports automated data exchange and business processes, improving the efficiency and convenience of collaboration.

[0104] In one embodiment of the present invention, the step S23 includes:

[0105] Assess the sensitivity of preliminary diagnostic results and key sensor data, and based on the assessment results, determine which data requires a higher level of encryption and protection;

[0106] The preliminary diagnostic results and key sensor data are compressed using a compression algorithm. The data is intelligently sharded using a hash function based on data size, type, and access frequency, ensuring balance and independence between shards.

[0107] A multi-layer encryption strategy is adopted for sensitive data. The outer layer uses public key encryption to ensure data transmission security, and the inner layer uses symmetric encryption to improve decryption efficiency.

[0108] For data that needs to be verified but whose specific content does not want to be disclosed, zero-knowledge proof technology is used to allow the verifier to confirm the correctness of the data without having to know the data itself.

[0109] The compressed data is fragmented and stored on multiple nodes in the blockchain network. Through an automated data verification process, when new data is uploaded to the blockchain, the integrity and authenticity of the data are verified through a consensus mechanism and recorded on the blockchain to form an unalterable chain of evidence.

[0110] Regularly conduct security audits on blockchain networks to check for potential security vulnerabilities and threats, and build data backup mechanisms.

[0111] The above technical solution works as follows: First, the system assesses the sensitivity of preliminary diagnostic results and key sensor data. This step determines which data requires higher levels of encryption and protection based on the data's nature, potential risks, and relevant legal and regulatory requirements. Based on the assessment results, the system employs a multi-layered encryption strategy to protect sensitive data. The outer layer uses public key encryption to ensure data security during transmission and prevent interception by unauthorized third parties. The inner layer uses symmetric encryption to improve decryption efficiency and meet data security requirements during storage and processing. To save storage space and improve transmission efficiency, the system compresses preliminary diagnostic results and key sensor data. The compression algorithm selected depends on the data type and characteristics to achieve the optimal compression ratio while maintaining data integrity. The system then uses hash functions to intelligently shard the data based on its size, type, and access frequency. This step ensures balance and independence among shards, enabling efficient storage and access within the blockchain network. The sharding strategy also considers data redundancy and fault tolerance to improve data reliability and availability. For data that requires verification but does not want its specific content disclosed, the system employs zero-knowledge proof technology. This technology allows verifiers to confirm data accuracy without knowing the specific content, thereby protecting data privacy. When new data is uploaded to the blockchain, the system triggers an automated data verification process. Using a consensus mechanism (such as PoW, PoS, or Delegated Proof of Stake), multiple nodes in the blockchain network jointly verify the integrity and authenticity of the data. Once verified, the data is recorded on the blockchain, forming an immutable chain of evidence. Compressed data shards are distributed and stored across multiple nodes in the blockchain network. This decentralized storage method improves data reliability and availability, while leveraging the blockchain's consensus mechanism to ensure data consistency and immutability. The system regularly conducts security audits of the blockchain network to identify potential vulnerabilities and threats. Furthermore, a data backup mechanism is implemented to mitigate potential data loss or corruption. These measures collectively ensure the security, stability, and sustainability of the blockchain network and its data.

[0112] The above technical solution achieves the following: Through sensitivity assessment and a multi-layered encryption strategy, preliminary diagnostic results and critical sensor data are protected with differentiated protection, ensuring a higher level of encryption and security for sensitive data. Outer public key encryption ensures data security during transmission, while inner symmetric encryption improves decryption efficiency. The combination of these two effectively prevents the risk of data leakage and tampering. Data compression significantly reduces storage space and transmission bandwidth requirements, lowering storage and transmission costs. Furthermore, intelligent sharding technology shards data based on size, type, and access frequency, ensuring balance and independence between shards and improving data storage and access efficiency within the blockchain network. The introduction of zero-knowledge proof technology enables data verification without revealing its specific content. This protects data privacy while satisfying the verifier's need for data accuracy, promoting secure data sharing and collaboration. The blockchain's decentralization and consensus mechanism ensure the integrity and authenticity of data uploaded to the blockchain. Through an automated data verification process, newly uploaded data is verified by the consensus mechanism and recorded on the blockchain, forming an immutable chain of evidence. This effectively prevents the possibility of malicious data tampering or forgery. Sharded data storage and the decentralized nature of blockchain improve system reliability and availability. Even if some nodes fail or are attacked, data is not lost and access can be quickly restored. Furthermore, data backup mechanisms further ensure long-term data security and recoverability. Regular security audits of the blockchain network to identify potential security vulnerabilities and threats help identify and remediate issues promptly, improving the overall security of the system. This measure also helps meet relevant regulatory requirements for data governance and compliance, protecting the legitimate rights and interests of businesses and users.

[0113] In one embodiment of the present invention, S3 includes:

[0114] S31. Collect data through a comprehensive data integration platform, including data from edge computing nodes, blockchain platforms, and other data sources, and use streaming data processing technologies (such as Apache Kafka and Apache Flink) to efficiently process real-time data streams;

[0115] S32. Establish a data warehouse, use a distributed storage system (such as Hadoop HDFS, Amazon S3) to store historical data, and perform a pre-processing module on the stored data;

[0116] S33. Conduct preliminary exploratory analysis of data using statistical analysis and data mining techniques (e.g., PCA, SVD, cluster analysis, etc.) to discover hidden patterns and anomalies;

[0117] S34. Build complex machine learning prediction models (such as deep learning neural networks, random forests, gradient boosting trees, etc.) based on grid operation data for fault prediction and diagnosis, and train the models. Integrate real-time data feedback mechanisms to continuously monitor model performance and update and optimize the models based on new data.

[0118] S35: Accurately diagnose and locate faults through an intelligent diagnostic engine, combined with the output of machine learning prediction models and expert knowledge base;

[0119] S36. Decision support systems automatically generate treatment plans based on diagnostic results, including emergency response measures, maintenance plans, and resource allocation.

[0120] The above technical solution works as follows: Through a comprehensive data integration platform, the system collects data from multiple data sources (including edge computing nodes, blockchain platforms, and other data sources). This data may be real-time or historical. To process real-time data streams, the system uses streaming data processing technologies (such as Apache Kafka and Apache Flink) to ensure efficient data processing and low latency. These technologies can handle high-throughput data streams and enable real-time data access, processing, and forwarding. The collected data (both real-time and historical) is stored in distributed storage systems (such as Hadoop HDFS and Amazon S3) to ensure scalability and reliability. After data storage, the system performs preprocessing on the data, including data cleaning, transformation, and normalization, to ensure data quality and consistency, laying the foundation for subsequent data analysis and modeling. The system performs preliminary exploratory analysis of the data using statistical analysis and data mining techniques (such as PCA, SVD, and cluster analysis). These analytical techniques help discover hidden patterns, trends, and anomalies in the data, providing valuable insights for subsequent machine learning modeling. This process enables the system to better understand the inherent structure and characteristics of the data. Based on grid operation data, the system builds complex machine learning predictive models (such as deep learning neural networks, random forests, and gradient boosting trees) for fault prediction and diagnosis. These models are trained to learn patterns and regularities in the data and are able to predict future faults when new data is input. To maintain model accuracy and timeliness, the system also integrates a real-time data feedback mechanism to continuously monitor model performance and update and optimize the model based on new data. Through an intelligent diagnostic engine, the system combines the output of the machine learning predictive model with an expert knowledge base to accurately diagnose and locate faults. The intelligent diagnostic engine integrates multiple aspects of information (including model predictions, historical fault cases, and expert experience) to accurately determine the fault type and location. This step is critical for rapid response and effective handling of faults. Finally, through a decision support system, the system automatically generates action plans based on the diagnostic results. These action plans may include emergency response measures, maintenance plans, and resource allocation, aiming to quickly resolve the fault and restore normal grid operation. The decision support system can provide the optimal action plan based on the specific situation and constraints, minimizing the impact and losses on the grid.

[0121] The above technical solution achieves this goal: Through a comprehensive data integration platform and streaming data processing technologies (such as Apache Kafka and Apache Flink), the system can collect and efficiently process data from multiple sources, including edge computing nodes and blockchain platforms, in real time. This real-time performance ensures data freshness and accuracy, providing strong support for subsequent fault prediction and diagnosis. Using distributed storage systems (such as Hadoop HDFS and Amazon S3) to establish a data warehouse effectively addresses the storage challenges of massive amounts of historical data. Distributed storage not only improves data reliability and availability but also offers high scalability, enabling flexible expansion as data volumes grow. Using statistical analysis and data mining techniques (such as PCA, SVD, and cluster analysis), the system conducts preliminary exploratory data analysis, uncovering hidden patterns, trends, and anomalies. These insights provide critical input for subsequent machine learning modeling, facilitating more accurate prediction and diagnosis of power grid faults. Complex machine learning predictive models (such as deep learning neural networks, random forests, and gradient boosting trees) built on power grid operational data can enable intelligent prediction and diagnosis of power grid faults. Through training and learning, these models can accurately capture regularities and patterns in data, improving the accuracy and efficiency of predictions and diagnoses. By integrating a real-time data feedback mechanism, the system continuously monitors the performance of machine learning models and updates and optimizes them based on new data. This dynamic adjustment mechanism ensures that the models remain optimally configured, improving the accuracy and timeliness of predictions. By combining the output of machine learning prediction models with an expert knowledge base through an intelligent diagnostic engine, the system can accurately diagnose and locate faults. This diagnostic approach, which combines artificial intelligence with expert experience, significantly improves the accuracy and efficiency of fault handling. Based on the diagnostic results, the decision support system automatically generates solutions, including emergency response measures, maintenance plans, and resource allocation. This automated solution generation not only saves time and labor costs, but also improves the relevance and effectiveness of solutions.

[0122] In one embodiment of the present invention, the S4 includes:

[0123] S41. Edge computing nodes continuously monitor the operating status of power grid equipment, capture sensor data in real time, and perform preliminary processing. When an abnormal signal or potential fault is detected by the edge computing node, it is immediately marked and prepared to upload the relevant data.

[0124] S42. Upload preliminary diagnostic results and key sensor data to the blockchain platform through a secure data transmission protocol;

[0125] S43. The blockchain platform verifies the received data, stores it in the distributed ledger, and shares the verified data with all authorized nodes.

[0126] S44. When the blockchain platform detects newly uploaded abnormal data, it automatically triggers the deep learning model in the data center for further analysis;

[0127] S45. The deep learning model obtains data from the blockchain platform, performs pre-processing, and uses the trained deep learning model to conduct in-depth analysis of the processed data to identify the fault type, location, and possible cause.

[0128] S46. Generate a detailed diagnostic report based on the analysis results, including fault description, impact scope, severity, etc. The deep learning model returns the diagnostic results to the edge computing node through a secure channel and displays them to the operation and maintenance personnel on the user interface.

[0129] The working principle of the above technical solution is as follows: In the power grid system, edge computing nodes are deployed near grid equipment to continuously monitor their operating status. These nodes, connected to a sensor network, capture real-time data from equipment operations and perform preliminary processing. When an edge computing node detects an abnormal signal or potential fault, it flags the data and prepares to upload it to a higher-level processing system for more in-depth analysis and diagnosis. To ensure the security and integrity of data during transmission, the edge computing node uses secure data transmission protocols (such as HTTPS and TLS) to upload preliminary diagnostic results and key sensor data to a blockchain platform. The blockchain platform, with its decentralized and tamper-proof nature, provides a secure and reliable environment for data storage and sharing. Upon receiving the data, the blockchain platform first verifies its integrity and authenticity. Verified data is stored in a distributed ledger and shared with all authorized nodes through a consensus mechanism. This distributed storage and sharing mechanism not only improves data reliability but also enables multiple nodes to access and analyze the data simultaneously, improving processing efficiency. When the blockchain platform detects newly uploaded abnormal data, it automatically triggers a deep learning model within the data center for further analysis. Deep learning models are powerful data analysis tools that can identify anomalies and faults in new data by learning patterns and regularities from historical data. After receiving data from the blockchain platform, the deep learning model first preprocesses it to eliminate noise and redundant information. The trained deep learning model then conducts in-depth analysis of the processed data. Using a complex neural network structure and algorithms, the model identifies the type, location, and possible cause of the fault, providing strong support for subsequent troubleshooting. Based on the deep learning model's analysis results, the system generates a detailed diagnostic report. This report includes key information such as the fault description, impact scope, and severity, helping operations personnel quickly understand the fault and develop appropriate solutions. The deep learning model also transmits diagnostic results to the edge computing node via a secure channel and displays them to operations personnel on a user interface. This allows personnel to obtain fault information immediately and take appropriate countermeasures.

[0130] The above technical solution achieves the following: The continuous monitoring and real-time data capture capabilities of edge computing nodes enable rapid detection of any abnormal signals or potential faults in power grid equipment. This real-time fault detection mechanism significantly shortens fault discovery time, saving valuable time for subsequent troubleshooting. Data is transmitted to the blockchain platform via a secure data transmission protocol, ensuring the security and integrity of the data during transmission. The distributed ledger technology and decentralized nature of the blockchain platform make data more difficult to tamper with or delete, effectively protecting data privacy and security. The blockchain platform not only stores verified data but also shares it with all authorized nodes through a consensus mechanism. This data sharing mechanism improves data transparency and traceability, enabling multiple departments or teams to collaborate and analyze based on the same data. The introduction of deep learning models makes the analysis and diagnosis of power grid equipment faults more intelligent and accurate. By learning from the patterns and regularities in historical data, the model can conduct in-depth analysis of new abnormal data and accurately identify the fault type, location, and possible cause. Based on the analysis results of the deep learning model, the system can automatically generate a detailed diagnostic report. This report contains comprehensive fault information, including fault description, impact scope, and severity, providing operators with clear and accurate fault information. Furthermore, diagnostic results are returned to the edge computing node via a secure channel and displayed on the user interface, improving the efficiency and convenience of information transmission. Through automated and intelligent fault detection, analysis, and diagnosis processes, the S4 technology solution significantly improves the efficiency of power grid operations and maintenance. Operators can more quickly obtain fault information, develop solutions, and implement appropriate countermeasures. Furthermore, the automated process reduces the risk of human error and operational costs. Real-time, accurate fault detection and diagnosis help grid operators promptly eliminate potential safety hazards and fault points, thereby enhancing the stability and reliability of the power grid. This is crucial for ensuring the continuity and security of power supply.

[0131] In one embodiment of the present invention, the S5 includes:

[0132] S51. Operations and maintenance personnel view the deep learning model's diagnostic results and decision support information through the user interface and evaluate the feasibility of the fault handling solution and the rationality of the resource allocation suggestion.

[0133] S52. Execute fault handling operations based on the decision support information and adjust the power grid operation strategy based on the impact of the fault;

[0134] S53. During the fault handling process, continuously monitor the operating status of the power grid equipment and provide real-time feedback on the handling effect;

[0135] S54. Record the entire process of troubleshooting, organize the results into documents or reports, upload them through the blockchain platform, and form a complete troubleshooting file together with the preliminary diagnosis results and diagnostic report.

[0136] The working principle of this technical solution is as follows: Operations and maintenance personnel first view the diagnostic results and decision-making support information generated by the deep learning model through a user interface. This information includes a detailed description of the fault, its scope, severity, and the system's recommended troubleshooting solutions and resource allocation suggestions. Based on their professional knowledge and experience, operations and maintenance personnel evaluate the feasibility and rationality of this information. They consider factors such as implementation difficulty, time required, and resource costs, and comprehensively determine and select the most appropriate troubleshooting solution. Once a troubleshooting solution is determined, the operations and maintenance personnel immediately implement the relevant actions. This may include remotely adjusting equipment parameters, dispatching a maintenance team for on-site repairs, or replacing faulty components. Furthermore, operations and maintenance personnel adjust the grid's operating strategy based on the fault's impact on grid operations to ensure grid stability and reliability. These adjustments may involve power dispatch, load distribution, and backup power source switching. During the troubleshooting process, the system continuously monitors the operating status of grid equipment and provides real-time feedback to operations and maintenance personnel on the results of the troubleshooting. Through real-time data monitoring and analysis, operations and maintenance personnel can promptly understand the progress and effectiveness of the troubleshooting process, allowing them to make necessary adjustments and optimizations to the solution. This real-time monitoring and feedback mechanism ensures timely and effective troubleshooting. After troubleshooting is complete, the system automatically records the entire process. This record includes detailed information such as the time taken, measures taken, and results. Operations and maintenance personnel compile this information into documents or reports and upload them via the blockchain platform. These documents or reports, along with the preliminary diagnosis results and diagnostic report, form a complete troubleshooting archive. The blockchain platform's distributed storage and immutability ensure the authenticity and traceability of the troubleshooting archive, providing strong support for subsequent analysis and improvement.

[0137] The above technical solution provides the following benefits: Operations and maintenance personnel can directly view the diagnostic results and decision-making support information generated by the deep learning model through a user interface. This information, derived from big data and machine learning algorithms, is highly accurate and reliable. Operations and maintenance personnel can quickly assess the feasibility of fault resolution solutions and the rationality of resource allocation recommendations, enabling them to make rapid decisions and improve handling efficiency. Decision-making support information includes resource allocation recommendations, such as dispatching maintenance teams and adjusting spare parts inventory. These recommendations, based on real-time data and historical experience, help operations and maintenance personnel more effectively allocate resources, avoid waste and shortages, and improve resource utilization efficiency. Operations and maintenance personnel can also adjust grid operation strategies in a timely manner based on the impact of faults on grid operations. This immediate adjustment helps minimize the impact of faults on grid stability and ensure the continuity and reliability of power supply. During the fault handling process, the operating status of grid equipment is continuously monitored, providing real-time feedback on the handling results. This real-time monitoring mechanism allows operations and maintenance personnel to promptly understand the progress and effectiveness of fault handling, making necessary adjustments and optimizations to the handling plan to ensure a thorough resolution of the fault. The entire fault handling process is recorded, including handling time, measures taken, and results, and this information is compiled into documents or reports. This standardized document management approach helps preserve historical data, providing strong support for subsequent analysis and improvement. Furthermore, documents or reports uploaded via the blockchain platform, along with preliminary diagnostic results and reports, form a complete troubleshooting archive, enhancing data security and traceability. The blockchain platform's distributed storage and tamper-proof nature ensure the authenticity and credibility of the troubleshooting archive. Any modifications or deletions to the archive are recorded on the blockchain, forming a traceable historical record. This transparency helps enhance the credibility of O&M work and strengthens user trust in grid operations. Complete troubleshooting archives not only document the process and results, but also encompass the O&M personnel's decision-making process and experience. These archives serve as a knowledge base for other O&M personnel to learn from and reference, promoting knowledge sharing and experience accumulation. Furthermore, through analysis and summary of historical archives, valuable patterns and insights can be extracted to guide future troubleshooting efforts.

[0138] One embodiment of the present invention provides a remote fault diagnosis system for power grid operation and maintenance, comprising a memory, a processor, and a computer program stored in and executable on the memory, wherein the processor executes the program to implement any of the remote fault diagnosis methods for power grid operation and maintenance described above.

[0139] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A remote fault diagnosis method for power grid operation and inspection, characterized in that: The method comprises: S1. Collect the operating status data of power grid equipment in real time through the sensor network, and process and preliminarily analyze the operating status data in real time through the edge computing node; S2, upload the preliminary diagnosis results and key sensor data of the edge computing node to the blockchain platform, and store and verify the preliminary diagnosis results and key sensor data through the distributed ledger of the blockchain; S3. Build a data set, train the deep learning model through the data set, interact with the blockchain platform, and receive data from edge nodes for more in-depth analysis and diagnosis; S4, information is transmitted through real-time data communication mechanism. When the edge computing node finds abnormal signals or potential faults, the relevant information is uploaded to the blockchain platform for verification and sharing, and the deep learning model is triggered for further analysis and diagnosis; S5. Obtain remote decision support based on the diagnostic results of the deep learning model, and perform fault handling based on the decision support information. After the fault handling is completed, the processing results will be recorded and uploaded to the blockchain platform to form a complete fault handling file.

2. A method for remote fault diagnosis for power grid operation and inspection according to claim 1, characterized in that: Said S1 comprises: S11. Determine the deployment location of edge computing nodes through optimization algorithms based on the grid topology, equipment density, and fault history data; S12. Integrate various types of sensors to form a comprehensive monitoring network, and monitor the status of power grid equipment through the comprehensive monitoring network; S13. Based on the Internet of Things, sensor data is collected in real time and the collected data is transmitted to the edge computing node; S14, the edge computing node performs real-time processing and preliminary analysis on the received sensor data; S15: Based on the preset threshold, perform preliminary fault diagnosis on the data after preliminary analysis to identify possible abnormalities or fault signs.

3. A method for remote fault diagnosis for power grid operation and inspection according to claim 2, characterized in that: The S15 comprises: Using historical fault data and current operation data, combined with machine learning algorithms, the thresholds of various monitoring parameters are dynamically adjusted; Through the threshold adaptive learning mechanism, the threshold adjustment effect is evaluated regularly, and the algorithm parameters are adjusted according to the evaluation results to form a closed-loop optimization system; Integrate data fusion algorithms on edge computing nodes to fuse multi-source data from different sensors and extract comprehensive feature vectors that reflect the health status of the equipment; Use the fused feature vectors to perform pattern recognition and identify the characteristic patterns of the equipment under different operating conditions; Based on the anomaly detection algorithm, the fused data is subjected to real-time anomaly detection to identify data points that deviate from the normal operating mode as potential fault signals; For the detected abnormal patterns, cluster analysis or case-based reasoning methods are used to match the abnormal patterns with known fault types to preliminarily predict the possible fault types and their severity.

4. A method for remote fault diagnosis for power grid operation and inspection according to claim 3, characterized in that: The threshold adaptive learning mechanism is used to regularly evaluate the effect of threshold adjustment, and the algorithm parameters are adjusted according to the evaluation result feedback to form a closed-loop optimization system, including: Establish a data feedback channel to transmit the actual thresholds and diagnostic results used by edge computing nodes in the initial fault diagnosis process to the central control system in real time or regularly; In the central control system, the threshold usage effect data from each edge computing node is processed through the data processing and analysis module, and a data set for threshold optimization is constructed; Using reinforcement learning algorithm, combined with the operating characteristics and failure modes of power grid equipment, a threshold optimization strategy is designed; Use historical fault data and real-time feedback data to train and iterate the reinforcement learning model, and continuously optimize the parameters of the threshold adjustment strategy; Regularly evaluate the effectiveness of the optimized thresholds, and automatically adjust the parameters of the reinforcement learning model based on the evaluation results to form a closed-loop system of continuous optimization; timely correct or replace threshold adjustment strategies that perform poorly.

5. A method for remote fault diagnosis for power grid operation and inspection according to claim 1, characterized in that: The S2 comprises: S21. Encrypt the preliminary diagnosis results and key sensor data through encryption algorithms, and transmit the encrypted data to the blockchain platform. S22, based on smart contracts, is automatically triggered when edge computing nodes upload data to execute data verification and storage processes; S23. Through the decentralized nature of blockchain, preliminary diagnostic results and key data are stored in multiple nodes, and the uploaded data is verified through the consensus mechanism of blockchain.

6. A method for remote fault diagnosis for power grid operation and inspection according to claim 1, characterized in that: The S23 comprises: Evaluate the sensitivity of preliminary diagnostic results and key sensor data; The initial diagnosis results and key sensor data are compressed using compression algorithms; hash functions are used to intelligently slice data based on data size, type, and access frequency; The compressed data is fragmented and stored in multiple nodes in the blockchain network. Through the automated data verification process, when new data is uploaded to the blockchain, the integrity and authenticity of the data are verified through the consensus mechanism and recorded on the blockchain to form an unalterable chain of evidence. Regularly conduct security audits on blockchain networks, check for potential security vulnerabilities and threats, and build data backup mechanisms.

7. A remote fault diagnosis method for power grid operation and inspection according to claim 1, characterized in that: The S3 includes: S31. Collect data through a comprehensive data integration platform and use streaming data processing technology to process real-time data streams; S32, establish a data warehouse, use a distributed storage system to store historical data, and perform a preprocessing module on the stored data; S33. Conduct preliminary exploratory analysis of data through statistical analysis and data mining techniques to discover hidden patterns and anomalies; S34. Build a machine learning prediction model based on power grid operation data and train the model; integrate a real-time data feedback mechanism to continuously monitor model performance and update and optimize the model based on new data; S35. Use the intelligent diagnosis engine, combined with the output of the machine learning prediction model and the expert knowledge base, to diagnose and locate faults; S36. Automatically generate a treatment plan based on the diagnosis results through a decision support system.

8. A method for remote fault diagnosis for power grid operation and inspection according to claim 1, characterized in that: The S4 comprises: S41. The edge computing node continuously monitors the operating status of the power grid equipment, captures sensor data in real time and performs preliminary processing; when the edge computing node detects an abnormal signal or potential fault, it immediately marks and prepares to upload the relevant data; S42, upload preliminary diagnostic results and key sensor data to the blockchain platform through a secure data transmission protocol; S43. The blockchain platform verifies the received data and stores it in the distributed ledger, while sharing the verified data with all authorized nodes; S44. After the blockchain platform detects the newly uploaded abnormal data, it automatically triggers the deep learning model in the data center for further analysis; S45. The deep learning model obtains data from the blockchain platform, performs preprocessing, and uses the trained deep learning model to conduct in-depth analysis on the processed data; S46. A detailed diagnostic report is generated based on the analysis results. The deep learning model returns the diagnostic results to the edge computing node through a secure channel and displays them to the operation and maintenance personnel on the user interface.

9. A method for remote fault diagnosis for power grid operation and inspection according to claim 1, characterized in that: The S5 comprises: S51. Operation and maintenance personnel view the diagnosis results and decision support information of the deep learning model through the user interaction interface; evaluate the feasibility of the fault handling solution and the rationality of the resource allocation suggestion; S52, executing fault handling operations according to the decision support information, and adjusting the power grid operation strategy according to the impact of the fault; S53. During the fault handling process, the operating status of the power grid equipment is continuously monitored and the handling effect is fed back in real time; S54. Record the entire process of troubleshooting, organize the processing results into documents or reports, upload them through the blockchain platform, and form a complete troubleshooting file together with the preliminary diagnosis results and diagnostic reports.

10. A remote fault diagnosis system for power grid operation and inspection, characterized in that: It includes a memory, a processor, and a computer program stored in the memory and executable on the memory, wherein the processor executes the program to implement the remote fault diagnosis method for power grid operation and inspection as described in any one of claims 1 to 9.

Images