Electric power data cloud edge cooperation system

Abstract

The invention discloses a power data cloud edge cooperation system, which comprises a data processing module, a line evaluation module, a prediction module and a communication module, and is characterized in that the data processing module is used for formatting a historical repair log to obtain a quantization matrix; the line evaluation module is used for acquiring a fault type and a fault reason, the prediction module is used for constructing a multi-state evaluation system according to diversified operation states of the power equipment, and the communication module is used for transmitting a line abnormal state of the power equipment predicted by a cloud to local equipment of a user. According to the method, whether a potential fault exists or not can be judged according to real-time data, the fault type and the fault reason can be quickly identified through comparison and data analysis of historical repair logs, so that accurate positioning of the fault position is provided, the health state of the power equipment can be calculated through the cloud, multi-dimensional and all-around comprehensive evaluation is carried out, and the reliability of the power equipment is improved. The health state index can accurately reflect the actual operation condition of the line.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment evaluation, and in particular to a power data cloud-edge collaboration system. Background Art

[0002] With the continuous growth of global energy demand and the aging of power infrastructure, the stability and reliability of power systems have become increasingly important. As a vital component of the power system, the safety and stability of power equipment directly impacts the reliable supply of electricity. Efficiently monitoring, evaluating, and predicting the operating status of power equipment has become a major challenge facing the power industry. Traditional power equipment monitoring methods rely primarily on manual inspections and periodic testing, which have many limitations. Manual inspections cannot provide real-time monitoring and are easily affected by human factors, making it difficult to detect potential faults in a timely manner. Periodic testing, on the other hand, can easily miss sudden equipment damage or environmental changes. As the scope of power equipment continues to expand, traditional manual inspection methods can no longer meet the needs of efficient, comprehensive, and safe management.

[0003] Currently, a Chinese invention patent application with application number CN202111362694.X discloses a method for assessing the status of power equipment. This application includes obtaining online monitoring data of power equipment and constructing a status evaluation index system; constructing a model for the time-varying confidence level of power equipment status assessment results based on inspection data; and calculating a comprehensive evaluation value of the operating status of overhead power equipment that comprehensively considers both inspection data and online monitoring data. This method can comprehensively utilize inspection data and online monitoring data of power equipment, taking into account the differences in the timeliness characteristics of different data, to achieve a comprehensive and effective assessment of the operating status of power equipment, providing an effective basis for condition-based maintenance. However, existing technologies cannot classify fault types through detailed data quantification and iteration, conduct detailed detection and fault analysis of power equipment status, or build a sound multi-state assessment system for real-time prediction of power equipment faults. Summary of the Invention

[0004] The technical problem solved by the present invention is that the existing technology cannot classify fault types through detailed data quantification and iteration, conduct detailed detection and fault analysis of power equipment status, and cannot build a sound multi-state evaluation system to predict power equipment faults in real time.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: a power data cloud-edge collaborative system, comprising a data processing module, a line evaluation module, a prediction module, and a communication module:

[0006] The data processing module is used to format the historical repair log, and quantize the formatted historical repair log using confidence to obtain a quantization matrix;

[0007] The line evaluation module is used to establish a key parameter sequence system based on the quantization matrix and the evaluation data type definition, compare the currently extracted key parameters with the signal characteristic parameters in the historical repair log, determine the category and cause of the line fault, obtain the fault type and cause, calculate the line status index through a fuzzy comprehensive evaluation method, and calculate the influence of the status index in the evaluation process;

[0008] The prediction module is used to build a multi-state evaluation system according to the diverse operating states of the power equipment, and predict the state of line abnormalities and faults in real time through the multi-state evaluation system;

[0009] The communication module is used to transmit the abnormal line status of the power equipment predicted by the cloud to the user's local device, and at the same time send the historical repair log record of the user's local device to the cloud through the user's local device.

[0010] Preferably, the communication module is used to collect historical maintenance data of the power equipment, and the historical maintenance data includes the date, reason, repaired equipment and maintenance method of the historical maintenance.

[0011] Preferably, the data processing module is used to format the historical repair log, including converting unstructured and semi-structured data in the historical repair data into structured data of the same standard, and the conversion logic includes:

[0012] Extract keywords from the unstructured data and semi-structured data through natural language processing (NLP), extract relationships between the keywords, and obtain semi-structured data in a unified format;

[0013] Unify the field names of the semi-structured data, merge duplicate information, and construct a data table based on the unified field names of the semi-structured data. The data table is used to standardize the fields of all historical repair logs. The fields of the data table include equipment identification, fault type, fault description, maintenance date, repair measures, fault cause, and maintenance personnel. The construction logic includes:

[0014] Extract the field names, remove duplicate field names, classify all field names using the k-means algorithm, add type labels to the classified types, map the classified field names and type labels as a whole to the data table, obtain structured data, and store the structured data in a structured database.

[0015] Preferably, quantifying the structured data by confidence measurement includes:

[0016] The type label of the structured data is used as a state indicator to be quantized, and Huffman coding is used to set confidence levels for the structured data under the same state indicator. The confidence levels represent the state degree of the structured data under the state indicator. The confidence levels are combined into a quantization matrix, where each row of the quantization matrix represents the structured data, each column represents the quantization indicator, and each matrix element corresponds to a confidence level of the quantization indicator. The confidence level also represents the credibility of the structured data.

[0017] The principal component analysis (PCA) is used to reduce the dimension of the quantified structured data and extract key parameters.

[0018] Preferably, the line evaluation module is used to establish a key parameter sequence system by limiting the key parameters and evaluation data types, and the establishment logic includes:

[0019] Setting a confidence threshold, selecting key parameters corresponding to structured data corresponding to a confidence level not less than the confidence threshold, and monotonically arranging the key parameters in chronological order and confidence level to construct a time series system, wherein the time series system is used to describe the evolution process of the operating status and fault characteristics of the power equipment;

[0020] Comparing the currently extracted key parameters with the signal characteristic parameters in the historical repair logs and analyzing the similarity between the key parameters and the signal characteristic parameters in the historical repair logs includes:

[0021] Using dynamic time warping (DTW), the similarity between each key parameter and the signal characteristic parameters in the historical repair log is calculated. A similarity threshold is set for each historical repair log. When the similarity is not less than the similarity threshold of the historical repair log corresponding to the compared signal characteristic parameter, it is detected that the line state corresponding to the key parameter is the same as the line state corresponding to the historical repair log.

[0022] Fault category determination and fault cause analysis for line status include:

[0023] The fault types include short circuit, overload and corrosion, and the fault causes include environmental factors and electrical factors. Based on the similarity calculation results, the structured data are nested and classified into the categories of the fault type and fault cause using the k-means algorithm, and a fault category label and a fault cause label are set for each structured data.

[0024] Preferably, mapping the key parameter sequence to the line state index by a fuzzy comprehensive evaluation method comprises: using fuzzy logic to comprehensively evaluate the key parameters in the key parameter sequence to obtain the line state index of the power equipment;

[0025] The influence of the state indicators in the mapping process is calculated by weighted average, which is the weighted average of conductor temperature, conductor sag and insulator contamination. The weight coefficient is automatically adjusted according to the predicted probability distribution of the comprehensive evaluation.

[0026] Preferably, the prediction module is used to construct a multi-state evaluation system, and the construction logic includes:

[0027] The diverse operating states of power equipment include normal operation, overload, critical state, and fault state. Each of these diverse operating states corresponds to a line state index. The multi-state evaluation system also assesses line performance based on different operating scenarios, including load changes, weather impacts, and equipment aging.

[0028] Real-time data is acquired by installing sensors on power equipment. The sensors include current sensors, voltage sensors, temperature sensors, vibration sensors, and humidity sensors. Real-time data acquired by different sensors are converted into a unified unit using physical formulas. The real-time data is transmitted to a centralized monitoring system using Internet of Things technology. The real-time data is then transmitted to a big data platform for storage using a cloud computing platform. A prediction model is constructed based on the real-time data and historical data, including historical repair logs of power equipment. The construction logic includes:

[0029] Obtaining the line status index corresponding to the real-time data through the line evaluation module, and comparing the line status index corresponding to the real-time data with the line status index corresponding to the historical data includes: using the line status index corresponding to the historical data as a category label to train a support vector machine, respectively obtaining the power equipment status characteristics corresponding to each category label, inputting the line status index corresponding to the real-time data as a validation set into the trained support vector machine, obtaining the power equipment status characteristics corresponding to each line status index, and performing real-time abnormal state detection on the power equipment status characteristics through the anomaly detection algorithm isolation forest. The anomaly detection algorithm isolation forest is based on online learning and incremental learning, and the anomaly detection algorithm isolation forest is updated according to real-time new data. When the anomaly detection algorithm isolation forest detects an abnormal state, the abnormal state is input into the random forest in real time to predict and output a probability vector of the fault type. The probability vector of the fault type represents a potential defect of the power equipment. The potential defect includes loose conductor strands and broken insulators. The potential defect is located by a multi-classifier fusion strategy based on evidence theory.

[0030] Preferably, the logic for locating potential defects includes:

[0031] The probability vector is classified and predicted, and the classification type is obtained by the category label corresponding to the probability vector. The prediction results of each category are plotted as a continuous probability distribution diagram. The probability distribution represents the confidence level that the line belongs to different fault types. The probability distribution is combined into a layer using the Dempster-Shafer theory, and the degree of the fault type is characterized by visual elements. The visual elements include color, shape and size. The combined layer is applied to a three-dimensional power equipment virtual environment using three-dimensional visualization technology.

[0032] Preferably, when the prediction model predicts an abnormal state of the power equipment, a warning is issued to the maintenance personnel to prompt them to check the circuit.

[0033] Preferably, the transfer of the state of the power device is predicted by a Markov model, and the transition of the power device from one state to another is dynamically tracked through real-time collected data, and the transition state is used as the predicted output of the power device.

[0034] Beneficial effects of the present invention: The present invention can collect and analyze the status data of power equipment in real time through the user's local equipment, promptly identify the abnormal status of the line, and determine whether there is a potential fault based on the real-time data. This can effectively provide early warning, avoid the occurrence of major faults, reduce equipment damage and downtime, and quickly identify the type of fault and the cause of the fault through comparison and data analysis of historical repair logs, thereby providing accurate positioning of the fault location, helping operation and maintenance personnel to respond and handle it quickly. Through the fuzzy comprehensive evaluation method, combined with multiple key indicators, the system can conduct a multi-dimensional and comprehensive evaluation of the health status of power equipment. The health status index can accurately reflect the actual operating status of the line, making it easier for power operators to understand the status of the equipment in a timely manner. This method constructs a multi-state evaluation system for the various operating states of power equipment, and can perform accurate evaluations in different operating scenarios. This ensures that the system can cope with different environments and working scenarios, provides a strong guarantee for the stable operation of the line, and effectively reduces the risk of failure caused by the failure to monitor the status changes in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A basic flow chart of a power data cloud-edge collaborative system provided for one embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0037] Reference Figure 1, as an embodiment of the present invention, provides a power data cloud-edge collaboration system, including a data processing module, a line evaluation module, a prediction module, and a communication module:

[0038] The data processing module is used to format the historical repair log, and quantize the formatted historical repair log using confidence to obtain a quantization matrix;

[0039] The line evaluation module is used to establish a key parameter sequence system based on the quantization matrix and the evaluation data type definition, compare the currently extracted key parameters with the signal characteristic parameters in the historical repair log, determine the category and cause of the line fault, obtain the fault type and cause, calculate the line status index through a fuzzy comprehensive evaluation method, and calculate the influence of the status index in the evaluation process;

[0040] The prediction module is used to build a multi-state evaluation system according to the diverse operating states of the power equipment, and predict the state of line abnormalities and faults in real time through the multi-state evaluation system;

[0041] The communication module is used to transmit the abnormal line status of the power equipment predicted by the cloud to the user's local device, and at the same time send the historical repair log record of the user's local device to the cloud through the user's local device.

[0042] The present invention classifies fault types through detailed data quantification and iteration, performs meticulous detection and fault analysis on the power equipment status, and builds a sound multi-state evaluation system to predict the faults of power equipment in real time.

[0043] The communication module is used to collect historical maintenance data of the power equipment, and the historical maintenance data includes the date, reason, repaired equipment and maintenance method of the historical maintenance.

[0044] The data processing module is used to format the historical repair log, including converting unstructured and semi-structured data in the historical repair data into structured data of the same standard. The conversion logic includes:

[0045] Extract keywords from the unstructured data and semi-structured data through natural language processing (NLP), extract relationships between the keywords, and obtain semi-structured data in a unified format;

[0046] Unify the field names of the semi-structured data, merge duplicate information, and construct a data table based on the unified field names of the semi-structured data. The data table is used to standardize the fields of all historical repair logs. The fields of the data table include equipment identification, fault type, fault description, maintenance date, repair measures, fault cause, and maintenance personnel. The construction logic includes:

[0047] Extract the field names, remove duplicate field names, classify all field names using the k-means algorithm, add type labels to the classified types, map the classified field names and type labels as a whole to the data table to obtain structured data, store the structured data in a structured database to ensure high availability and queryability of the data, convert historical maintenance data from different sources and formats into a unified format that is easy to store, query and analyze, and provide support for subsequent fault prediction, maintenance optimization, etc.

[0048] Quantifying the structured data by confidence metrics includes:

[0049] The type label of the structured data is used as a state indicator to be quantified. Huffman coding is used to set confidence levels for the structured data under the same state indicator. The confidence levels represent the state degree of the structured data under the state indicator. For example, if the fault types include short circuit, overload, and corrosion, the short circuit confidence level is set to 1, the overload confidence level is set to 2, and the corrosion confidence level is set to 3. The confidence levels are combined into a quantization matrix, where each row of the quantization matrix represents the structured data, each column represents the quantization indicator, and each matrix element corresponds to the confidence level of the quantization indicator. The confidence level also represents the credibility of the structured data, that is, the credibility of each historical repair log.

[0050] The quantization matrix usually contains multiple dimensions, and many dimensions may be highly correlated or redundant. Therefore, it is necessary to reduce the dimensionality of the quantization matrix to extract key parameters.

[0051] Principal component analysis (PCA) is used to reduce the dimension of the quantified structured data and extract the key parameters that best reflect the operating status of the equipment. Principal component analysis (PCA) selects the most important components that explain the data variance, reduces redundant information, retains key features, and manually selects the most suitable number of dimensions after dimensionality reduction.

[0052] The line evaluation module is used to establish a key parameter sequence system through the key parameters and evaluation data type restrictions, and the establishment logic includes:

[0053] Manually set a confidence threshold, select key parameters corresponding to structured data with a confidence level not less than the confidence threshold, and monotonically arrange the key parameters in chronological order and confidence level to construct a time series system. The time series system is used to describe the evolution of the operating status and fault characteristics of the power equipment. For each historical repair log, the change trend of each key parameter in the record can be tracked to form time series data;

[0054] Comparing the currently extracted key parameters with the signal characteristic parameters in the historical repair logs and analyzing the similarity between the key parameters and the signal characteristic parameters in the historical repair logs includes:

[0055] Using dynamic time warping (DTW), the similarity between each key parameter and the signal characteristic parameters in the historical repair log is calculated. A similarity threshold is manually set for each historical repair log. When the similarity is not less than the similarity threshold of the historical repair log corresponding to the compared signal characteristic parameter, it is detected that the line state corresponding to the key parameter is the same as the line state corresponding to the historical repair log.

[0056] Fault category determination and fault cause analysis for line status include:

[0057] The fault types include short circuit, overload and corrosion, and the fault causes include environmental factors and electrical factors. Based on the similarity calculation results, the structured data are nested and classified into the categories of the fault type and fault cause using the k-means algorithm, and a fault category label and a fault cause label are set for each structured data.

[0058] By using confidence levels to quantify, reduce the dimension, extract key parameters, and perform similarity analysis on historical repair logs, we can effectively classify and analyze the causes of line faults, providing a basis for fault prediction. This process utilizes machine learning and data analysis techniques to help improve the fault detection and early warning capabilities of power equipment, thereby improving the safety and reliability of the power system.

[0059] Mapping the key parameter sequence to a line status index using a fuzzy comprehensive evaluation method includes: using fuzzy logic to comprehensively evaluate the key parameters in the key parameter sequence to obtain a line status index for the power equipment, wherein the status index can reflect the overall health status of the line and provide a quantitative basis for further fault prediction and diagnosis;

[0060] The influence of the state indicators in the mapping process is calculated by weighted average, which is the weighted average of conductor temperature, conductor sag and insulator contamination. The weight coefficient is automatically adjusted according to the predicted probability distribution of the comprehensive evaluation.

[0061] The prediction module is used to build a multi-state evaluation system, and the construction logic includes:

[0062] The diverse operating states of power equipment include normal operation, overload, critical state, and fault state. Each of these diverse operating states corresponds to a line state index. The multi-state evaluation system also assesses line performance based on different operating scenarios, including load changes, weather impacts, and equipment aging.

[0063] Real-time data is acquired by installing sensors on power equipment. The sensors include current sensors, voltage sensors, temperature sensors, vibration sensors, and humidity sensors. Real-time data acquired by different sensors is converted into a unified unit using physical formulas. The real-time data is transmitted to a centralized monitoring system using Internet of Things technology to ensure real-time data updates and use for status assessment. The real-time data is transmitted to a big data platform for storage using a cloud computing platform. A prediction model is constructed based on the real-time data and historical data, including historical repair logs of power equipment. The construction logic includes:

[0064] The line status index corresponding to the real-time data is obtained through the line evaluation module, and the line status index corresponding to the real-time data is compared with the line status index corresponding to the historical data, including: using the line status index corresponding to the historical data as a category label for training through a support vector machine, respectively obtaining the power equipment status characteristics corresponding to each category label, inputting the line status index corresponding to the real-time data as a validation set into the trained support vector machine, obtaining the power equipment status characteristics corresponding to each line status index, performing real-time abnormal state detection on the power equipment status characteristics through the anomaly detection algorithm isolation forest, the anomaly detection algorithm isolation forest is based on online learning and incremental learning, and the anomaly detection algorithm isolation forest is updated according to real-time new data to improve the prediction accuracy of the model, when the anomaly detection algorithm isolation forest detects an abnormal state, the abnormal state is input into the random forest in real time, and a probability vector of the fault type is predicted and output, the probability vector of the fault type represents the potential defects of the power equipment, the potential defects include loose conductor strands and broken insulators, and the potential defects are located based on the multi-classifier fusion strategy of evidence theory.

[0065] The logic for locating potential defects includes:

[0066] The probability vector is classified and predicted, and the classification type is obtained by the category label corresponding to the probability vector. The prediction results of each category are plotted as a continuous probability distribution diagram. The probability distribution represents the confidence level that the line belongs to different fault types. The probability distribution is combined into a layer using the Dempster-Shafer theory, and the degree of the fault type is characterized by visual elements. The visual elements include color, shape and size. The combined layer is applied to a three-dimensional power equipment virtual environment using three-dimensional visualization technology.

[0067] When the prediction model predicts an abnormal state of the power equipment, it warns the maintenance personnel and prompts them to check the lines. By evaluating the operating quality of the power equipment, the Internet of Things automatically determines the load balance, stability and overall operating efficiency of the power equipment.

[0068] The Markov model is used to predict the transition of power equipment status; for example, if the line is in a "slightly abnormal" state, the system can predict that it may transition to a "moderately abnormal" or "fault precursor" state and take measures in advance.

[0069] By using data collected in real time, the conversion of the power equipment from one state to another is dynamically tracked, and the converted state is used as the predicted output of the power equipment. The state is automatically classified by the Markov model.

[0070] The present invention can collect and analyze the status data of power equipment in real time through the user's local device, promptly identify the abnormal status of the line, and determine whether there is a potential fault based on the real-time data. This can effectively provide early warning, avoid the occurrence of major faults, reduce equipment damage and downtime, and quickly identify the type and cause of the fault by comparing and analyzing historical repair logs, thereby providing accurate positioning of the fault location and helping operation and maintenance personnel to respond and handle it quickly. Through the fuzzy comprehensive evaluation method and combining multiple key indicators, the system can conduct a multi-dimensional and comprehensive evaluation of the health status of power equipment. The health status index can accurately reflect the actual operating status of the line, making it easier for power operators to understand the equipment status in a timely manner. This method constructs a multi-state evaluation system for the various operating states of power equipment and can perform accurate evaluations in different operating scenarios. This ensures that the system can cope with different environments and working scenarios, provides a strong guarantee for the stable operation of the line, and effectively reduces the risk of failure caused by the failure to monitor the status changes in a timely manner.

[0071] Those skilled in the art will appreciate that embodiments of the present invention may provide methods, systems, or computer program products. Therefore, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium may be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0072] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A power data cloud-edge collaborative system, characterized by: Including data processing module, line evaluation module, prediction module and communication module: The data processing module is used to format the historical repair log, and quantize the formatted historical repair log using confidence to obtain a quantization matrix; The line evaluation module is used to establish a key parameter sequence system based on the quantization matrix and the evaluation data type definition, compare the currently extracted key parameters with the signal characteristic parameters in the historical repair log, determine the category and cause of the line fault, obtain the fault type and cause, calculate the line status index through a fuzzy comprehensive evaluation method, and calculate the influence of the status index in the evaluation process; The prediction module is used to build a multi-state evaluation system according to the diverse operating states of the power equipment, and predict the state of line abnormalities and faults in real time through the multi-state evaluation system; The communication module is used to transmit the abnormal line status of the power equipment predicted by the cloud to the user's local device, and at the same time send the historical repair log record of the user's local device to the cloud through the user's local device.

2. The power data cloud-edge collaboration system according to claim 1, characterized in that: The communication module is used to collect historical maintenance data of the power equipment, and the historical maintenance data includes the date, reason, repaired equipment and maintenance method of the historical maintenance.

3. The power data cloud-edge collaboration system according to claim 2, characterized in that: The data processing module is used to format the historical repair log, including converting unstructured and semi-structured data in the historical repair data into structured data of the same standard. The conversion logic includes: Extract keywords from the unstructured data and semi-structured data through natural language processing (NLP), extract relationships between the keywords, and obtain semi-structured data in a unified format; Unify the field names of the semi-structured data, merge duplicate information, and construct a data table based on the unified field names of the semi-structured data. The data table is used to standardize the fields of all historical repair logs. The fields of the data table include equipment identification, fault type, fault description, maintenance date, repair measures, fault cause, and maintenance personnel. The construction logic includes: Extract the field names, remove duplicate field names, classify all field names using the k-means algorithm, add type labels to the classified types, map the classified field names and type labels as a whole to the data table, obtain structured data, and store the structured data in a structured database.

4. The power data cloud-edge collaboration system according to claim 3, characterized in that: Quantifying the structured data by confidence metrics includes: The type label of the structured data is used as a state indicator to be quantized, and Huffman coding is used to set confidence levels for the structured data under the same state indicator. The confidence levels represent the state degree of the structured data under the state indicator. The confidence levels are combined into a quantization matrix, where each row of the quantization matrix represents the structured data, each column represents the quantization indicator, and each matrix element corresponds to a confidence level of the quantization indicator. The confidence level also represents the credibility of the structured data. The principal component analysis (PCA) is used to reduce the dimension of the quantified structured data and extract key parameters.

5. The power data cloud-edge collaboration system according to claim 4, characterized in that: The circuit evaluation module is used to establish a key parameter sequence system by limiting the key parameters and evaluation data types, and the establishment logic includes: Setting a confidence threshold, selecting key parameters corresponding to structured data corresponding to a confidence level not less than the confidence threshold, and monotonically arranging the key parameters in chronological order and confidence level to construct a time series system, wherein the time series system is used to describe the evolution process of the operating status and fault characteristics of the power equipment; Comparing the currently extracted key parameters with the signal characteristic parameters in the historical repair logs and analyzing the similarity between the key parameters and the signal characteristic parameters in the historical repair logs includes: Using dynamic time warping (DTW), the similarity between each key parameter and the signal characteristic parameters in the historical repair log is calculated. A similarity threshold is set for each historical repair log. When the similarity is not less than the similarity threshold of the historical repair log corresponding to the compared signal characteristic parameter, it is detected that the line state corresponding to the key parameter is the same as the line state corresponding to the historical repair log. Fault category determination and fault cause analysis for line status include: The fault types include short circuit, overload and corrosion, and the fault causes include environmental factors and electrical factors. Based on the similarity calculation results, the structured data are nested and classified into the categories of the fault type and fault cause using the k-means algorithm, and a fault category label and a fault cause label are set for each structured data.

6. The power data cloud-edge collaboration system according to claim 5, characterized in that: Mapping the key parameter sequence to a line state index by a fuzzy comprehensive evaluation method includes: using fuzzy logic to comprehensively evaluate the key parameters in the key parameter sequence to obtain a line state index of the power equipment; The influence of the state indicators in the mapping process is calculated by weighted average, which is the weighted average of conductor temperature, conductor sag and insulator contamination. The weight coefficient is automatically adjusted according to the predicted probability distribution of the comprehensive evaluation.

7. The power data cloud-edge collaboration system according to claim 6, characterized in that: The prediction module is used to build a multi-state evaluation system, and the construction logic includes: The diverse operating states of power equipment include normal operation, overload, critical state, and fault state. Each of these diverse operating states corresponds to a line state index. The multi-state evaluation system also assesses line performance based on different operating scenarios, including load changes, weather impacts, and equipment aging. Real-time data is acquired by installing sensors on power equipment. The sensors include current sensors, voltage sensors, temperature sensors, vibration sensors, and humidity sensors. Real-time data acquired by different sensors are converted into a unified unit using physical formulas. The real-time data is transmitted to a centralized monitoring system using Internet of Things technology. The real-time data is then transmitted to a big data platform for storage using a cloud computing platform. A prediction model is constructed based on the real-time data and historical data, including historical repair logs of power equipment. The construction logic includes: Obtaining the line status index corresponding to the real-time data through the line evaluation module, and comparing the line status index corresponding to the real-time data with the line status index corresponding to the historical data includes: using the line status index corresponding to the historical data as a category label to train a support vector machine, respectively obtaining the power equipment status characteristics corresponding to each category label, inputting the line status index corresponding to the real-time data as a validation set into the trained support vector machine, obtaining the power equipment status characteristics corresponding to each line status index, and performing real-time abnormal state detection on the power equipment status characteristics through the anomaly detection algorithm isolation forest. The anomaly detection algorithm isolation forest is based on online learning and incremental learning, and the anomaly detection algorithm isolation forest is updated according to real-time new data. When the anomaly detection algorithm isolation forest detects an abnormal state, the abnormal state is input into the random forest in real time to predict and output a probability vector of the fault type. The probability vector of the fault type represents a potential defect of the power equipment. The potential defect includes loose conductor strands and broken insulators. The potential defect is located by a multi-classifier fusion strategy based on evidence theory.

8. The power data cloud-edge collaboration system according to claim 7, characterized in that: The logic for locating potential defects includes: The probability vector is classified and predicted, and the classification type is obtained by the category label corresponding to the probability vector. The prediction results of each category are plotted as a continuous probability distribution diagram. The probability distribution represents the confidence level that the line belongs to different fault types. The probability distribution is combined into a layer using the Dempster-Shafer theory, and the degree of the fault type is characterized by visual elements. The visual elements include color, shape and size. The combined layer is applied to a three-dimensional power equipment virtual environment using three-dimensional visualization technology.

9. The power data cloud-edge collaboration system according to claim 8, characterized in that: When the prediction model predicts an abnormal state of the power equipment, a warning is issued to the maintenance personnel, prompting them to check the circuit.

10. The power data cloud-edge collaboration system according to claim 9, characterized in that: The transfer of the state of the power equipment is predicted by the Markov model, and the conversion of the power equipment from one state to another is dynamically tracked through the real-time collected data, and the converted state is used as the predicted output of the power equipment.

Images