Substation main equipment self-checking system based on multi-source heterogeneous data fusion technology

Abstract

The invention relates to the technical field of power systems, in particular to a substation main equipment self-checking system based on a multi-source heterogeneous data fusion technology. The self-inspection system comprises a data acquisition and preprocessing module, a self-adaptive data fusion algorithm module, an equipment fault mode dynamic identification and intelligent repair module, a multi-scale time sequence prediction and emergency plan generation module, an equipment health state evaluation and residual life prediction module, and a fault diagnosis and knowledge base updating module. A user interaction and remote monitoring management module, and an intelligent prediction and optimization scheduling module; according to the substation main equipment self-checking system based on the multi-source heterogeneous data fusion technology, comprehensive monitoring and intelligent management of substation equipment are achieved by integrating various data acquisition and processing technologies.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a substation main equipment self-checking system based on multi-source heterogeneous data fusion technology. Background Art

[0002] With the continuous development of the power system, substations are important nodes for power transmission and distribution, and the reliability and stability of their equipment are particularly important. Traditional substation equipment detection methods mainly rely on manual inspections and regular maintenance, which not only consumes a lot of manpower and material resources, but also makes it difficult to detect potential equipment failures in a timely manner, resulting in high equipment failure rates, high maintenance costs, and high risks of power outages.

[0003] Although some automated detection methods have been introduced into substation equipment, such as online monitoring systems and intelligent diagnosis systems, these systems can usually only process a single data source and it is difficult to effectively integrate data from different sensors and devices, resulting in insufficient detection accuracy and reliability. In addition, the existing systems also have many limitations in fault pattern recognition, fault prediction, and emergency plan generation, and cannot achieve comprehensive, real-time monitoring and intelligent management of substation equipment. Summary of the invention

[0004] The purpose of the present invention is to provide a substation main equipment self-checking system based on multi-source heterogeneous data fusion technology to solve the problems raised in the above background technology.

[0005] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a substation main equipment self-inspection system based on multi-source heterogeneous data fusion technology, the self-inspection system includes a data acquisition and preprocessing module, an adaptive data fusion algorithm module, a dynamic identification and intelligent repair module for equipment failure mode, a multi-scale time series prediction and emergency plan generation module, an equipment health status assessment and remaining life prediction module, a fault diagnosis and knowledge base update module, a user interaction and remote monitoring management module, and an intelligent prediction and optimization scheduling module; the data acquisition and preprocessing module is used to collect electrical signal data and environmental data on the main equipment of the substation, and perform corresponding preprocessing; the adaptive data fusion algorithm module is used to adjust the weights of different data sources in real time to improve the fusion accuracy and robustness of multi-source heterogeneous data; the dynamic identification of equipment failure mode and the intelligent repair module are used to improve the fusion accuracy and robustness of multi-source heterogeneous data; the self-inspection system includes a data acquisition and preprocessing module, an adaptive data fusion algorithm module, a dynamic identification and intelligent repair module for equipment failure mode, a multi-scale time series prediction and emergency plan generation module, an equipment health status assessment and remaining life prediction module, a fault diagnosis and knowledge base update module, a user interaction and remote monitoring management module, and an intelligent prediction and optimization scheduling module; the data acquisition and preprocessing module is used to collect electrical signal data and environmental data on the main equipment of the substation, and perform corresponding preprocessing; the adaptive data fusion algorithm module is used to adjust the weights of different data sources in real time to improve the fusion accuracy and robustness of multi-source heterogeneous data; the self-inspection system includes a data acquisition and preprocessing module, an adaptive data fusion algorithm module, a dynamic identification and intelligent repair module for equipment failure mode, and a multi-scale time series prediction and emergency plan generation module. The identification and intelligent repair module is used to identify known and unknown failure modes of substation equipment and provide self-repair suggestions; the multi-scale time series prediction and emergency plan generation module is used to predict failure trends at different time scales and generate dynamic emergency plans; the equipment health status assessment and remaining life prediction module is used to assess the health status of the main equipment of the substation, predict its remaining service life, and provide maintenance and replacement decision support; the fault diagnosis and knowledge base update module is used to identify substation equipment failures and update the fault knowledge base to achieve intelligent fault handling; the user interaction and remote monitoring management module is used to implement a user-friendly interface for substation equipment and support remote monitoring and management; the intelligent prediction and optimization scheduling module is used for intelligent prediction and optimization algorithms to improve the scheduling efficiency and resource utilization of substation equipment.

[0006] Preferably, the data processing method of the adaptive data fusion algorithm module includes the following steps: S1, extracting data features; S2, adaptively adjusting the weights of data sources; S3, fusing and outputting data; S4, self-learning and optimizing the algorithm model.

[0007] Preferably, the method for adaptive weight adjustment of the data source in the above step S2 is to first construct an adaptive fusion model, in which the model input is the feature vector extracted by each sensor, dynamically adjust the weight of each sensor by comparing historical data with real-time data, embed an adaptive weight adjustment layer in the deep learning model, and use a loss function based on mean square error or cross entropy to optimize the model and train it to ensure that the weight of each data source is adaptively adjusted with factors such as environmental changes and sensor status.

[0008] Preferably, in the above step S3, the data fusion method is to sum the feature vectors of different sensors according to weights to generate a fused feature vector, and the data output method is to input the fused feature into a classifier or regression model to obtain the state prediction or fault detection result of the equipment.

[0009] Preferably, the identification and repair method of the equipment fault mode dynamic identification and intelligent repair module includes the following steps: S1, fault mode identification; S2, fault mode dynamic update and expansion; S3, intelligent repair suggestion generation; S4, self-optimization and feedback.

[0010] Preferably, the dynamic emergency plan generation method of the multi-scale time series prediction and emergency plan generation module includes the following steps: S1, time series feature extraction; S2, fault trend prediction; S3, generating an emergency plan based on the fault prediction results; S4, performing dynamic emergency plan optimization; S5, performing simulation drills and plan verification.

[0011] Preferably, the evaluation and prediction method of the equipment health status evaluation and remaining life prediction module includes the following steps: S1, perform equipment health status evaluation; S2, perform remaining life prediction; S3, generate adaptive maintenance strategy; S4, analyze and visualize equipment health trend; S5, decision support and optimization.

[0012] Preferably, the intelligent fault handling method of the fault diagnosis and knowledge base update module includes the following steps: S1, fault diagnosis technology implementation; S2, fault analysis and decision support; S3, knowledge base update and self-learning; S4, fault mode and trend analysis; S5, feedback and improvement.

[0013] Preferably, the prediction and optimization method of the intelligent prediction and optimization scheduling module includes the following steps: S1, load prediction and trend analysis; S2, scheduling strategy generation and optimization; S3, resource allocation and dynamic adjustment; S4, simulation and optimization evaluation; S5, user feedback and iterative improvement; S6, feeding back the results to the user's visual scheduling interface and generating a report.

[0014] Preferably, the load forecasting and trend analysis method in the above step S1 is to use a time series analysis method to predict the load demand of the equipment, and then use historical load data to build a model to predict future load change trends and analyze the impact of different seasons and time periods on the load; the scheduling strategy generation and optimization method in the above step S2 is that according to the load forecast results, the system generates a preliminary scheduling plan, and then uses optimization algorithms such as genetic algorithms or particle swarm optimization to optimize the preliminary scheduling plan.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention uses an adaptive data fusion algorithm module to adjust the weights of different data sources in real time, improve the fusion accuracy and robustness of multi-source heterogeneous data, ensure the accuracy and reliability of data processing, and the equipment failure mode dynamic identification and intelligent repair module can identify known and unknown failure modes and provide self-repair suggestions, reduce equipment failure rate, and improve the stability of equipment operation.

[0016] 2. The present invention uses a multi-scale time series prediction and emergency plan generation module to predict fault trends on different time scales and generate dynamic emergency plans, thereby improving emergency response capabilities and reducing the risk of power outages. The equipment health status assessment and remaining life prediction module can assess the health status of the equipment and predict its remaining service life, provide maintenance and replacement decision support, extend the equipment service life, and reduce maintenance costs.

[0017] 3. The intelligent prediction and optimization scheduling module of the present invention improves the scheduling efficiency and resource utilization of substation equipment and optimizes the operation of the power system through intelligent prediction and optimization algorithms. It can significantly improve the detection accuracy and management efficiency of the main equipment of the substation, reduce equipment failure rate and maintenance costs, and improve the reliability and stability of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 Schematic diagram of the self-checking system of the present invention.

[0019] Figure 2 It is a flowchart of the data processing method of the present invention.

[0020] Figure 3 It is a flowchart of the adaptive weight adjustment method of the present invention.

[0021] Figure 4 It is a flowchart of the data fusion and output method of the present invention.

[0022] Figure 5 It is a flowchart of the identification and repair method of the present invention.

[0023] Figure 6 The figure is a flowchart of the method for generating a dynamic emergency plan according to the present invention.

[0024] Figure 7 It is a flowchart of the evaluation and prediction method of the present invention.

[0025] Figure 8 It is a flowchart of the intelligent fault handling method of the present invention.

[0026] Fig. 9 It is a flowchart of the prediction and optimization method of the present invention.

[0027] Fig.10 It is a flowchart of the load forecasting and trend analysis method of the present invention.

[0028] Fig.11 This is a flowchart of the scheduling strategy generation and optimization method of the present invention. DETAILED DESCRIPTION

[0029] The present invention is further described below in conjunction with the accompanying drawings and specific implementation methods. It should be noted that, under the premise of no conflict, the various embodiments or technical features described below can be arbitrarily combined to form a new embodiment. It should be noted that the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0030] See also Figures 1 to 11 The present invention provides a technical solution: a substation main equipment self-test system based on multi-source heterogeneous data fusion technology, the system includes the following modules: 1. The data acquisition and preprocessing module collects electrical signal data and environmental data from the main equipment of the substation through multi-source sensors, and performs corresponding preprocessing to meet the needs of the self-test system operation.

[0031] When it is working, it transmits the collected data to other modules and performs corresponding self-checking work.

[0032] Data collection and preprocessing phase of the adaptive data fusion algorithm module: Collect heterogeneous data from multiple sensors or devices. Data sources include electrical signal data (such as voltage, current), environmental data (temperature, humidity) and historical operating status data of the equipment. The collected data is preprocessed, including denoising, normalization and filtering of abnormal data. For example, Kalman filter or moving average filter is used to reduce the impact of noise, and the range of different data sources is normalized by Z-Score.

[0033] Data collection and preprocessing stage of the equipment failure mode dynamic identification and intelligent repair module: real-time collection of equipment operation data, such as current, voltage, temperature, humidity and other key parameters. These data will be extracted from multi-source sensors and synchronously transmitted to the central processing system to preprocess the collected data, including data cleaning, denoising, anomaly detection and time alignment. Statistical analysis methods such as principal component analysis (PCA) or independent component analysis (ICA) are used to eliminate irrelevant noise data.

[0034] Multi-scale time series prediction and emergency plan generation module data collection and multi-scale time segmentation stage: collect real-time monitoring data of the equipment, including key parameters such as current, voltage, temperature, and pressure. These data are transmitted to the system in real time through the sensor network, and the collected data is processed by multi-scale time segmentation, and the data is divided into different time periods for analysis. Common time scales include short-term (such as seconds and minutes) and long-term (such as hours and days) to capture short-term fluctuations and long-term trends of equipment.

[0035] Data collection and preliminary processing stage of the equipment health status assessment and remaining life prediction module: Real-time collection of key performance indicators (KPIs) of equipment, such as temperature, voltage, current, vibration and other sensor data, as well as historical maintenance records, operating time and other related data. Use IoT sensors to achieve all-weather monitoring to ensure the continuity and integrity of data, and pre-process the collected raw data, including data cleaning, denoising, and normalization to ensure data consistency and validity. Common algorithms such as mean filtering and Kalman filtering can be used to remove noise.

[0036] Fault data collection and recording phase of the fault diagnosis and knowledge base update module: Real-time monitoring of key operating parameters of the equipment (such as current, voltage, temperature, vibration, etc.), and sending data to the fault diagnosis module through the data acquisition system. Ensure that the frequency of data collection is higher than the frequency of fault occurrence, so as to capture fault signs in time and establish a fault event recording system. The system will automatically record the time of each fault, equipment status, operating environment and repair measures taken, forming a detailed fault case database.

[0037] Data collection and environmental analysis phase of the intelligent prediction and optimization scheduling module: real-time collection of equipment operation data and environmental data, including load data, electricity prices, meteorological conditions, etc., using sensors and external data sources (such as weather forecast API) to obtain information, and pre-processing the collected data to ensure data accuracy and consistency. Data cleaning includes removing outliers and filling missing values ​​to improve the accuracy of the prediction model.

[0038] 2. Adaptive data fusion algorithm module Used to adjust the weights of different data sources in real time to improve the fusion accuracy and robustness of multi-source heterogeneous data.

[0039] Feature extraction stage: Data features are extracted through convolutional neural networks (CNNs). The signal of each sensor is processed by different convolutional layers to extract features at different levels. For example, the time series data of a temperature sensor can be extracted for its short-term fluctuation features through a one-dimensional convolutional network. The features of different sensor data are weighted using an attention mechanism. The attention mechanism calculates the importance weight of each sensor data by learning the correlation between different features.

[0040] Adaptive weight adjustment phase: Build an adaptive fusion model with the feature vector extracted from each sensor as the model input. Dynamically adjust the weight of each sensor by comparing historical data with real-time data. Embed an adaptive weight adjustment layer in the deep learning model. Use a loss function based on mean square error (MSE) or cross entropy to optimize the model training to ensure that the weight of each data source is adaptively adjusted with factors such as environmental changes and sensor status.

[0041] Data fusion and output stage: The feature vectors of different sensors are weighted and summed to generate a fused feature vector. This vector is a comprehensive expression of multi-source data, reflecting the importance of each data source in the current environment. The fused features are input into the classifier or regression model to finally obtain the state prediction or fault detection results of the equipment.

[0042] Model self-learning and optimization stage: The model is updated regularly through the feedback loop. The system can use the reinforcement learning mechanism to adjust the weight strategy of each data source according to the error between the actual detected device state and the predicted result. In this way, the accuracy of the data fusion model can be gradually optimized as the device operation time increases.

[0043] This module uses a dynamic attention mechanism and an adaptive weight adjustment algorithm to enable multi-source heterogeneous data to maintain high fusion accuracy in any environment. The model's self-learning and optimization are achieved by combining convolutional neural networks with reinforcement learning algorithms.

[0044] 3. Dynamic identification of equipment failure modes and intelligent repair module Capable of identifying known and unknown failure modes of substation equipment and providing self-repair recommendations.

[0045] Failure mode identification phase: Use graph neural networks to build association models between devices, treat data from different devices and sensors as network nodes, and use edges to represent the associations between them. In this way, the state changes of devices and their mutual influence can be discovered by analyzing the network structure. Models such as support vector machines or deep belief networks are trained based on historical data to classify and detect known equipment failure modes. These models can detect known types of failures, such as low voltage, high temperature, and abnormal frequency, through classification algorithms. Use autoencoders in deep learning technology to identify abnormal patterns that have never been seen before. When the autoencoder model cannot effectively reconstruct the current data, it indicates that there may be new unknown failure modes. The model will give an alarm prompt based on the abnormality score.

[0046] Failure mode dynamic update and expansion stage: The fault mode library is dynamically updated through online learning technology. The system will regularly fine-tune the model for new fault data, identify and classify new fault modes. This enables the system to automatically expand its fault identification capabilities. It connects with the maintenance logs and external fault databases provided by the equipment manufacturer to automatically import new fault modes. This ensures the continuous expansion and updating of the system fault identification library.

[0047] Smart repair suggestion generation phase: Generate the best repair suggestions based on reinforcement learning algorithms. The reinforcement learning model uses historical data on device status to build a reward function and find the best repair path and operation suggestions. For example, when a voltage anomaly is detected, the system will recommend checking a specific switch device or performing a repair operation. The system provides multiple levels of intelligent repair solutions, ranging from simple device restarts to complex component replacement solutions. Depending on the severity of the fault and the operating environment, the system will provide a variety of repair strategies for engineers to choose from.

[0048] Self-optimization and feedback mechanism: By monitoring the operation of the equipment after the repair operation is performed, the system will automatically evaluate the repair effect and optimize the repair strategy through reinforcement learning feedback. If a repair measure is not effective, the system will reduce the priority of the strategy and gradually optimize the repair process.

[0049] This module combines graph neural networks with traditional machine learning methods to achieve dynamic identification and correlation analysis of equipment faults. It uses autoencoder technology to identify unknown faults and expand the fault mode library. It generates self-optimizing intelligent repair solutions through reinforcement learning, achieving highly automated equipment fault handling.

[0050] 4. Multi-scale time series prediction and emergency plan generation module Predict fault trends at different time scales and generate dynamic emergency response plans.

[0051] Time series feature extraction stage: Use wavelet transform or Fourier transform to perform frequency domain analysis on time series data to extract frequency components and their changing trends. Wavelet transform can capture features in different time periods at multiple scales, and is particularly suitable for detecting emergencies (such as current shocks). Use long short-term memory networks (LSTM) or gated recurrent units (GRU) to build time series prediction models to model and learn device operation data. LSTM has the ability to remember long-term dependencies and can effectively predict the operating trends of devices in the future.

[0052] Fault trend prediction stage: On a short-term time scale, the LSTM model predicts the state of the equipment in the next few minutes or hours, mainly used to detect sudden failure risks, such as a sudden drop in voltage or a rapid rise in temperature. On a long-term time scale, historical data is combined to predict the chronic failure or performance degradation trend that may occur in the equipment. For example, by analyzing the data of the past few months, the life attenuation or maintenance needs of the equipment can be predicted. The Bayesian prediction model is used to estimate the probability of failure and its time range, and generate a failure risk curve. The model combines the historical failure data of the equipment with the current status data to infer the probability that the equipment may fail in the future.

[0053] Emergency plan generation stage: Based on the results of fault prediction, the system automatically generates a corresponding emergency plan. The emergency plan generates a multi-level response plan based on the type of equipment, fault category and environmental conditions. For example, for a slight temperature rise, the system will generate an alarm and inspection suggestion; for a serious voltage anomaly, the system will trigger the equipment to shut down or switch to a backup line. The system generates the priority and response schedule of the emergency plan based on the importance and current operating status of the equipment to ensure that key equipment is given priority. This process uses the Analytic Hierarchy Process (AHP) to evaluate the priority of emergency measures based on the failure risk and impact range of the equipment.

[0054] Dynamic emergency plan optimization stage: By monitoring the status of equipment and changes in the operating environment in real time, the system will dynamically adjust the emergency plan. For example, when the ambient temperature changes dramatically, the emergency plan will be updated in real time to ensure that the plan matches the current environmental conditions. Connecting with the dispatch system and personnel communication interface, the system will send an alarm to the operator based on the predicted time and severity of the fault, and provide detailed emergency operation steps and preventive maintenance suggestions.

[0055] Simulation drill and plan verification phase: The system supports simulation drills for emergency plans. During normal operation, the system simulates various failure scenarios to verify the effectiveness of emergency plans. Through simulation drills, operators can become familiar with the plan process, and the system can also evaluate the execution time and effect of the plan. After the plan is executed, the system collects feedback data, analyzes the execution effect of the plan, and further optimizes it according to actual conditions. Using reinforcement learning, the system can adjust the strategy in the plan based on the results of the drill to improve the efficiency and accuracy of the plan's execution.

[0056] This module uses multi-scale time series analysis to capture both short-term and long-term equipment failure trends, taking into account both sudden failures and chronic problems. Combining the LSTM model for equipment trend prediction can improve the prediction accuracy of complex equipment state changes. The emergency plan has dynamic adjustment capabilities and can be automatically updated according to real-time data and environmental changes to ensure the adaptability and flexibility of the plan. Through multi-scale analysis and dynamic plan generation, this module enhances the system's emergency handling capabilities in complex environments.

[0057] 5. Equipment health status assessment and remaining life prediction module It can evaluate the health status of the main equipment in the substation, predict its remaining service life, and provide maintenance and replacement decision support.

[0058] Equipment health status assessment phase: Through a model based on health indicators, multiple performance indicators of the equipment are comprehensively analyzed to generate the current health score of the equipment. The health score is calculated by weighting the weights of each indicator. For example, temperature and current anomalies may have a greater impact on the health score, while slight vibrations may have a smaller impact. Use ensemble learning algorithms such as random forests or extreme gradient boosting to train health assessment models based on historical data. These algorithms generate health scores for equipment and predict their possible degradation trends by analyzing the relationship between past failure modes of the equipment and current data. Use the hierarchical analysis method to dynamically adjust the weights of each indicator to ensure the accuracy of equipment health assessment under different operating environments. For example, in a high temperature environment, the weight of the impact of temperature on equipment health may increase.

[0059] Remaining life prediction stage: Based on the long short-term memory network (LSTM) or recurrent neural network (RNN) model in deep learning, the remaining life of the equipment is predicted using time series data. By analyzing the data change trend of each sensor of the equipment, the degradation trajectory of the equipment is predicted and the estimated time of the remaining life is given. Combined with external factors such as the operating environment and load conditions of the equipment, the Bayesian inference model is used to give a probabilistic prediction of the equipment life under different conditions. For example, under high load and high temperature, the equipment life may be significantly shortened, and the system will generate multiple life prediction curves based on different external conditions.

[0060] Adaptive maintenance strategy generation phase: Generate adaptive maintenance strategies based on the equipment health status and remaining life prediction results. The system will recommend whether preventive maintenance, calibration, or component replacement is needed based on the current health score and predicted life of the equipment. For equipment with low health scores and insufficient remaining life, the system will prioritize maintenance. Dynamically adjust maintenance strategies through reinforcement learning algorithms. The system will learn the actual use of equipment and maintenance effects, and continuously optimize maintenance cycles and content. For example, the system will record the equipment status after each maintenance, evaluate whether the maintenance measures have extended the life of the equipment, and adjust subsequent maintenance plans.

[0061] Equipment health trend analysis and visualization stage: Through time series analysis, the system generates equipment health trend charts to show the historical changes in equipment health scores and future forecast trends. Operators can view equipment status changes through a visual interface and foresee possible equipment degradation or failures in advance. It provides a function to compare equipment health status in different time periods to help operators analyze equipment performance differences in different time periods. For example, you can compare equipment health score changes in winter and summer to analyze seasonal impacts.

[0062] Decision support and optimization stage: The system works in conjunction with the asset management platform to provide decision-making recommendations for equipment replacement, upgrade, and maintenance based on health status assessment and remaining life prediction. The system can also combine the historical cost data of the equipment to provide the best maintenance or replacement plan to balance equipment maintenance costs and operational risks. Through multi-objective optimization algorithms, the system can find a balance between equipment life, operational risks, and maintenance costs, and provide decision makers with the best maintenance time and plan recommendations. For example, a particle swarm optimization algorithm (PSO) or a genetic algorithm (GA) is used for solution optimization.

[0063] This module combines deep learning with Bayesian inference models to accurately predict the remaining life of equipment and adapt to changes in equipment life under different operating conditions. It uses reinforcement learning to dynamically optimize equipment maintenance strategies to ensure that equipment maintenance decisions can self-adjust as equipment status changes. It provides a maintenance decision system supported by multi-objective optimization algorithms to achieve the optimal balance between operating risks and costs. This module enhances the maintenance management capabilities of equipment through multi-level health status assessment and remaining life prediction.

[0064] 6. Fault diagnosis and knowledge base update module Through fault diagnosis technology, substation equipment faults are identified and the fault knowledge base is updated to achieve intelligent fault handling.

[0065] Fault diagnosis technology implementation stage: Using rule-based fault diagnosis systems, equipment operation data is analyzed in real time according to pre-defined fault rules and conditions. For example, when the temperature exceeds the set threshold, the system will trigger the relevant fault alarm. Combined with fuzzy logic systems to handle uncertainty, it is able to handle fuzzy and imprecise data. For example, for the vibration of the equipment, the fuzzy logic system can evaluate the different ranges of vibration values ​​and their impact on the health of the equipment. Machine learning methods such as support vector machines or decision trees are used to classify and predict equipment failures. By analyzing historical fault data, models are built to identify similar fault types, thereby achieving automated fault diagnosis.

[0066] Fault analysis and decision support phase: When a fault is detected, the system generates a detailed fault analysis report that includes the fault type, possible causes and influencing factors, and provides repair suggestions. For example, if voltage instability is detected, the report may indicate that it is caused by transformer overload. The system provides decision support tools for fault handling to help operators quickly judge and select appropriate treatment measures when encountering a fault. For example, based on historical fault data, the system can recommend priority replacement of parts or calibration in specific fault situations.

[0067] Knowledge base update and self-learning stage: Every time a fault is handled, the system will record the new fault case and handling results into the fault knowledge base to form a dynamically updated knowledge base. The knowledge base contains information such as fault type, historical cases, handling strategies and results. Through deep learning algorithms (such as long short-term memory network LSTM), the system can analyze historical fault data in the knowledge base, extract valuable information, help discover new fault modes and update diagnostic models. For example, the system can automatically identify the root causes of common faults and their solutions.

[0068] Failure mode and trend analysis phase: Use data mining technology to analyze historical fault data and identify trends in fault patterns. The system clusters similar faults together through cluster analysis to help maintenance personnel discover the patterns of fault occurrence. For example, you can analyze the high incidence of faults in a specific season or under specific load conditions. Provide a visualization interface for fault patterns to help engineers intuitively understand the distribution and trends of fault patterns. For example, use a heat map to show the relationship between the frequency and time of fault occurrence, or use a time series graph to show the seasonal changes in fault occurrence.

[0069] Feedback and improvement mechanism: A feedback mechanism is established. After each fault handling, the operator can provide feedback on the system's fault diagnosis results and handling suggestions. The system will optimize the diagnosis model and knowledge base content based on the feedback information. This feedback mechanism will improve the intelligence level of the system. The system regularly evaluates the completeness and accuracy of the knowledge base, reviews and adjusts it according to new fault cases, and ensures that the knowledge base always reflects the latest equipment status and fault handling information.

[0070] This module combines fuzzy logic and machine learning methods to enhance the intelligence and accuracy of fault diagnosis and can handle complex and uncertain fault conditions. The dynamically updated knowledge base can continuously improve fault handling capabilities and self-learn and improve in fault pattern identification and analysis. It provides visualization tools to help operators quickly identify fault patterns and trends and improve decision-making efficiency. This module enhances the reliability and safety of substation equipment through intelligent fault diagnosis and knowledge base update mechanisms.

[0071] 7. User interaction and remote monitoring management module Implement a user-friendly interface for substation equipment, support remote monitoring and management, and improve operation and maintenance efficiency and safety.

[0072] User interface design and function planning stage: Design a user-friendly interface to ensure that operators can easily view equipment status, historical data, and fault alarms. Use responsive design so that the interface can be displayed smoothly on different devices (such as desktops, tablets, and mobile phones). Plan module functions, including real-time data monitoring, fault alarms, historical data queries, maintenance record management, equipment health status assessment, etc., to ensure that the functions cover user needs.

[0073] Data visualization and interaction design stage: Implement data visualization, use charts to display equipment operating status and historical data, and help users quickly obtain equipment health information. Visualization tools can use open source libraries such as D3.js or Chart.js. Design interactive dashboards so that users can select different equipment and time periods by clicking icons or sliders, and dynamically update and display corresponding data. For example, users can choose to view the temperature changes of a transformer in the past month.

[0074] Remote monitoring and alarm system stage: Build a remote monitoring system, use IoT technology to collect device data in real time and upload it to the cloud or local server to ensure the security and real-time nature of data transmission. Use MQTT or HTTP protocol for data transmission. Implement real-time alarm function. When the device status is abnormal or a fault occurs, the system can send an alarm to notify the operator immediately. The alarm method can be pushed via SMS, email or mobile app to ensure that the information can be quickly conveyed to relevant personnel.

[0075] Remote control and management function stage: Provides remote control function, users can control the device through the interface, such as start, stop, restart and other operations. All operations will be recorded in the system log to ensure the traceability of the operation. Implements device parameter configuration function, users can adjust the operating parameters of the device (such as threshold settings) as needed, and it will take effect immediately. The system will verify all configuration changes to avoid device failures caused by improper settings.

[0076] Historical data query and analysis stage: Develop a historical data query function so that users can query historical data by time, equipment type, fault type, etc., to help analyze equipment performance and fault causes. The system should support multiple data export formats for subsequent analysis. Provide data analysis tools to support trend analysis and prediction of historical data to help users identify potential problems in equipment operation. For example, by statistically analyzing historical fault data, identify the high incidence period of a certain type of fault and formulate preventive measures in advance.

[0077] User feedback and optimization mechanism stage: Establish a user feedback mechanism so that users can provide feedback and suggestions at any time during use. The system will collect this feedback regularly and optimize the user interface and function settings accordingly. Through user behavior analysis, the system can automatically identify functions that users use frequently and optimize these functions to improve the user experience. For example, if a function is frequently used by users, the system can highlight it in the interface.

[0078] Security and permission management stage: Implement user rights management. The system sets different access rights according to user roles to ensure that important operations and data are only open to specific personnel. Rights management can be divided into roles such as administrators, operation and maintenance personnel, and ordinary users. Ensure the security of data transmission and storage, and use encryption technology (such as SSL / TLS) to protect the security of data during network transmission to prevent data leakage or tampering.

[0079] This module uses IoT technology to achieve real-time remote monitoring and management, improving the management efficiency and response speed of substation equipment. Combined with data visualization and interactive design, users can easily obtain equipment information and perform management operations, enhancing user experience. It has security and permission management functions to ensure the security of the system and the traceability of operations, reducing potential risks. This module improves the operation and maintenance efficiency and safety of substation equipment through user-friendly interactive design and remote monitoring and management functions.

[0080] 8. Intelligent prediction and optimization scheduling module Through intelligent prediction and optimization algorithms, the dispatching efficiency and resource utilization of substation equipment can be improved to ensure efficient operation of equipment.

[0081] Load forecasting and trend analysis phase: Use time series analysis methods (such as ARIMA, LSTM, etc.) to predict the load demand of equipment. Use historical load data to build a model to predict future load change trends so that scheduling strategies can be formulated in advance. Analyze the impact of different seasons and time periods on load, such as the peak power consumption in summer and winter. The model will automatically adjust the weight of the prediction algorithm to improve the accuracy of the prediction.

[0082] Scheduling strategy generation and optimization phase: Based on the load forecast results, the system generates a preliminary dispatching plan. The dispatching plan includes equipment start-up and shutdown plans, load distribution, etc., to ensure that the equipment can operate normally when the load is high, and some equipment can be reasonably shut down when the load is low. The preliminary dispatching plan is optimized using optimization algorithms such as genetic algorithms or particle swarm optimization. The algorithm will consider factors such as equipment operating costs, maintenance requirements, energy prices, etc., to find the optimal dispatching plan to maximize economic benefits.

[0083] Resource allocation and dynamic adjustment stage: The dispatching plan is adjusted dynamically according to the real-time operation status and external conditions. For example, when a certain device fails or the load demand suddenly increases, the system can quickly adjust the operation plan to ensure the stability of power supply. The system can automatically select the best combination of equipment for operation to meet the current load demand at the lowest cost. For example, during periods of high electricity prices, lower-cost equipment is used for power supply first.

[0084] Simulation and optimization evaluation phase: Provides a simulation function for scheduling plans. Users can simulate the plan before executing the scheduling to evaluate its feasibility and effect. By simulating different operating scenarios, users can better understand the advantages and disadvantages of the scheduling plan. Establish a performance evaluation system to analyze the effect of the scheduling plan after implementation. Evaluation indicators include cost savings, equipment operating efficiency, failure rate, etc., in order to provide a basis for the optimization of subsequent scheduling plans.

[0085] User feedback and iterative improvement phase: The system collects user feedback during the scheduling process and analyzes user satisfaction and suggestions on the scheduling effect. Based on the actual experience of users, the scheduling strategy and algorithm parameters are optimized. The scheduling strategy is continuously learned and optimized using machine learning technology. The system will automatically adjust the model parameters to adapt to different operating conditions and load patterns, and improve the accuracy of prediction and scheduling.

[0086] Visual scheduling interface and report generation stage: Develop a visual dispatch interface so that users can intuitively view the current equipment dispatch status, load forecast results and dispatch plans. Display real-time data and early warning information through the dashboard to help users make decisions quickly. The system can automatically generate dispatch reports, record the dispatch decision process, implementation effects and optimization suggestions, and help management conduct subsequent reviews and decisions.

[0087] This module combines time series analysis and optimization algorithms to achieve intelligent load forecasting and scheduling, and improve the utilization rate of substation resources. Dynamic adjustment and real-time feedback mechanisms enhance the flexibility and adaptability of scheduling, ensuring stable operation in complex environments. It provides a visual scheduling interface and automatic report generation functions to improve user experience and management efficiency, and help operation and maintenance personnel make scientific decisions. This module significantly improves the operating efficiency and economic benefits of substations through intelligent prediction and optimized scheduling.

[0088] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. The substation main equipment self-test system based on multi-source heterogeneous data fusion technology is characterized by: The self-check system includes data acquisition and preprocessing module, adaptive data fusion algorithm module, equipment fault mode dynamic identification and intelligent repair module, multi-scale time series prediction and emergency plan generation module, equipment health status assessment and remaining life prediction module, fault diagnosis and knowledge base update module, user interaction and remote monitoring management module, intelligent prediction and optimization scheduling module; The data acquisition and preprocessing module is used to collect electrical signal data and environmental data on the main equipment of the substation and perform corresponding preprocessing; The adaptive data fusion algorithm module is used to adjust the weights of different data sources in real time to improve the fusion accuracy and robustness of multi-source heterogeneous data; The equipment failure mode dynamic identification and intelligent repair module is used to identify known and unknown failure modes of substation equipment and provide self-repair suggestions; The multi-scale time series prediction and emergency plan generation module is used to predict fault trends at different time scales and generate dynamic emergency plans; The equipment health status assessment and remaining life prediction module is used to assess the health status of the main equipment of the substation, predict its remaining service life, and provide maintenance and replacement decision support; The fault diagnosis and knowledge base update module is used to identify substation equipment faults and update the fault knowledge base to achieve intelligent fault handling; The user interaction and remote monitoring management module is used to implement a user-friendly interface for substation equipment and support remote monitoring and management; The intelligent prediction and optimization scheduling module is used for intelligent prediction and optimization algorithms to improve the scheduling efficiency and resource utilization of substation equipment.

2. The substation main equipment self-checking system based on multi-source heterogeneous data fusion technology according to claim 1 is characterized by: The data processing method of the adaptive data fusion algorithm module includes the following steps: S1, extracting data features; S2, adaptively adjusting the weight of the data source; S3, fusing and outputting data; S4, self-learning and optimizing the algorithm model.

3. The substation main equipment self-checking system based on multi-source heterogeneous data fusion technology according to claim 2 is characterized by: The method for adaptive weight adjustment of the data source in step S2 is to first build an adaptive fusion model, where the model input is the feature vector extracted by each sensor, dynamically adjust the weight of each sensor by comparing historical data with real-time data, embed an adaptive weight adjustment layer in the deep learning model, and use a loss function based on mean square error or cross entropy to optimize the model and train it to ensure that the weight of each data source is adaptively adjusted with factors such as environmental changes and sensor status.

4. The substation main equipment self-checking system based on multi-source heterogeneous data fusion technology according to claim 2 is characterized in that: In step S3, the data fusion method is to sum the feature vectors of different sensors according to the weights to generate a fused feature vector, and the data output method is to input the fused feature into a classifier or regression model to obtain the state prediction or fault detection result of the equipment.

5. The substation main equipment self-checking system based on multi-source heterogeneous data fusion technology according to claim 1 is characterized in that: The identification and repair method of the equipment fault mode dynamic identification and intelligent repair module includes the following steps: S1, fault mode identification; S2, fault mode dynamic update and expansion; S3, intelligent repair suggestion generation; S4, self-optimization and feedback.

6. The substation main equipment self-checking system based on multi-source heterogeneous data fusion technology according to claim 1 is characterized by: The dynamic emergency plan generation method of the multi-scale time series prediction and emergency plan generation module includes the following steps: S1, time series feature extraction; S2, fault trend prediction; S3, generating an emergency plan according to the fault prediction result; S4, performing dynamic emergency plan optimization; S5, performing simulation drills and plan verification.

7. The substation main equipment self-checking system based on multi-source heterogeneous data fusion technology according to claim 1 is characterized by: The evaluation and prediction method of the equipment health status evaluation and remaining life prediction module includes the following steps: S1, performing equipment health status evaluation; S2, performing remaining life prediction; S3, generating an adaptive maintenance strategy; S4, analyzing and visualizing equipment health trends; S5, decision support and optimization.

8. The substation main equipment self-checking system based on multi-source heterogeneous data fusion technology according to claim 1 is characterized by: The intelligent fault handling method of the fault diagnosis and knowledge base update module includes the following steps: S1, fault diagnosis technology implementation; S2, fault analysis and decision support; S3, knowledge base update and self-learning; S4, fault mode and trend analysis; S5, feedback and improvement.

9. The substation main equipment self-checking system based on multi-source heterogeneous data fusion technology according to claim 1 is characterized in that: The prediction and optimization method of the intelligent prediction and optimization scheduling module includes the following steps: S1, load prediction and trend analysis; S2, scheduling strategy generation and optimization; S3, resource allocation and dynamic adjustment; S4, simulation and optimization evaluation; S5, user feedback and iterative improvement; S6, feeding back the results to the user's visual scheduling interface and generating a report.

10. The substation main equipment self-checking system based on multi-source heterogeneous data fusion technology according to claim 9 is characterized in that: The load forecasting and trend analysis method in step S1 is to use the time series analysis method to predict the load demand of the equipment, and then use the historical load data to build a model to predict the future load change trend and analyze the impact of different seasons and time periods on the load; the scheduling strategy generation and optimization method in step S2 is that according to the load forecast results, the system generates a preliminary scheduling plan, and then uses optimization algorithms such as genetic algorithms or particle swarm optimization to optimize the preliminary scheduling plan.

Images