Power grid equipment information visualization method, system, equipment and medium

Abstract

The invention relates to the technical field of information, and provides a power grid equipment information visualization method and system, equipment and a medium, and the method comprises the steps: carrying out the region division of main network equipment through employing a hierarchical clustering algorithm according to the attribute of the main network equipment; when it is monitored that the line loss rate of a certain line exceeds a preset line loss rate threshold value, automatically obtaining region division information of equipment to which the corresponding line belongs, and extracting historical operation data of the equipment from a cloud database of the equipment in the corresponding region; an Apriori association rule mining algorithm is adopted to analyze the relevance between the historical operation data of the equipment and the line loss rate abnormity, and an equipment defect and line loss rate relevance analysis result is obtained; and acquiring real-time monitoring data of equipment corresponding to the line with the abnormal line loss rate, and verifying and correcting an equipment defect reason judgment result by comparing and analyzing the real-time monitoring data with historical operation data of the equipment and adopting an isolation forest anomaly detection algorithm. According to the invention, rapid positioning of line loss rate abnormity and accurate prediction and visual management of equipment defects are realized.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a method, system, device and medium for visualizing information of power grid equipment. Background Art

[0002] Abnormal line loss rate is a complex problem in the operation of power systems. When the line loss rate suddenly increases, it is crucial to quickly and accurately locate the cause of the fault and predict possible equipment defects, but this process faces many challenges. First, the main network equipment is numerous, widely distributed, and highly interrelated. It is difficult to find the root cause of the problem in time by traditional manual troubleshooting alone. Secondly, line loss abnormality may be caused by the superposition of multiple factors, such as equipment aging, load changes, weather effects, etc., and a large amount of historical and real-time data needs to be comprehensively analyzed to accurately judge. Furthermore, even if suspicious equipment is found, it is difficult to predict the specific defect type and severity, which affects the scientific nature of maintenance decisions. In addition, due to the lack of intuitive visualization methods, it is difficult for operation and maintenance personnel to quickly grasp the overall situation and development trend of line loss abnormality. These factors lead to inefficient line loss management, affecting the safe and stable operation of the power grid and the reliable power supply of users. How to quickly locate line loss abnormalities, accurately predict equipment defects, and intuitively present the analysis results in a complex main network environment has become a technical problem that needs to be solved urgently. Summary of the invention

[0003] The present invention provides a method for visualizing power grid equipment information, which mainly includes: According to the attributes of the main network devices, a hierarchical clustering algorithm is used to divide the main network devices into regions; When the line loss rate of a certain line is monitored to exceed the preset line loss rate threshold, the regional division information of the equipment to which the corresponding line belongs is automatically obtained, and the historical operation data of the equipment is extracted from the cloud database of the equipment in the corresponding area; The Apriori association rule mining algorithm is used to analyze the correlation between the historical operation data of the equipment and the abnormal line loss rate, and the correlation analysis results between equipment defects and line loss rate are obtained; The Bayesian network model is trained based on the results of the correlation analysis between equipment defects and line loss rate. Based on the trained Bayesian network model and the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, the probability of the equipment defect cause causing the abnormal line loss rate is estimated and preliminarily judged. Combined with the historical maintenance work orders, the judgment result of the equipment defect cause is obtained. Based on the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, through comparative analysis with the historical operation data of the equipment, the isolation forest anomaly detection algorithm is used to further verify and correct the judgment result of the cause of the equipment defect; According to the corrected equipment defect cause judgment results, combined with the equipment attributes of the corresponding equipment, the decision tree algorithm is used to match similar cases from the equipment failure case library, analyze and predict the specific defect type of the equipment, and obtain the equipment defect analysis and prediction results; The rule-based expert system evaluates the equipment defect analysis and prediction results, generates a priority ranking for equipment maintenance or replacement, and automatically generates corresponding equipment maintenance or replacement suggestions. The equipment defect analysis and prediction results and suggestions are visualized through the work order management system.

[0004] The present application also proposes a power grid equipment information visualization system, the system comprising: Device area division module: used to divide the main network devices into regions according to the attributes of the main network devices using a hierarchical clustering algorithm; Historical operation data extraction module: when the line loss rate of a certain line is monitored to exceed the preset line loss rate threshold, it automatically obtains the regional division information of the equipment to which the corresponding line belongs, and extracts the historical operation data of the equipment from the cloud database of the equipment in the corresponding area; Association rule mining module: used to train the Bayesian network model according to the results of the correlation analysis between equipment defects and line loss rate. Based on the trained Bayesian network model and the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, the probability of the equipment defect cause causing the abnormal line loss rate is estimated and preliminarily judged. Combined with the historical maintenance work order, the judgment result of the equipment defect cause is obtained. Anomaly detection and verification module: Based on the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, through comparative analysis with the historical operation data of the equipment, the isolation forest anomaly detection algorithm is used to further verify and correct the judgment result of the cause of the equipment defect; Defect type prediction module: used to judge the cause of the equipment defect according to the corrected result, combined with the equipment attributes of the corresponding equipment, using the decision tree algorithm, matching similar cases from the equipment failure case library, analyzing and predicting the specific defect type of the equipment, and obtaining the equipment defect analysis and prediction results; Maintenance suggestion generation and visual feedback module: used for rule-based expert systems to evaluate equipment defect analysis and prediction results, generate equipment maintenance or replacement priority rankings, and automatically generate corresponding equipment maintenance or replacement suggestions, and visualize the equipment defect analysis and prediction results and suggestions through the work order management system.

[0005] The present application also proposes a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0006] The present application also proposes a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of any one of the above methods when executed by a processor.

[0007] The technical solution provided by the embodiment of the present invention may have the following beneficial effects: The present invention discloses a method for analyzing the abnormal line loss rate of main network equipment and predicting defects. When an abnormal increase in the line loss rate of a line is detected, the present invention automatically obtains the regional division information and historical operation data of the equipment to which the corresponding line belongs, and uses the Apriori algorithm to analyze the correlation between the equipment historical data and the abnormal line loss rate. Combined with the equipment maintenance work order data, a Bayesian network is used to make a preliminary judgment on the cause of the equipment defect that causes the abnormal line loss rate. By obtaining the real-time monitoring data of the corresponding equipment, the isolation forest algorithm is used to further verify and correct the defect cause judgment result. Based on the equipment attributes, a decision tree is used to match similar cases from a fault case library to predict the specific defect type. Finally, the present invention uses a rule expert system to evaluate the analysis results, generate equipment maintenance priority ranking and suggestions, and visually feedback to the operation and maintenance personnel through the work order management system. The method realizes the rapid positioning of line loss rate anomalies and the accurate prediction of equipment defects, thereby improving the efficiency and reliability of power grid operation and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 The present invention is a flowchart of a method for visualizing power grid equipment information.

[0009] Figure 2 The present invention is a schematic block diagram of a power grid equipment information visualization system.

[0010] Figure 3 The figure is a schematic diagram of the structure of a computer device of the present invention. DETAILED DESCRIPTION

[0011] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0012] like Figure 1 , a method for visualizing power grid equipment information in this embodiment may specifically include: S101. According to the attributes of the main network devices, a hierarchical clustering algorithm is used to divide the main network devices into regions.

[0013] Acquire the multi-dimensional attribute data of the main network equipment and construct the attribute feature vector; calculate the similarity between the devices based on the device attribute feature vector; cluster the devices using the hierarchical clustering algorithm based on the similarity; analyze the attribute characteristics of each category of equipment based on the clustering results and determine the regional division plan; determine the regional affiliation of each device based on the regional division plan; and obtain the regional division results of the main network equipment.

[0014] Specifically, obtaining multi-dimensional attribute data of main network equipment is the basis for equipment analysis and classification. Taking the power main network equipment as an example, the attribute data of the equipment such as voltage level, capacity, service life, and failure rate can be collected. For example, the attributes of a transformer may include: voltage level 500kV, capacity 1000MVA, service life 15 years, and annual average failure rate 0.5 times. These multi-dimensional attributes together constitute the feature vector of the equipment. Calculating the similarity between devices based on the device attribute feature vector is the key to clustering. Commonly used similarity calculation methods include Euclidean distance, cosine similarity, etc. Taking cosine similarity as an example, the cosine value of the angle between the feature vectors of two devices can be calculated. The closer the value is to 1, the more similar it is. For example, the similarity of two transformers is 0.95, which means that their attributes are very similar. Clustering equipment using a hierarchical clustering algorithm can reveal the hierarchical relationship between equipment. The process of hierarchical clustering is to continuously merge the most similar categories from bottom to top until the preset number of categories is reached. For example, substation equipment may eventually be clustered into several major categories such as "main transformer", "circuit breaker", and "transformer". Analyzing the characteristics of the equipment attributes of each category will help determine a reasonable regional division scheme. For example, the common characteristics of the "main transformer" are large capacity and high voltage level, which may be suitable for division into the "core area"; while the "transformer" has a small capacity and may be suitable for division into the "peripheral area". This regional division based on equipment characteristics is more in line with actual operation needs. Determining the regional affiliation of each device is an important step in achieving regional management. Each device can be classified into the most similar area based on its attribute characteristics. For example, a newly added 500kV transformer is likely to be classified into the "core area" based on its high voltage level characteristics. Further analysis of the functional attributes of the equipment in each area can determine a more refined functional module division scheme. For example, in the "core area", "power conversion module", "power control module", etc. can be divided according to the specific use of the equipment. This functional module division is conducive to more professional equipment management and maintenance. Finally, the functional module affiliation of each device is determined according to the functional module division scheme, and the regional division result of the main network equipment is obtained. For example, a 500kV transformer is not only divided into the "core area" but also belongs to the "power conversion module". This multi-level division result provides a more comprehensive perspective for equipment management. This equipment classification method based on data analysis has many advantages. First, it can objectively reflect the actual characteristics of the equipment and avoid the subjectivity and inconsistency that may be caused by artificial division. Secondly, through cluster analysis, the potential similarity relationship between equipment can be discovered, which is helpful to optimize the equipment management strategy. Furthermore, the division of regions and functional modules provides a basis for differentiated management, and more targeted maintenance and operation plans can be formulated for different types of equipment. In addition, this method also has good scalability and adaptability.As new equipment is added or the properties of existing equipment change, the classification results can be dynamically updated to ensure that the classification always reflects the latest equipment status. This is particularly important for a complex system that is constantly evolving, such as the power system. In general, this equipment classification method based on multidimensional attribute data achieves scientific and refined management of main grid equipment in a data-driven manner. It not only improves the efficiency and accuracy of equipment management, but also provides data support for the full life cycle management of equipment, which is conducive to improving the operational reliability and economy of the entire power system.

[0015] S102. When it is monitored that the line loss rate of a certain line exceeds a preset line loss rate threshold, the regional division information of the equipment to which the corresponding line belongs is automatically obtained, and the historical operation data of the equipment is extracted from the cloud database of the equipment in the corresponding area.

[0016] Based on the installation of smart meters and sensors on each line, the operating parameters of the line are collected in real time, and the collected data is uploaded to the cloud database for storage. The historical operating data in the cloud database is mined and analyzed using big data analysis technology. By establishing a line loss rate prediction model, the line loss rate of each line is calculated in real time and compared with the preset line loss rate threshold. If the real-time line loss rate of a line exceeds the preset threshold, it is determined that the corresponding line has a line loss abnormality, triggering a line loss abnormality alarm, and automatically obtaining the regional division information of the equipment to which the corresponding line belongs. Based on the obtained equipment regional division information, the equipment historical operation data of all equipment in the corresponding area is extracted from the cloud database.

[0017] Specifically, smart meters and sensors are key equipment for real-time monitoring of power lines. Taking a city's distribution network as an example, a group of smart meters and current and voltage sensors are installed every 500 meters on a 10kV line. These devices collect the current, voltage, active power and other parameters of the line every 5 minutes, and upload the data to the cloud database through a wireless communication network. The cloud database adopts a distributed architecture and can store massive historical operation data. Through big data analysis technology, the law of line operation can be mined. For example, the average load of a line during the day on weekdays is 2MW and 1MW at night; the load on weekends is relatively low. Based on these laws, a line loss rate prediction model can be established. The line loss rate prediction model uses machine learning algorithms, such as random forests or neural networks, with inputs including time, weather, load and other features, and outputs the predicted line loss rate. The model is regularly trained with the latest data to adapt to changes in line operation conditions. The system calculates the real-time line loss rate every 15 minutes and compares it with a preset threshold (such as 5%). If the real-time line loss rate of a line exceeds the threshold, the system will trigger an abnormal line loss alarm. At the same time, the regional information to which the corresponding line belongs is automatically queried. For example, an abnormal line belongs to the eastern commercial district of the city. The system then extracts the historical operation data of all equipment in the area from the cloud database, including the operating status and maintenance records of equipment such as transformers and switches. This line loss management method based on big data and artificial intelligence has many advantages. First, real-time monitoring and automatic alarm can quickly detect line loss anomalies, which is conducive to timely handling of faults and improving power supply reliability. Secondly, by analyzing historical data, potential problems such as equipment aging and uneven load distribution can be discovered, providing a basis for preventive maintenance. Thirdly, accurate line loss calculation helps to optimize power grid operation, reduce line loss rate and improve economic benefits. In practical applications, this method can also be combined with other technologies. For example, the location of line loss anomalies can be visualized using a geographic information system (GIS), and the fault point can be quickly located in combination with drone inspections. In addition, by comparing line loss rates in different regions and different time periods, decision support can be provided for power grid planning and transformation. With the development of technologies such as 5G and the Internet of Things, larger-scale and more accurate power grid monitoring and management can be achieved in the future, further improving the level of intelligence of the power grid.

[0018] S103. Using the Apriori association rule mining algorithm, the correlation between the historical operation data of the equipment and the abnormal line loss rate is analyzed to obtain the correlation analysis result between the equipment defects and the line loss rate.

[0019] Based on the acquired historical operation data of the equipment, a data set is constructed and preprocessed. The Apriori association rule mining algorithm is used to set the support and confidence thresholds to mine the association rules between the equipment operation parameters and the abnormal line loss rate in the historical operation data of the equipment. According to the mined association rules, the association strength between the equipment operation parameters and the abnormal line loss rate is judged to obtain the correlation analysis results between equipment defects and line loss rate.

[0020] Specifically, constructing a data set is the basis for data analysis. In the power system, the historical operation data of equipment usually includes parameters such as voltage, current, and power factor. The preprocessing process may involve steps such as data cleaning and normalization to ensure data quality and consistency. For example, for a 10kV distribution line, the operation data of the past year may be collected, including parameters such as voltage, current, active power, reactive power, and the corresponding line loss rate per hour. The Apriori algorithm is a classic association rule mining algorithm that is suitable for discovering frequent item sets in large data sets. In line loss analysis, the equipment operation parameters can be used as item sets and the line loss rate anomaly can be used as the target item. The support indicates the frequency of an item set in the entire data set, while the confidence indicates the probability of the subsequent item appearing when the previous item appears. For example, setting the support threshold to 0.1 and the confidence threshold to 0.8 means that only the rules that appear in at least 10% of the samples and the probability of the subsequent item appearing with more than 80% when the previous item appears are focused on. Through the Apriori algorithm, the following association rule may be found: when the voltage fluctuation exceeds the rated value by 5% and the power factor is less than 0.9, the probability of abnormal line loss rate is 85%. The support of this rule may be 0.15, indicating that this situation is observed in 15% of the samples; the confidence is 0.85, indicating that when the premise conditions are met, abnormal line loss rate will occur in 85% of the cases. The judgment of association strength is usually based on indicators such as support, confidence, and lift. Lift represents the ratio of the probability of the latter item occurring under the condition of the former item to the probability of the latter item occurring. For example, if the basic probability of abnormal line loss rate is 10%, and the probability is 85% when the voltage fluctuation is large and the power factor is low, the lift is 8.5, indicating that this rule has strong predictive ability. Through this analysis, the correlation results between equipment defects and line loss rate can be obtained. For example, it may be found that the transformer load factor is too high (such as more than 80%) and there is a strong correlation with abnormal line loss rate, which suggests that the heavily loaded transformer may need to be upgraded or the load transferred. For example, it is found that the insufficient cross-sectional area of ​​conductors in certain sections is highly correlated with abnormal line loss rate, which may mean that line reconstruction is needed. This method based on data mining is more objective and comprehensive than traditional manual experience judgment. It can discover potential laws that are difficult for humans to detect from massive data, and provide data support for power grid operation and maintenance decisions. For example, through this analysis, it may be found that some seemingly unrelated factors (such as ambient temperature, load type, etc.) have unexpected correlations with abnormal line loss rate, thus providing new ideas for refined management. However, it should be noted that association rules are not equivalent to causal relationships. For example, even if it is found that low power factor is highly correlated with abnormal line loss rate, it cannot be directly concluded that low power factor causes high line loss. This may be due to common potential factors (such as equipment aging) affecting both at the same time. Therefore, in practical applications, further analysis and verification are required in combination with professional knowledge.

[0021] S104. Train a Bayesian network model based on the results of the correlation analysis between equipment defects and line loss rate. Based on the trained Bayesian network model and the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, make a probability estimate and preliminary judgment on the cause of the equipment defect that causes the abnormal line loss rate. Combined with historical maintenance work orders, obtain the judgment result of the cause of the equipment defect.

[0022] Based on the results of the correlation analysis between equipment defects and line loss rate, the equipment defect causes and the first line loss rate data are extracted to construct a training data set of the Bayesian network model; wherein the first line loss rate data includes the historical operation data of the equipment corresponding to the abnormal line loss rate; according to the training data set of the Bayesian network model, the prior probability and conditional probability table of the Bayesian network model are estimated by the parameter learning algorithm, and the probability dependence relationship between the equipment defect cause and the abnormal line loss rate is established; if the abnormal line loss rate currently occurs, the second line loss rate data is input into the trained Bayesian network model, and the posterior probability of each equipment defect cause is calculated by the inference algorithm to obtain the probability distribution of the equipment defect cause that causes the abnormal line loss rate; wherein the second line loss rate data includes the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate; according to the posterior probability of the equipment defect cause and the preset probability threshold, the equipment defect cause of the abnormal line loss rate is preliminarily judged, and the equipment defect cause that is most likely to cause the abnormality is determined; the preliminarily judged equipment defect cause is matched with the equipment maintenance work order, the accuracy of the judgment result is verified by the historical maintenance work order, and the judgment result is optimized and adjusted to obtain the judgment result of the equipment defect cause. The real-time detection data and the historical operation data of the equipment both belong to the operation data of the equipment.

[0023] Specifically, in the power system, abnormal line loss rate is an important problem that needs to be discovered and solved in time. To this end, a correlation analysis model between equipment defects and line loss rate can be established. First, obtain equipment operation data and historical maintenance work orders, which contain information such as equipment operation parameters, fault records, and maintenance conditions. Data preprocessing is the basis for model establishment. The original data needs to be cleaned to remove outliers and missing values. For example, an outlier of -50°C appears in the temperature data of a transformer, which needs to be removed. Then data integration is performed to integrate data from different sources together. For example, the voltage, current and other parameters in the equipment operation data are associated with the fault type, maintenance time and other information in the maintenance work order. Finally, data conversion is performed to convert the original data into a format suitable for model use. For example, the fault type is converted into a numerical code, such as "insulation fault" is coded as 1, "mechanical fault" is coded as 2, etc. The Bayesian network model is a probabilistic graphical model that is suitable for analyzing causal relationships in complex systems. In this example, a Bayesian network is used to establish the relationship between equipment defects and abnormal line loss rate. First, a network structure is constructed based on the preprocessed data. The nodes represent variables (such as equipment parameters, fault types, line loss rates, etc.), and the edges represent the dependencies between variables. Then, the probability distribution of the model is estimated through the parameter learning algorithm. For example, it may be found that when the transformer oil temperature exceeds 80°C, the probability of abnormal line loss rate increases significantly. When the line loss rate is abnormal, the abnormal data is input into the trained Bayesian network. Through the inference algorithm, various possible causes of equipment defects and their probabilities can be calculated. For example, the model may give the result that the probability of transformer insulation aging is 60%, the probability of poor contact is 30%, and the probability of other causes is 10%. According to the preset probability threshold (such as 50%), it can be preliminarily judged that transformer insulation aging is the main cause of abnormal line loss rate. In order to verify the accuracy of the judgment result, the result of the preliminary judgment is matched with the historical maintenance work order. If it is found that similar line loss rate abnormalities have indeed been caused by transformer insulation aging many times in the past, then the judgment confidence can be enhanced. On the contrary, if historical data shows that poor contact is more common, it may be necessary to adjust the model parameters or re-evaluate the judgment criteria. This method can not only quickly locate the possible causes of abnormal line loss rate, but also continuously optimize the model to improve the accuracy of judgment. This is of great significance for preventive maintenance and fault diagnosis of power systems, and can help operation and maintenance personnel to carry out maintenance more targetedly, reduce power outage time, and improve power supply reliability.

[0024] S105. Based on the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, through comparative analysis with the historical operation data of the equipment, an isolation forest anomaly detection algorithm is used to further verify and correct the judgment result of the cause of the equipment defect.

[0025] The real-time monitoring data and historical operation data of the equipment associated with the line with abnormal line loss rate are obtained, and the obtained real-time monitoring data and historical operation data are preprocessed to obtain standardized real-time monitoring data and historical operation data. Based on the standardized real-time monitoring data and the equipment historical operation data, an isolation forest anomaly detection model is constructed. Through the isolation forest anomaly detection algorithm, the abnormality of the standardized real-time monitoring data relative to the standardized equipment historical operation data is calculated to determine whether the standardized real-time monitoring data has an abnormality; if the real-time monitoring data has an abnormality, the corresponding abnormal data is obtained, and the equipment defect cause judgment result is verified based on the abnormal sample of the isolation forest anomaly detection model, and the specific defect cause of the equipment is determined by analyzing the characteristics of the abnormal data; based on the verified specific defect cause of the equipment, the equipment defect cause judgment result is corrected based on the expert knowledge base to obtain a corrected equipment defect cause judgment result.

[0026] Specifically, in the power system, abnormal line loss rate may be caused by a variety of equipment defects. In order to accurately determine the cause of the defect, it is first necessary to obtain the real-time monitoring data and historical operation data of the relevant equipment. These data may include parameters such as transformer oil temperature, load current, and voltage. Data preprocessing is a key step. Standardization can eliminate the influence of different dimensions and make the data comparable. For example, the transformer oil temperature is converted from degrees Celsius to a standard fraction so that it is on the same scale as indicators such as current and voltage. Isolation forest is an efficient anomaly detection algorithm. Its core idea is that abnormal data is easier to be "isolated". In practical applications, the isolation forest model can be constructed using historical operation data. Assume that the historical load current average value of a transmission line is 500A and the standard deviation is 50A. When the real-time monitored current value is 700A, the isolation forest model may judge it as abnormal because the value is far beyond the normal range. By analyzing the characteristics of abnormal data, the cause of equipment defects can be further verified and refined. For example, if the current is found to be abnormally high and the voltage is normal, it may point to problems such as loose wire joints or insulation aging. If the temperature rises abnormally, it may be a fault in the transformer cooling system. These analysis results can be compared with the previous Bayesian network model judgment results to improve the accuracy of diagnosis. The expert knowledge base plays an important role in equipment defect diagnosis. It contains a large number of historical cases and expert experience, which can help correct and improve the judgment results. For example, a diagnosis initially determined that the transformer oil temperature was abnormal, but the expert knowledge base recorded similar cases that may be false alarms caused by temperature sensor failure. Through this correction, unnecessary power outages and maintenance can be avoided, and the accuracy and efficiency of diagnosis can be improved. This multi-level and multi-angle diagnostic method can significantly improve the accuracy of equipment defect cause judgment. By combining statistical models, machine learning algorithms and expert experience, problems can be quickly located, misjudgments can be reduced, and the reliability of power grid operation can be improved. At the same time, this method also provides a basis for predictive maintenance, which helps to detect potential problems in a timely manner and reduce equipment failure rates and line loss rates.

[0027] S106. Based on the corrected equipment defect cause judgment result and in combination with the equipment attributes of the corresponding equipment, a decision tree algorithm is used to match similar cases from the equipment failure case library, analyze and predict the specific defect type of the equipment, and obtain equipment defect analysis and prediction results.

[0028] The C4.5 decision tree algorithm is used to train the corrected equipment defect cause judgment results in the historical case data in the equipment failure case library, and the equipment attribute information of the corresponding equipment is obtained. The corrected equipment defect cause judgment results and equipment attribute information are used as the input of the C4.5 decision tree algorithm to obtain the equipment defect analysis prediction model. From the equipment failure case library, historical failure cases similar to the current equipment to be analyzed are retrieved according to the equipment attribute information. The equipment attribute information of the current equipment is input into the equipment defect analysis prediction model, and the equipment defect analysis prediction model is used to analyze and predict the specific defect type of the equipment; if the confidence of the analysis and prediction result is lower than the preset threshold, the KNN algorithm is used to further match the most similar case from similar historical failure cases according to the equipment attribute information; the prediction results of the equipment defect analysis prediction model and the most similar case matched by the KNN algorithm are combined to make a final analysis and judgment on the specific defect type of the equipment, and the equipment defect analysis and prediction results are obtained.

[0029] Specifically, the equipment defect cause judgment results and equipment attribute information are key inputs for analyzing equipment failures. Equipment attribute information includes equipment type, operating years, rated parameters, etc. For example, for a transformer, its attribute information may include a capacity of 500MVA, an operating year of 15 years, and a cooling method of OFAF. Together with the defect cause judgment results, this information provides rich feature data for the C4.5 decision tree algorithm. The C4.5 decision tree algorithm is a commonly used classification algorithm that classifies data by building a decision tree. In equipment defect analysis, the C4.5 algorithm can effectively use historical case data to learn the relationship between equipment attributes, defect causes, and specific defect types. For example, by analyzing historical data, the algorithm may find that when the transformer has been in operation for more than 20 years and the recent dissolved gas analysis results in the oil are abnormal, there is a high probability that the winding insulation will age. The trained equipment defect analysis prediction model can predict possible specific defect types based on the newly input equipment attributes and defect causes. For example, for a 500kV circuit breaker that has been in operation for 15 years, if the cause of the defect is determined to be "abnormal operating mechanism", the model may predict the specific defect type as "aging of the operating mechanism seal ring". In order to improve the accuracy of the analysis, the system will retrieve similar historical cases from the equipment fault case library. This step uses the equipment attribute information as the retrieval condition to find historical fault cases similar to the current equipment. For example, for the above circuit breaker, the system may retrieve historical fault cases of circuit breakers of the same model and similar operating years. When the confidence of the prediction result of the C4.5 decision tree model is low, the system will use the KNN (K nearest neighbor) algorithm for further analysis. The KNN algorithm finds the most similar case by calculating the distance between the current equipment and the historical case in each attribute dimension. For example, if there is a historical case whose circuit breaker model, operating years, and use environment are very close to the current equipment, and the cause of the fault in the case is determined to be "broken operating mechanism spring", then this case will be selected as the most similar case by the KNN algorithm. Finally, the system will combine the prediction results of the C4.5 decision tree model and the most similar cases matched by the KNN algorithm to obtain the final equipment defect analysis and prediction results. This multi-algorithm combination method can make full use of historical data and expert experience to improve the accuracy and reliability of the analysis. For example, if there is a difference between the "aging of the operating mechanism seal ring" predicted by the C4.5 model and the "fracture of the operating mechanism spring" matched by KNN, the system may further analyze the symptoms and impacts of the two faults, and finally determine the more likely defect type based on the specific situation of the current equipment. This equipment defect analysis method based on machine learning and case reasoning can quickly and accurately identify potential equipment problems, provide an important basis for timely repairs and preventive maintenance, and thus improve the reliability and safety of the power system.

[0030] S107. A rule-based expert system evaluates the equipment defect analysis and prediction results, generates a priority ranking for equipment maintenance or replacement, and automatically generates corresponding equipment maintenance or replacement suggestions, and visualizes the equipment defect analysis and prediction results and suggestions through a work order management system.

[0031] According to the historical operation data of the equipment and the rules in the expert knowledge base, a rule-based expert system is used to analyze the equipment. Combined with factors such as the importance of the equipment and the degree of impact of the fault, a multi-criteria decision-making method is used to evaluate and score the severity of the equipment defects. The equipment defect analysis and prediction results for equipment maintenance or replacement are determined based on the evaluation results to perform evaluation priority sorting. According to the type and severity of the equipment defect, the corresponding maintenance or replacement measures are matched from the expert knowledge base, and targeted equipment maintenance suggestions or replacement suggestions are automatically generated, and the suggestions are written into the corresponding work order tasks. Through the work order management system, the expert system's equipment defect analysis prediction results, the maintenance or replacement priority sorting obtained by the evaluation, and the automatically generated maintenance or replacement suggestions are visualized to form an intuitive equipment health report and operation and maintenance decision reference. The equipment health report is pushed to the relevant operation and maintenance personnel. If the operation and maintenance personnel confirm that the equipment needs to be repaired or replaced, corresponding measures are taken according to the work order content, and the processing process and results are fed back to the work order system to form a complete equipment operation and maintenance closed loop. According to the actual effect of equipment operation and maintenance, feedback from operation and maintenance personnel is collected to optimize and improve the rule base and knowledge base of the expert system to improve the accuracy of equipment defect analysis prediction and maintenance decision-making. Machine learning algorithms such as decision trees and support vector machines are used to establish equipment defect prediction models based on equipment historical operation data and operation and maintenance knowledge, and equipment health predictions are carried out regularly to warn of potential equipment problems in advance, guide preventive operation and maintenance, and improve equipment operation reliability.

[0032] Specifically, a rule-based expert system is an intelligent system that simulates the decision-making process of human experts. It evaluates and analyzes the equipment defect analysis and prediction results through a preset rule base. For example, for transformer equipment, the expert system may contain such a rule: if the analysis results of dissolved gas in oil show that the acetylene content exceeds 50ppm, it is judged that the transformer may have a discharge fault. When the real-time monitoring data triggers the rule, the system will issue an early warning signal. The multi-criteria decision-making method can comprehensively consider multiple factors and comprehensively evaluate the severity of equipment defects. Taking the circuit breaker as an example, the evaluation factors may include the importance of the equipment (such as the main transformer circuit breaker is more important than the ordinary line circuit breaker), the scope of the fault impact (such as the possibility of large-scale power outages), the operating years of the equipment, etc. Through methods such as hierarchical analysis, each factor can be given a weight, and finally a comprehensive score is obtained. For example, a circuit breaker scores 85 points (out of 100 points), indicating that its maintenance priority is high. For different types of equipment defects, the expert knowledge base stores corresponding treatment measures. For example, for abnormal dissolved gas in transformer oil, the system may automatically generate the following maintenance suggestions: conduct further chromatographic analysis, check whether the transformer body has abnormal sound, and conduct partial discharge test if necessary. These suggestions will be written into the work order system to form specific maintenance tasks. The work order management system uses visualization technology to convert complex data analysis results into intuitive charts and reports. For example, the health status of the equipment can be displayed in the form of a dashboard, the maintenance priority ranking of different equipment can be displayed in a bar chart, and the recommended maintenance steps can be displayed in a flow chart. This visual display method can help operation and maintenance personnel quickly understand the equipment status and make correct decisions. The push and feedback mechanism of equipment health reports forms a complete operation and maintenance closed loop. When the operation and maintenance personnel receive a report that a transformer needs maintenance, they can take corresponding measures according to the system recommendations. For example, conduct an on-site inspection, replace aging seals, or adjust the cooling system. After the maintenance is completed, the operation and maintenance personnel will enter the processing process and results into the system, and this information can be used for subsequent analysis and decision optimization. By continuously collecting operation and maintenance feedback and actual results, the knowledge base of the expert system can be continuously optimized. For example, if the system predicts that a certain type of circuit breaker may have an operating mechanism failure, but the actual maintenance finds that it is caused by a humid environment, then relevant rules can be added to the knowledge base to improve the accuracy of future predictions. The introduction of machine learning algorithms makes equipment health predictions more intelligent. Taking the decision tree algorithm as an example, it can automatically discover the relationship between equipment parameters and faults by learning historical data. For example, by analyzing a large amount of transformer operation data, the algorithm may find that when the load rate exceeds 90% for a long time and the ambient temperature is higher than 35°C, the transformer insulation aging rate will be significantly accelerated. This discovery can guide preventive maintenance, such as appropriately reducing the transformer load in high temperature seasons.

[0033] Reference Figure 2 , the present application also proposes a power grid equipment information visualization system, the system comprising: Device area division module 100: used to divide the main network devices into regions using a hierarchical clustering algorithm according to the attributes of the main network devices; The historical operation data extraction module 200 is used to automatically obtain the regional division information of the equipment to which the corresponding line belongs when the line loss rate of a certain line is monitored to exceed the preset line loss rate threshold, and extract the historical operation data of the equipment from the cloud database of the equipment in the corresponding area; Association rule mining module 300: used to use Apriori association rule mining algorithm to analyze the correlation between historical equipment operation data and abnormal line loss rate, and obtain the analysis result of the correlation between equipment defects and line loss rate; The module 400 for initially judging the cause of equipment defects is used to train a Bayesian network model according to the results of the correlation analysis between equipment defects and line loss rates, and to estimate the probability and initially judge the cause of equipment defects that cause abnormal line loss rates based on the trained Bayesian network model and the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rates, and to obtain the judgment result of the cause of equipment defects in combination with historical maintenance work orders; Anomaly detection and verification module 500: used to further verify and correct the result of the determination of the cause of the equipment defect based on the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, by comparing and analyzing the historical operation data of the equipment, and using the isolation forest anomaly detection algorithm; Defect type prediction module 600: used to match similar cases from the equipment failure case library according to the corrected equipment defect cause judgment result and the equipment attributes of the corresponding equipment, and analyze and predict the specific defect type of the equipment by using a decision tree algorithm to obtain equipment defect analysis and prediction results; Maintenance suggestion generation and visual feedback module 700: used for rule-based expert system to evaluate equipment defect analysis and prediction results, generate equipment maintenance or replacement priority ranking, and automatically generate corresponding equipment maintenance or replacement suggestions, and visualize the equipment defect analysis and prediction results and suggestions through the work order management system.

[0034] Reference Figure 3 In the embodiment of the present application, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3As shown. The computer device includes a processor, an internal memory, a storage medium (non-volatile storage medium) and a network interface connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes the above-mentioned storage medium (non-volatile storage medium) and the internal memory. The storage medium (non-volatile storage medium) stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the storage medium (non-volatile storage medium). The database of the computer device is used to store usage data, etc. The network interface of the computer device is used to communicate with an external terminal through a network connection. Furthermore, the above-mentioned computer device can also be provided with an input device and a display screen, etc. When the above-mentioned computer program is executed by a processor, a method for visualizing information of power grid equipment is implemented, including the following steps: according to the attributes of the main network equipment, a hierarchical clustering algorithm is used to divide the main network equipment into regions; when it is monitored that the line loss rate of a certain line exceeds a preset line loss rate threshold, the regional division information of the equipment to which the corresponding line belongs is automatically obtained, and the historical operation data of the equipment is extracted from the cloud database of the equipment in the corresponding area; the Apriori association rule mining algorithm is used to analyze the correlation between the historical operation data of the equipment and the abnormal line loss rate, and the analysis results of the correlation between the equipment defects and the line loss rate are obtained; according to the analysis results of the correlation between the equipment defects and the line loss rate, a Bayesian network model is trained, and based on the trained Bayesian network model and the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, the probability of the equipment defect cause causing the abnormal line loss rate is analyzed. Estimate and make preliminary judgments, combined with historical maintenance work orders, to obtain the results of the equipment defect cause judgment; based on the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, through comparative analysis with the historical operation data of the equipment, the isolation forest anomaly detection algorithm is used to further verify and correct the equipment defect cause judgment results; according to the corrected equipment defect cause judgment results, combined with the equipment attributes of the corresponding equipment, the decision tree algorithm is used to match similar cases from the equipment failure case library, analyze and predict the specific defect types of the equipment, and obtain equipment defect analysis and prediction results; based on the rule-based expert system, the equipment defect analysis and prediction results are evaluated, the priority ranking of equipment maintenance or replacement is generated, and the corresponding equipment maintenance or replacement suggestions are automatically generated, and the equipment defect analysis and prediction results and suggestions are visualized through the work order management system. Figure 3 The structure shown in is merely a block diagram of a portion of the structure related to the present application solution and does not constitute a limitation on the computer device to which the present application solution is applied.

[0035] An embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, a visual early warning analysis method for power equipment damage is implemented, including the following steps: according to the attributes of the main network equipment, a hierarchical clustering algorithm is used to divide the main network equipment into regions; when it is monitored that the line loss rate of a certain line exceeds a preset line loss rate threshold, the regional division information of the equipment to which the corresponding line belongs is automatically obtained, and the historical operation data of the equipment is extracted from the cloud database of the equipment in the corresponding area; the Apriori association rule mining algorithm is used to analyze the correlation between the historical operation data of the equipment and the line loss rate anomaly, and the analysis results of the correlation between the equipment defects and the line loss rate are obtained; according to the analysis results of the correlation between the equipment defects and the line loss rate, a Bayesian network model is trained, and based on the trained Bayesian network model and the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, the Bayesian network model is trained. The probability of the equipment defect cause causing the abnormal line loss rate is estimated and preliminarily judged, and the equipment defect cause judgment result is obtained by combining the historical maintenance work order; based on the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, through comparative analysis with the historical operation data of the equipment, the isolation forest anomaly detection algorithm is used to further verify and correct the equipment defect cause judgment result; according to the corrected equipment defect cause judgment result, combined with the equipment attributes of the corresponding equipment, the decision tree algorithm is used to match similar cases from the equipment failure case library, analyze and predict the specific defect type of the equipment, and obtain the equipment defect analysis and prediction results; based on the rule-based expert system, the equipment defect analysis and prediction results are evaluated, the priority ranking of equipment maintenance or replacement is generated, and the corresponding equipment maintenance or replacement suggestions are automatically generated, and the equipment defect analysis and prediction results and suggestions are visualized through the work order management system. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0036] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided in this application and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0037] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device system, article or method including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, device system, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the existence of other identical elements in the process, device system, article or method including the element.

[0038] The above description is only a preferred embodiment of the present application, and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for visualizing power grid equipment information, characterized in that: The method comprises: According to the attributes of the main network devices, a hierarchical clustering algorithm is used to divide the main network devices into regions; When the line loss rate of a certain line is monitored to exceed the preset line loss rate threshold, the regional division information of the equipment to which the corresponding line belongs is automatically obtained, and the historical operation data of the equipment is extracted from the cloud database of the equipment in the corresponding area; The Apriori association rule mining algorithm is used to analyze the correlation between the historical operation data of the equipment and the abnormal line loss rate, and the correlation analysis results between equipment defects and line loss rate are obtained; The Bayesian network model is trained based on the results of the correlation analysis between equipment defects and line loss rate. Based on the trained Bayesian network model and the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, the probability of the equipment defect cause causing the abnormal line loss rate is estimated and preliminarily judged. Combined with the historical maintenance work orders, the judgment result of the equipment defect cause is obtained. Based on the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, through comparative analysis with the historical operation data of the equipment, the isolation forest anomaly detection algorithm is used to further verify and correct the judgment result of the cause of the equipment defect; According to the corrected equipment defect cause judgment results, combined with the equipment attributes of the corresponding equipment, the decision tree algorithm is used to match similar cases from the equipment failure case library, analyze and predict the specific defect type of the equipment, and obtain the equipment defect analysis and prediction results; The rule-based expert system evaluates the equipment defect analysis and prediction results, generates a priority ranking for equipment maintenance or replacement, and automatically generates corresponding equipment maintenance or replacement suggestions. The equipment defect analysis and prediction results and suggestions are visualized through the work order management system.

2. The method according to claim 1, characterized in that The method of dividing the main network devices into regions by using a hierarchical clustering algorithm according to the attributes of the main network devices includes: Obtain multi-dimensional attribute data of main network equipment and construct attribute feature vectors; Calculate the similarity between devices based on the attribute feature vectors of the devices; Based on the similarity, a hierarchical clustering algorithm is used to cluster the devices; Based on the clustering results, analyze the attributes of each category of equipment and determine the regional division plan; Determine the region to which each device belongs based on the regional division plan.

3. The method according to claim 1, characterized in that When the line loss rate of a certain line is monitored to exceed a preset line loss rate threshold, the area division information of the equipment to which the corresponding line belongs is automatically obtained, and the historical operation data of the equipment is extracted from the cloud database of the equipment in the corresponding area, including: Based on the installation of smart meters and sensors on each line, the line operating parameters are collected in real time, and the collected data is uploaded to the cloud database for storage; Use big data analysis technology to mine and analyze historical operation data in the cloud database, establish a line loss rate prediction model, calculate the line loss rate of each line in real time, and compare it with the preset line loss rate threshold; If the line loss rate calculated in real time for a line exceeds the preset threshold, the corresponding line is judged to have abnormal line loss, triggering an abnormal line loss alarm, and automatically obtaining the area division information of the equipment to which the corresponding line belongs; According to the acquired regional division information, the historical equipment operation data of all equipment in the corresponding area is extracted from the cloud database.

4. The method according to claim 1, characterized in that: The Apriori association rule mining algorithm is used to analyze the correlation between the historical operation data of the equipment and the abnormal line loss rate, and the analysis results of the correlation between the equipment defects and the line loss rate are obtained, including: Building a data set based on the acquired historical operation data of the equipment, and preprocessing the data set; The Apriori association rule mining algorithm is used to set support and confidence thresholds to mine the association rules between equipment operating parameters and line loss rate anomalies in the equipment historical operation data. According to the mined association rules, the correlation strength between the equipment operating parameters and the line loss rate anomaly is judged, and the correlation analysis results between equipment defects and line loss rate are obtained.

5. The method according to claim 1, characterized in that The Bayesian network model is trained according to the analysis results of the correlation between equipment defects and line loss rate. Based on the trained Bayesian network model and the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, the probability estimation and preliminary judgment of the equipment defect cause causing the abnormal line loss rate are performed. Combined with the historical maintenance work order, the judgment result of the cause of the equipment defect is obtained, including: Based on the results of the correlation analysis between equipment defects and line loss rates, the causes of equipment defects and first line loss rate data are extracted to construct a training data set for the Bayesian network model; wherein the first line loss rate data includes historical equipment operation data corresponding to abnormal line loss rates; Based on the training data set of the Bayesian network model, the prior probability and conditional probability table of the Bayesian network model are estimated through the parameter learning algorithm, and the probability dependence relationship between the cause of equipment defects and abnormal line loss rate is established; If the line loss rate is abnormal, the second line loss rate data is input into the trained Bayesian network model, and the posterior probability of the defect cause of each device is calculated by the inference algorithm to obtain the probability distribution of the defect cause of the equipment causing the abnormal line loss rate; wherein the second line loss rate data includes the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate; Based on the posterior probability of the equipment defect cause and the preset probability threshold, a preliminary judgment is made on the equipment defect cause of the abnormal line loss rate, and the equipment defect cause that is most likely to cause the abnormality is determined; The preliminary judged cause of the equipment defect is matched with the equipment maintenance work order, the accuracy of the judgment result is verified through the historical maintenance work order, the judgment result is optimized and adjusted, and the judgment result of the cause of the equipment defect is obtained.

6. The method according to claim 1, characterized in that The real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate is compared with the historical operation data of the equipment, and the isolation forest anomaly detection algorithm is used to further verify and correct the judgment result of the cause of the equipment defect, including: Obtain the real-time monitoring data and historical operation data of the equipment corresponding to the line with abnormal line loss rate, pre-process the acquired real-time monitoring data and historical operation data, and obtain standardized real-time monitoring data and historical operation data; Based on the standardized real-time monitoring data and the equipment historical operation data, an isolation forest anomaly detection model is constructed. Through the isolation forest anomaly detection algorithm, the anomaly degree of the standardized real-time monitoring data relative to the standardized equipment historical operation data is calculated to determine whether there is an anomaly in the standardized real-time monitoring data; If there are anomalies in the real-time monitoring data and the corresponding abnormal data is obtained, the judgment result of the cause of the equipment defect is verified according to the abnormal samples of the isolation forest anomaly detection model, and the specific cause of the equipment defect is determined by analyzing the characteristics of the abnormal data; According to the specific defect cause of the verified equipment, the equipment defect cause judgment result is corrected based on the expert knowledge base to obtain a corrected equipment defect cause judgment result.

7. The method according to claim 1, characterized in that According to the corrected equipment defect cause judgment result, combined with the equipment attributes of the corresponding equipment, a decision tree algorithm is used to match similar cases from the equipment failure case library, analyze and predict the specific defect type of the equipment, and obtain equipment defect analysis and prediction results, including: The decision tree algorithm is used to train the corrected equipment defect cause judgment results in the historical case data in the equipment failure case library, and the equipment attribute information of the corresponding equipment is obtained. The corrected equipment defect cause judgment results and equipment attribute information are used as inputs of the decision tree algorithm to obtain an equipment defect analysis and prediction model; From the equipment failure case library, retrieve historical failure cases similar to the equipment to be analyzed based on the equipment attribute information; Input the device attribute information of the current device into the device defect analysis and prediction model, and use the device defect analysis and prediction model to analyze and predict the specific defect type of the device; If the confidence of the analysis and prediction results is lower than the preset threshold, the KNN algorithm is used to further match the most similar cases from similar historical fault cases based on the device attribute information; The prediction results of the equipment defect analysis and prediction model are combined with the most similar cases matched by the KNN algorithm to make a final analysis and judgment on the specific defect type of the equipment, and obtain the equipment defect analysis and prediction results.

8. A power grid equipment information visualization system, characterized in that: The system comprises: Device area division module: used to divide the main network devices into regions according to the attributes of the main network devices using a hierarchical clustering algorithm; Historical operation data extraction module: when the line loss rate of a certain line is monitored to exceed the preset line loss rate threshold, it automatically obtains the regional division information of the equipment to which the corresponding line belongs, and extracts the historical operation data of the equipment from the cloud database of the equipment in the corresponding area; Association rule mining module: used to use the Apriori association rule mining algorithm to analyze the correlation between the historical operation data of the equipment and the abnormal line loss rate, and obtain the analysis results of the correlation between equipment defects and line loss rate; Equipment defect cause preliminary judgment module: used to train the Bayesian network model according to the results of the correlation analysis between equipment defects and line loss rate. Based on the trained Bayesian network model and the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, the module makes a probability estimate and preliminary judgment on the equipment defect cause that causes the abnormal line loss rate. Combined with the historical maintenance work orders, the judgment result of the equipment defect cause is obtained. Anomaly detection and verification module: Based on the real-time monitoring data of the equipment corresponding to the line with abnormal line loss rate, through comparative analysis with the historical operation data of the equipment, the isolation forest anomaly detection algorithm is used to further verify and correct the judgment result of the cause of the equipment defect; Defect type prediction module: used to judge the cause of the equipment defect according to the corrected result, combined with the equipment attributes of the corresponding equipment, using the decision tree algorithm, matching similar cases from the equipment failure case library, analyzing and predicting the specific defect type of the equipment, and obtaining the equipment defect analysis and prediction results; Maintenance suggestion generation and visual feedback module: used for rule-based expert systems to evaluate equipment defect analysis and prediction results, generate equipment maintenance or replacement priority rankings, and automatically generate corresponding equipment maintenance or replacement suggestions, and visualize the equipment defect analysis and prediction results and suggestions through the work order management system.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images