Method and system for monitoring high-pressure gas relay and storage medium

Abstract

The invention relates to the technical field of relay monitoring and control, and discloses a method and system for monitoring a high-pressure gas relay and a storage medium. The method comprises the following steps: carrying out closed-loop control data acquisition on SF6 gas parameters of a high-pressure gas relay to obtain control loop operation data; calculating a control deviation and a residual vector for the data to obtain a control performance index; inputting the performance index into an isolation forest algorithm to identify abnormity, and obtaining a health state evaluation result; adaptive parameter adjustment is executed according to the evaluation result, PID parameters are optimized through iterative learning control, and a control parameter set is generated. According to the invention, closed-loop control system data acquisition is carried out on the SF6 gas parameters of the high-pressure gas relay, and the isolation forest algorithm is used for abnormal identification, so that the problem that a traditional fixed threshold monitoring method is easy to cause false alarm and missing alarm is solved.

Description

Technical Field

[0001] The present application relates to the field of relay monitoring and control technology, and in particular to a method, system and storage medium for monitoring a high-voltage gas relay. Background Art

[0002] High-voltage gas relays are important protective devices in power systems. They usually use sulfur hexafluoride (SF6) as an insulating medium and are widely used in high-voltage power transmission and distribution systems. Traditional high-voltage gas relay monitoring methods mainly rely on regular inspections and offline tests to evaluate the status of the equipment by detecting parameters such as SF6 gas density, pressure, and purity. With the advancement of smart grid construction, automated monitoring technology has been widely used. Existing technologies have begun to use real-time monitoring systems to collect online data on the operating parameters of high-voltage gas relays, and perform anomaly detection through simple judgment methods such as threshold methods. These monitoring systems usually use fixed thresholds to judge the status of the equipment, and use traditional PID controllers to maintain the stability of parameters such as gas pressure and density.

[0003] However, the existing high-voltage gas relay monitoring technology has obvious shortcomings. First, the anomaly detection method based on fixed thresholds is difficult to adapt to the complex and changeable operating environment and equipment aging process, and is prone to false alarms and missed alarms. Secondly, the traditional PID controller uses fixed parameters and cannot be adaptively adjusted according to the health status of the equipment and changes in the operating environment, resulting in the control performance gradually deteriorating with the aging of the equipment. In addition, the existing monitoring methods are insufficient in data preprocessing, fail to fully utilize the patterns and knowledge contained in the historical operating data, lack the ability to accurately evaluate and predict the status of the equipment, and are difficult to support predictive maintenance decisions.

[0004] At the same time, how to achieve accurate collection and control of SF6 gas parameters in high-voltage gas relays, how to scientifically analyze the collected data to obtain control performance indicators, how to use advanced algorithms to identify abnormal states and degrees, and how to dynamically adjust control parameters based on evaluation results to establish a complete closed-loop intelligent monitoring and control system. In particular, in terms of control parameter adjustment, how to adopt differentiated control strategies based on different abnormal levels and use iterative learning algorithms to achieve parameter optimization are key links that need to be broken through. Summary of the invention

[0005] The present application provides a method, system and storage medium for monitoring high-voltage gas relays. By collecting data of the SF6 gas parameters of the high-voltage gas relays through a closed-loop control system and using the isolation forest algorithm for abnormal identification, the problem of false alarms and missed alarms in the traditional fixed threshold monitoring method is solved. At the same time, the iterative learning control algorithm is used to adaptively adjust the PID parameters according to the health status assessment results, overcoming the defect that the traditional fixed parameter controller cannot cope with equipment aging and environmental changes.

[0006] In a first aspect, the present application provides a method for monitoring a high-voltage gas relay. The method for monitoring the high-voltage gas relay includes: performing data acquisition on the SF6 gas pressure, density, and temperature parameters of the high-voltage gas relay based on a closed-loop control system to obtain the operating data of the relay control loop; calculating the control deviation and analyzing the residual vector of the operating data of the relay control loop to obtain the relay control performance index; inputting the relay control performance index into an isolation forest algorithm for abnormal pattern recognition, determining the degree of abnormality by calculating the isolation path length of the sample, and obtaining the evaluation result of the health status of the control loop; and performing adaptive control parameter adjustment according to the evaluation result of the health status of the control loop, optimizing the PID parameters by using an iterative learning control algorithm, and generating a set of relay control parameters.

[0007] Optionally, the performing data acquisition on the SF6 gas pressure, density, and temperature parameters of the high-voltage gas relay based on a closed-loop control system to obtain the operating data of the relay control loop includes: Collecting the physical parameters of the SF6 gas through a pressure sensor, a density sensor, and a temperature sensor installed in the gas chamber of the high-voltage gas relay, and marking the collected data with time stamps to obtain the original parameter time series data; Performing outlier detection on the original parameter time series data according to the three-standard-deviation principle, replacing the single-point outlier with the average value of the previous and subsequent time points, and repairing the continuous outliers by using linear interpolation to obtain the cleaned parameter data; Inputting the cleaned parameter data into a Kalman filter, and performing state estimation based on the process noise covariance matrix and the measurement noise covariance matrix to obtain the filtered parameter data; Performing standardization processing on the filtered parameter data by a closed-loop control system, converting the data with different dimensions into standard distribution data, and obtaining the standardized control loop parameter characteristics; Calculating the statistics within a ten-minute sliding window based on the standardized control loop parameter characteristics, including the mean, variance, maximum value, minimum value, kurtosis, and skewness, to obtain a multi-dimensional control loop feature vector; Calculating the deviation between the controller output signal and the actual system response according to the multi-dimensional control loop feature vector and the action time data recorded by the relay action mechanism sensor, and generating the operating data of the relay control loop.

[0008] Optionally, the calculating the control deviation and analyzing the residual vector of the operating data of the relay control loop to obtain the relay control performance index includes: Performing segmentation processing on the operating data of the relay control loop, and dividing the data into multiple pressure adjustment cycles according to the change points of the SF6 gas pressure control instruction to obtain the gas pressure control segmented data; Calculate the difference between the pressure set value and the actual pressure value for each adjustment cycle in the gas pressure control segmented data to form a gas pressure control deviation sequence, and obtain the gas pressure control deviation data; Calculate the integral absolute error and the integral time absolute error of the gas relay pressure control loop according to the gas pressure control deviation data, perform weighted summation on the error data, and obtain the gas pressure control accuracy evaluation value; Compare the gas pressure control accuracy evaluation value with the relay historical reference value, calculate the gas pressure control performance degradation rate, and combine with the gas insulation strength analysis to obtain the relay insulation performance attenuation index; Construct a gas relay residual vector based on the gas pressure control deviation data, analyze the structure of the residual vector by calculating the difference between the theoretical gas pressure response and the actual gas pressure response, and obtain the relay leakage characteristic distribution; According to the relay leakage characteristic distribution and the relay insulation performance attenuation index, comprehensively evaluate the state of the relay gas system through a weighted fusion algorithm, and generate the relay control performance index.

[0009] Optionally, input the relay control performance index into the isolation forest algorithm for abnormal pattern recognition, determine the degree of abnormality by calculating the sample isolation path length, and obtain the control loop health status evaluation result, including: Perform data standardization processing on the relay control performance index and perform feature space transformation to obtain a standardized control performance feature set; Construct an isolation forest model based on the standardized control performance feature set, construct multiple decision trees by randomly selecting sample subsets and randomly selecting features, randomly select a feature and a splitting threshold at each node for data splitting until the samples are completely isolated or reach the preset maximum depth, and obtain the abnormal sample detection structure; Input the standardized control performance feature set into the abnormal sample detection structure, record the path length of each feature sample point from the root node to the leaf node, take the average value of the path lengths of each decision tree as the isolation path length, and obtain the abnormal separation data of the feature samples; Perform normalization processing according to the abnormal separation data of the feature samples, convert the average path length into a standardized abnormal index between zero and one, and the closer the value is to one, the higher the probability of abnormality, and obtain the relay control abnormal probability value; Set multi-level abnormal thresholds based on the relay control abnormal probability value, divide the abnormality into four levels: slight abnormality, moderate abnormality, severe abnormality, and critical abnormality, and calibrate the thresholds according to the characteristics of the high-pressure gas relay to obtain the relay abnormal level result; Analyze the contribution degree of each control performance index to the abnormality according to the abnormal level result of the relay, determine the key parameters affecting the relay performance through feature importance calculation, and generate the evaluation result of the health state of the control loop.

[0010] Optionally, construct an isolation forest model based on the standardized control performance feature set, construct multiple decision trees by randomly selecting sample subsets and randomly selecting features, randomly select a feature and a splitting threshold at each node for data splitting until the samples are completely isolated or reach the preset maximum depth, and obtain the abnormal sample detection structure, including: Perform sample subset partitioning on the standardized control performance feature set, use the random sampling method to extract multiple sample subsets from the feature set, and keep the number of samples in each subset equal to obtain a set of training sample subsets; Construct a decision tree forest based on the set of training sample subsets, construct a completely random decision tree for each sample subset, and set the total number of decision trees to the logarithm of the total number of samples to obtain the initial forest structure; Perform construction process control on each decision tree in the initial forest structure, set the maximum tree depth to the logarithm of the feature dimension to prevent overfitting, and obtain a set of decision trees with limited depth; Perform node splitting operations based on the set of decision trees with limited depth, randomly select a feature dimension for each internal node, and randomly select a splitting threshold within the value range of the feature to divide the data into left and right subtrees to obtain a randomly split decision tree structure; Calculate the sample path according to the randomly split decision tree structure, record the average path length of each normal sample in the training set from the root node to the leaf node, establish the path length distribution of the normal operation state of the high-pressure gas relay, and obtain the normal state reference model; Calibrate the parameters of the normal state reference model, adjust the abnormal determination boundary according to the characteristics of the gas relay, and set the conversion coefficient of the abnormal degree in combination with expert knowledge to form the abnormal sample detection structure.

[0011] Optionally, perform adaptive control parameter adjustment according to the evaluation result of the health state of the control loop, optimize the PID parameters using the iterative learning control algorithm, and generate a set of relay control parameters, including: Perform grading processing on the evaluation result of the health state of the control loop, determine the control parameter adjustment strategy according to the abnormal level and abnormal type, and determine different adjustment weights for different types of abnormalities to obtain the control parameter adjustment plan; Perform iterative learning control on the control parameter adjustment plan, design a learning gain matrix according to the control characteristics of the high-pressure gas relay, analyze and process the historical control data, and obtain the control trajectory optimization data; Apply the control trajectory optimization data to a PID controller, predict the deviation trend of the current control cycle by analyzing the error trajectory of the previous control cycle, adjust the control output, and obtain a compensated control signal; Perform automatic adjustment calculation of PID parameters according to the compensated control signal, and use the error convergence speed and overshoot index to dynamically optimize the proportional coefficient, integral coefficient, and differential coefficient through the gradient descent method to obtain an optimized PID parameter set; Verify the robustness of the optimized PID parameter set, conduct simulation tests for multiple operating points within the range of SF6 gas pressure and temperature changes, evaluate the stability of control performance, and obtain PID controller parameters with robustness; Form a complete control strategy based on the PID controller parameters with robustness and the characteristics of the gas relay, including a normal mode parameter set, a slightly abnormal mode parameter set, a moderately abnormal mode parameter set, and a safety mode parameter set, and generate the relay control parameter set.

[0012] Optionally, perform iterative learning control on the control parameter adjustment scheme, design a learning gain matrix for the control characteristics of the high-pressure gas relay, analyze and process historical control data, and obtain control trajectory optimization data, including: Batch process the historical operation data of the high-pressure gas relay, with each batch containing control data under similar working conditions, to obtain a control data learning set; Extract the control error trajectory characteristics based on the control data learning set, calculate the shape characteristics of the time-domain error curve and the frequency-domain error distribution characteristics of each control cycle, and obtain an error mode feature library; Perform similarity analysis on the error mode feature library, calculate the similarity between the current error trajectory and the historical error trajectory using the dynamic time warping algorithm, identify the historical error mode with the highest similarity, and obtain an error mode matching result; Construct a learning gain matrix according to the error mode matching result, and the learning gain value is dynamically adjusted according to the health status of the control loop. The higher the degree of abnormality, the smaller the learning gain, to obtain an adaptive learning gain matrix; Perform iterative calculation on the historical control input based on the adaptive learning gain matrix, correct the control input of the next cycle through the error propagation correction formula, and perform ten iterative operations to obtain an optimized control input sequence; Fuse the optimized control input sequence with the current control strategy, determine the fusion weight according to the evaluation result of the control loop health status, and smooth the control trajectory to generate the control trajectory optimization data.

[0013] In a second aspect, the present application provides a system for monitoring a high-pressure gas relay. The system for monitoring a high-pressure gas relay includes: A collection module, configured to perform data collection on the SF6 gas pressure, density, and temperature parameters of a high-voltage gas relay based on a closed-loop control system to obtain the operation data of the relay control circuit; An analysis module, configured to perform control deviation calculation and residual vector analysis on the operation data of the relay control circuit to obtain the relay control performance index; An identification module, configured to input the relay control performance index into an isolation forest algorithm for abnormal mode identification, determine the degree of abnormality by calculating the sample isolation path length, and obtain the evaluation result of the health state of the control circuit; An execution module, configured to perform adaptive control parameter adjustment according to the evaluation result of the health state of the control circuit, optimize the PID parameters by using an iterative learning control algorithm, and generate a relay control parameter set.

[0014] In a third aspect, a device for monitoring a high-voltage gas relay is provided, including: a memory and at least one processor, wherein instructions are stored in the memory; the at least one processor calls the instructions in the memory to enable the device for monitoring a high-voltage gas relay to execute the above-mentioned method for monitoring a high-voltage gas relay.

[0015] In a fourth aspect, a computer-readable storage medium is provided, wherein instructions are stored in the computer-readable storage medium, and when the instructions run on a computer, the computer is enabled to execute the above-mentioned method for monitoring a high-voltage gas relay.

[0016] In the technical solution provided by this application, the closed-loop control system is used to accurately collect the SF6 gas pressure, density, and temperature parameters, realizing the comprehensive perception of the operating state of the relay control circuit, greatly improving the accuracy and integrity of data collection. The control deviation calculation and residual vector analysis are carried out on the operating data of the relay control circuit to generate control performance indicators, enabling the equipment performance evaluation to move from qualitative to quantitative and solving the problem of fuzzy evaluation criteria in traditional monitoring methods. Inputting the relay control performance indicators into the isolation forest algorithm for anomaly pattern recognition is the core innovation point of this solution. As an unsupervised learning algorithm specifically designed for anomaly detection, the isolation forest algorithm determines the degree of anomaly by calculating the isolation path length of samples, abandoning the limitations of traditional fixed-threshold monitoring methods and significantly improving the accuracy of anomaly detection. This algorithm constructs a random decision tree forest using the distribution characteristics of the data itself, has good processing ability for high-dimensional data, and is particularly suitable for the complex state monitoring of equipment such as high-pressure gas relays. Its characteristics of high calculation efficiency and small required sample size also meet the requirements of real-time monitoring of power equipment. The evaluation result of the health state of the control circuit obtained by the isolation forest algorithm accurately reflects the actual condition of the equipment, providing a reliable basis for subsequent control optimization. According to the evaluation result of the health state of the control circuit, the adaptive control parameter adjustment is performed, and the iterative learning control algorithm is used to optimize the PID parameters, realizing the intelligent adaptability of the control strategy and solving the problem that the traditional fixed-parameter PID controller cannot cope with equipment aging and operating environment changes. The iterative learning control algorithm makes full use of the empirical knowledge in the historical operating data of the equipment, optimizes the control input by continuously learning the historical error trajectory, and has strong pertinence and adaptability. The application of this algorithm in the control of high-pressure gas relays enables the control parameters to be dynamically adjusted according to the health state of the equipment, adopting differentiated control strategies for different anomaly types and levels, greatly improving the control accuracy and reliability. The finally generated relay control parameter set covers various working conditions such as normal mode, mild anomaly mode, moderate anomaly mode, and safety mode, realizing the seamless switching of control strategies and ensuring the safe and stable operation of the equipment in various states.

[0017] The present invention combines artificial intelligence algorithms with traditional control theories, forms an intelligent monitoring and control method for the characteristics and application scenarios of high-pressure gas relays, effectively solves the technical problems faced by the monitoring and control of high-pressure gas relays in the power system, and provides strong support for improving the reliability of power grid operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0019] Figure 1 This is a schematic diagram of an embodiment of the method for monitoring a high-voltage gas relay in an embodiment of the present application; Figure 2 This is a schematic diagram of an embodiment of the system for monitoring a high-voltage gas relay in an embodiment of the present application; Figure 3 This is a structural schematic block diagram of the device for monitoring a high-voltage gas relay in an embodiment of the present invention. Detailed implementation manners

[0020] The embodiments of the present application provide a method, a system, and a storage medium for monitoring a high-voltage gas relay. The terms "first", "second", "third", "fourth", etc. (if any) in the specification, claims, and the above-mentioned drawings of the present application are used to distinguish similar objects, and do not necessarily need to be used to describe a specific order or sequence. It should be understood that such used data can be interchanged under appropriate circumstances so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" or "have" and any deformation thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily need to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0021] For ease of understanding, the specific process of the embodiments of the present application will be described below. Please refer to Figure 1 , an embodiment of the method for monitoring a high-voltage gas relay in an embodiment of the present application includes: Step S101: Perform data acquisition on the SF6 gas pressure, density, and temperature parameters of the high-voltage gas relay based on a closed-loop control system to obtain the operation data of the relay control loop; Step S102: Perform control deviation calculation and residual vector analysis on the operation data of the relay control loop to obtain the relay control performance index; Step S103: Input the relay control performance index into the isolation forest algorithm for abnormal pattern recognition, determine the degree of abnormality by calculating the isolation path length of the sample, and obtain the evaluation result of the health state of the control loop; Step S104: Perform adaptive control parameter adjustment according to the evaluation result of the health state of the control loop, optimize the PID parameters using the iterative learning control algorithm, and generate a relay control parameter set.

[0022] It can be understood that the execution subject of the present application can be a system for monitoring a high-voltage gas relay, or a terminal or a server. Specifically, it is not limited here. The embodiments of the present application will be described by taking the server as the execution subject as an example.

[0023] Specifically, data acquisition of the SF6 gas pressure, density, and temperature parameters of the high-voltage gas relay is performed based on a closed-loop control system. In this step, the physical parameters of the SF6 gas are first collected through pressure sensors, density sensors, and temperature sensors installed in the gas chamber of the high-voltage gas relay, and the collected data is marked with a time stamp accurate to milliseconds to form the original parameter time-series data. Subsequently, the three-sigma principle outlier detection is performed on these original time-series data, and the data points deviating more than three standard deviations from the mean are marked as outliers. For single-point outliers, the average of the previous and subsequent time points is used for replacement, and for continuous outliers, they are repaired by linear interpolation. The repaired data is input into a Kalman filter for noise elimination, and after standardization, the data with different dimensions is uniformly converted into standard distribution data. The statistical features within a ten-minute sliding window are calculated for the standardized data, including the mean, variance, maximum value, minimum value, kurtosis, and skewness. Finally, in combination with the action time data recorded by the relay action mechanism sensor, the deviation between the controller output signal and the actual system response is calculated to generate the relay control loop operation data. Control deviation calculation and residual vector analysis are performed on the relay control loop operation data. In this step, the data is first divided into multiple pressure adjustment cycles according to the change points of the SF6 gas pressure control instruction, and the difference between the pressure set value and the actual pressure value within each adjustment cycle is calculated to form a gas pressure control deviation sequence. Based on this sequence, the integral absolute error and the integral time absolute error are calculated, and the gas pressure control accuracy evaluation value is obtained through weighted summation. The evaluation value is compared with the relay historical reference value to calculate the gas pressure control performance degradation rate, and at the same time, the relay insulation performance attenuation index is obtained by combining the gas insulation strength analysis. By calculating the difference between the theoretical gas pressure response and the actual gas pressure response, a gas relay residual vector is constructed, and after structured analysis of the residual vector, the relay leakage characteristic distribution is obtained. Finally, the relay leakage characteristic distribution and the insulation performance attenuation index are comprehensively evaluated by a weighted fusion algorithm to generate the relay control performance index.

[0024] Input the relay control performance indicators into the isolation forest algorithm for abnormal pattern recognition. In this step, first perform data standardization processing and feature space transformation on the relay control performance indicators to obtain a standardized control performance feature set. Based on this feature set, construct an isolation forest model. Build multiple decision trees by randomly selecting sample subsets and features. At each node, randomly select features and splitting thresholds for data splitting until the samples are completely isolated or reach the preset maximum depth to obtain an abnormal sample detection structure. Input the standardized feature set into this structure, record the path length of each feature sample from the root node to the leaf node, take the average value as the isolation path length, and obtain abnormal separation data. Perform normalization processing on the abnormal separation data, convert the path length into an abnormal index between 0 and 1. The closer the value is to 1, the higher the probability of abnormality. Set multi-level abnormal thresholds according to the abnormal probability value, classify the abnormalities into four levels: mild, moderate, severe, and critical, and calibrate the thresholds according to the characteristics of the high-pressure gas relay. Finally, analyze the contribution degree of each index to the abnormality, determine the key parameters affecting the relay performance, and generate an evaluation result of the health status of the control loop.

[0025] Execute adaptive control parameter adjustment according to the evaluation result of the health status of the control loop. In this step, first determine the control parameter adjustment strategy according to the abnormal level and type, determine different adjustment weights for different types of abnormalities to obtain a control parameter adjustment plan. Apply this plan to the iterative learning control process, design a learning gain matrix for the control characteristics of the high-pressure gas relay, and analyze historical control data to obtain control trajectory optimization data. Apply this data to the PID controller, predict the deviation trend of the current cycle by analyzing the error trajectory of the previous control cycle, adjust the control output to obtain a compensated control signal. Based on this signal, perform automatic adjustment calculation of PID parameters, optimize the proportional, integral, and differential coefficients. After verifying the robustness of the optimized parameter group, form a complete control strategy, including four sets of parameter groups for normal mode, mild abnormal mode, moderate abnormal mode, and safety mode, and generate a relay control parameter set.

[0026] In a specific embodiment, the process of executing step S101 may specifically include the following steps: Collect the physical parameters of SF6 gas through the pressure sensor, density sensor, and temperature sensor installed in the gas chamber of the high-pressure gas relay, and mark the collected data with timestamps to obtain the original parameter time series data; Perform abnormal value detection on the original parameter time series data using the three-standard-deviation principle, replace single-point abnormal values with the average value of the previous and subsequent time points, and use linear interpolation to repair continuous abnormal values to obtain the cleaned parameter data; Input the cleaned parameter data into the Kalman filter, perform state estimation based on the process noise covariance matrix and the measurement noise covariance matrix to obtain the filtered parameter data; Perform closed-loop control system standardization processing on the filtered parameter data, convert data with different dimensions into standard distribution data, and obtain the standardized control loop parameter characteristics; Calculate the statistics within a ten-minute sliding window based on the standardized control loop parameter characteristics, including mean, variance, maximum value, minimum value, kurtosis, and skewness, to obtain a multi-dimensional control loop feature vector; Calculate the deviation between the controller output signal and the actual system response according to the multi-dimensional control loop feature vector and the action time data recorded by the relay action mechanism sensor, and generate the relay control loop operation data.

[0027] Specifically, install a pressure sensor, a density sensor, and a temperature sensor in the gas chamber of the high-voltage gas relay to collect the pressure, density, and temperature parameters of the SF6 gas respectively. The pressure sensor uses a piezoresistive sensor with a measurement range of 0 - 1 MPa and an accuracy of 0.01 MPa; the density sensor uses a thermal conductivity sensor with a measurement range of 0 - 50 kg / m³ and an accuracy of 0.001 kg / m³; the temperature sensor uses a PT100 platinum resistance with a measurement range of -40°C to +120°C and an accuracy of 0.1°C. The acquisition frequency is set to 5 times per second for pressure, and 1 time per second for density and temperature. At the same time, attach a time stamp accurate to milliseconds to each data point to form the original parameter time series data. Perform three-standard-deviation principle outlier detection on the original parameter time series data. This method first calculates the mean and standard deviation of each type of parameter data, and marks the data points that deviate from the mean by more than three standard deviations as outliers. For single-point outliers, that is, isolated outliers where the data points before and after are normal values, the average value of the previous and subsequent time points is used for replacement; for continuous outliers, that is, the situation where multiple consecutive data points are abnormal, linear interpolation is used for repair. For example, if the previous normal value of consecutive outliers is a, the subsequent normal value is b, and there are n outliers in the middle, the repaired value of the i-th outlier is a + (b - a) × i / (n + 1). After processing in this way, the cleaned parameter data is obtained.

[0028] Kalman filtering is a recursive estimation algorithm. By establishing a system state equation and an observation equation, and combining the process noise covariance matrix and the measurement noise covariance matrix, it performs an optimal estimation of the system state. In this method, the process noise covariance matrix represents the uncertainty in the system state transition process, and the measurement noise covariance matrix represents the uncertainty in the sensor measurement process. These two matrices are obtained through offline calibration using historical data. Kalman filtering effectively eliminates the random noise in the sensor data through a prediction-correction iterative process, obtaining the smoothed parameter data after filtering. The filtered parameter data is subjected to standardization processing for the closed-loop control system. Since the units and dimensions of the three physical quantities of pressure, density, and temperature are different, they need to be converted into dimensionless standard distribution data. The standardization processing uses the Z-score method, subtracting the mean from each parameter data point and then dividing by the standard deviation, so that different parameter data are all converted into standard normal distribution data with a mean of 0 and a standard deviation of 1. The processed data is convenient for comparison and fusion in subsequent analysis, obtaining the standardized control loop parameter characteristics.

[0029] Calculate the statistics within a ten-minute sliding window based on the standardized control loop parameter characteristics. The sliding window refers to a fixed-length time period that slides on the time series. In this method, it is set to ten minutes, and the window moves one minute each time. Within each window, calculate six statistical characteristics of each parameter: the mean reflects the average level of the parameter, the variance reflects the degree of fluctuation of the parameter, the maximum and minimum values reflect the extreme value range of the parameter, the kurtosis reflects the sharpness of the data distribution, and the skewness reflects the symmetry of the data distribution. By calculating these statistics, the time series data is converted into a feature vector, obtaining a multi-dimensional control loop feature vector. According to the multi-dimensional control loop feature vector and the action time data recorded by the relay action mechanism sensor, calculate the deviation between the controller output signal and the actual system response. The controller output signal refers to the control instruction issued by the system, and the actual system response refers to the action actually executed by the relay. By comparing the difference between the theoretical response time and the actual response time, calculate the time delay; by comparing the difference between the theoretical response amplitude and the actual response amplitude, calculate the amplitude error. Take the time delay and the amplitude error as control deviation indicators, and combine with the aforementioned multi-dimensional feature vector to generate the complete relay control loop operation data.

[0030] In a specific embodiment, the process of executing step S102 may specifically include the following steps: Perform segmented processing on the relay control loop operation data. According to the change points of the SF6 gas pressure control instruction, divide the data into multiple pressure adjustment cycles to obtain the gas pressure control segmented data; Calculate the difference between the pressure set value and the actual pressure value for each adjustment cycle in the gas pressure control segmented data to form a gas pressure control deviation sequence, obtaining the gas pressure control deviation data; Calculate the integral absolute error and integral time absolute error of the gas relay pressure control loop based on the air pressure control deviation data, perform weighted summation on the error data to obtain the evaluation value of the gas pressure control accuracy; Compare the evaluation value of the gas pressure control accuracy with the relay historical reference value, calculate the degradation rate of the gas pressure control performance, and combine with the analysis of the gas insulation strength to obtain the relay insulation performance attenuation index; Construct a residual vector of the gas relay based on the air pressure control deviation data, and through calculating the difference between the theoretical gas pressure response and the actual gas pressure response, perform structural analysis on the residual vector to obtain the relay leakage characteristic distribution; According to the relay leakage characteristic distribution and the relay insulation performance attenuation index, comprehensively evaluate the state of the relay gas system through a weighted fusion algorithm to generate the relay control performance index.

[0031] Specifically, perform segmented processing on the relay control loop operation data, and divide the data by identifying the change points of the SF6 gas pressure control instruction. The change point refers to the moment when the control instruction is adjusted from one pressure set value to another, representing the start of a control adjustment cycle. The slope change detection method is used to identify the change point, calculate the first-order difference of the pressure set value, and when the difference value exceeds the preset threshold (usually set to 0.05 MPa), it is determined as a change point. The data between two adjacent change points is divided into an adjustment cycle to form the gas pressure control segmented data. Each segmented data contains information such as time stamp, pressure set value, and actual pressure value. Process the segmented gas pressure control segmented data, and calculate the pressure control deviation within each adjustment cycle. The pressure control deviation is defined as the difference between the pressure set value and the actual pressure value, and is arranged in time series to form the gas pressure control deviation sequence. For the sampling moment t, the deviation value is equal to the pressure set value at that moment minus the actual pressure value. The deviation sequence reflects the tracking accuracy of the control system to the instruction, and the smaller the deviation, the better the control performance. These calculated air pressure control deviation data are the basic data for evaluating the performance of the control loop.

[0032] Based on the air pressure control deviation data, two important control performance indicators are calculated: the integral absolute error and the integral time absolute error. The integral absolute error refers to the cumulative sum of the absolute values of all deviations within an adjustment period, reflecting the overall magnitude of the deviations; the integral time absolute error accumulates the absolute value of each deviation multiplied by the corresponding time weight, paying more attention to the deviations with a long duration. The calculation of the integral absolute error is to sum the absolute values of the deviations at each sampling moment within the adjustment period; the calculation of the integral time absolute error is to sum the product of each absolute deviation value and its corresponding time (relative to the start time of the adjustment). These two error indicators are weighted and fused. Generally, weights of 0.6 and 0.4 are used for the integral absolute error and the integral time absolute error respectively, and weighted summation is performed to obtain the evaluation value of the gas pressure control accuracy. The smaller this value is, the higher the control accuracy. The calculated evaluation value of the gas pressure control accuracy is compared with the historical reference value of the relay. The historical reference value refers to the typical control accuracy evaluation value of this type of relay under normal operating conditions, usually taken from the test data in the initial stage of equipment commissioning or the reference data provided by the manufacturer. By calculating the ratio of the current evaluation value to the reference value, the control performance degradation rate is obtained. At the same time, combined with the analysis of the gas insulation strength, the gas insulation strength is affected by the purity, density, and temperature of the SF6 gas. The content of decomposition products in the SF6 gas, such as carbon tetrafluoride and sulfur tetrafluoride, is analyzed through a professional gas analyzer. An increase in the content of these decomposition products indicates a decrease in the gas insulation ability. Considering the control performance degradation rate and the gas insulation analysis results comprehensively, the relay insulation performance attenuation index is obtained, which reflects the degree of deterioration of the equipment insulation performance over time.

[0033] Based on the air pressure control deviation data, a residual vector of the gas relay is constructed. The residual vector refers to the difference sequence between the predicted gas pressure response of the theoretical model and the actual gas pressure response. The theoretical response is calculated based on the physical model and control parameters of the relay, reflecting the performance that the equipment should have under ideal conditions. When constructing the residual vector, first establish the dynamic model of the relay, then substitute the actual control input into the model to calculate the theoretical response, and subtract the actual response point by point to obtain the residual. Perform a structured analysis on the residual vector, including frequency domain analysis, time series feature extraction, etc., to identify the regular patterns in the residual. These patterns are usually related to specific types of equipment defects. Through residual analysis, the relay leakage feature distribution is obtained. The leakage feature shows a trend of slow pressure drop in the residual. Based on the relay leakage feature distribution and the relay insulation performance attenuation index, the state of the relay gas system is comprehensively evaluated through a weighted fusion algorithm. The weighted fusion algorithm assigns weights according to the influence degree of the two indicators on the equipment performance. The leakage feature is directly related to the maintenance of the gas pressure and is given a higher weight (such as 0.7); the insulation performance attenuation is closely related to the safe operation of the equipment and is given an appropriate weight (such as 0.3). The relay control performance index is calculated through weighted summation.

[0034] In a specific embodiment, the process of executing step S103 may specifically include the following steps: Perform data standardization processing on the relay control performance indicators and perform feature space transformation to obtain a standardized control performance feature set; Based on the standardized control performance feature set, construct an isolation forest model. Build multiple decision trees by randomly selecting sample subsets and randomly selecting features. Randomly select a feature and a splitting threshold at each node for data splitting until the samples are completely isolated or reach the preset maximum depth to obtain an abnormal sample detection structure; Specifically, divide the standardized control performance feature set into sample subsets. Use the random sampling method to extract multiple sample subsets from the feature set, keep the number of samples in each subset equal to obtain a set of training sample subsets; construct a decision tree forest based on the set of training sample subsets, build a completely random decision tree for each sample subset, and set the total number of decision trees to the logarithm of the total number of samples to obtain an initial forest structure; control the construction process of each decision tree in the initial forest structure, set the maximum tree depth to the logarithm of the feature dimension to prevent overfitting to obtain a set of decision trees with limited depth; perform node splitting operations based on the set of decision trees with limited depth, randomly select a feature dimension for each internal node, and randomly select a splitting threshold within the value range of this feature to divide the data into left and right subtrees to obtain a randomly split decision tree structure; calculate the sample path according to the randomly split decision tree structure, record the average path length of each normal sample in the training set from the root node to the leaf node, establish the path length distribution of the normal operating state of the high-pressure gas relay to obtain a normal state reference model; calibrate the parameters of the normal state reference model, adjust the abnormal determination boundary according to the characteristics of the gas relay, and set the conversion coefficient of the abnormal degree in combination with expert knowledge to form an abnormal sample detection structure.

[0035] Input the standardized control performance feature set into the abnormal sample detection structure, record the path length of each feature sample point from the root node to the leaf node, take the average value of the path lengths of each decision tree as the isolation path length to obtain the abnormal separation data of the feature samples; Perform normalization processing according to the abnormal separation data of the feature samples, convert the average path length into a standardized abnormal index between zero and one, and the closer the value is to one, the higher the abnormal probability, to obtain the relay control abnormal probability value; Set multi-level abnormal thresholds based on the relay control abnormal probability value, divide the abnormal conditions into four levels: minor abnormality, moderate abnormality, severe abnormality, and critical abnormality, and calibrate the thresholds according to the characteristics of the high-pressure gas relay to obtain the relay abnormal level result; Analyze the contribution degree of each control performance index to the abnormality according to the relay abnormality level result, determine the key parameters affecting the relay performance through feature importance calculation, and generate the evaluation result of the health state of the control loop.

[0036] Specifically, perform data standardization processing on the relay control performance indicators to eliminate the influence of different index dimension differences. The standardization processing adopts the maximum-minimum normalization method, subtracts the minimum value of each control performance index and then divides it by the range (the difference between the maximum value and the minimum value), and converts it into a value between 0 and 1. For example, for the integral absolute error index, if its range is 10 to 100 MPa·s, a data point with a measured value of 40 MPa·s will get (40 - 10) / (100 - 10) = 0.33 after standardization processing. After performing standardization processing on all control performance indicators respectively, perform feature space transformation, reduce the dimension of multi-dimensional features through the principal component analysis method, retain the main information, and obtain the standardized control performance feature set. Build an isolation forest model based on the standardized control performance feature set. This is an algorithm designed specifically for anomaly detection. The basic principle is that abnormal data points are more likely to be "isolated". The specific implementation is divided into six detailed steps: First, divide the standardized control performance feature set into sample subsets, use the random sampling method to extract multiple sample subsets from the feature set, each subset contains a part of the samples of the original data, and keep the sample numbers between subsets equal. For example, if there are 1000 historical operation data samples of high-pressure gas relays, they can be divided into 100 subsets, each subset contains 250 randomly selected samples, and thus a set of training sample subsets is obtained.

[0037] Construct a decision tree forest based on a set of training sample subsets, and build a completely random decision tree for each sample subset. "Completely random" means randomly selecting features and splitting thresholds when splitting nodes, rather than searching for the optimal split. The total number of decision trees is usually set to the logarithm of the total number of samples. For example, for 1000 samples, about log(1000) ≈ 7 trees can be set to form an initial forest structure. In the third step, control the construction process of each decision tree in the initial forest structure, and set the maximum tree depth to the logarithm of the feature dimension to prevent the decision tree from overgrowing and causing overfitting. For example, if the feature dimension is 16, the maximum tree depth is set to log(16) ≈ 4 to obtain a set of decision trees with limited depth. Perform node splitting operations based on the set of decision trees with limited depth. Randomly select a feature dimension for each internal node, such as the pressure control accuracy, insulation performance index, etc., and randomly select a splitting threshold within the value range of this feature to divide the data into left and right subtrees. This random splitting is different from the way of traditional decision trees to find the optimal splitting point, but emphasizes randomness, which is beneficial to detecting outliers. Through this splitting method, a decision tree structure with random splitting is formed. In the fifth step, calculate the sample paths according to the decision tree structure with random splitting, and record the average path length of each normal sample in the training set from the root node to the leaf node. Normal samples usually require more splits to be isolated, so they have a longer average path length. By statistically analyzing these path lengths, establish the path length distribution of the normal operating state of the high-pressure gas relay to obtain a normal state reference model.

[0038] Calibrate the parameters of the normal state reference model and adjust the anomaly determination boundary according to the characteristics of the gas relay. Gas relays of different models and different operating environments may require different determination criteria. Combine expert knowledge to set the conversion coefficient of the anomaly degree to form the final anomaly sample detection structure. After completing the construction of the isolation forest model, input the standardized control performance feature set into the anomaly sample detection structure, record the path length of each feature sample point from the root node to the leaf node in all decision trees, take the average value of the path lengths of each decision tree as the isolation path length, and obtain the anomaly separation data of the feature samples.

[0039] Normalize the abnormal separation data of the feature samples, and convert the average path length into a standardized abnormal index between 0 and 1. The conversion process takes into account the influence of the dataset size on the path length, and a correction factor is introduced when calculating the abnormal index. The closer the value is to 1, the higher the probability of abnormality. After this processing, the abnormal probability value of the relay control is obtained. Set multi-level abnormal thresholds based on the abnormal probability value of the relay control, and divide the abnormalities into four levels: minor abnormality (0.6 - 0.75), moderate abnormality (0.75 - 0.85), severe abnormality (0.85 - 0.95), and critical abnormality (>0.95). Calibrate the thresholds according to the characteristics of the high-pressure gas relay, considering factors such as the importance of the equipment and the operating environment, and obtain the abnormal level results of the relay. Analyze the contribution degree of each control performance index to the abnormality according to the abnormal level results of the relay. Determine the key parameters affecting the relay performance through feature importance calculation. The method is to analyze the frequency of each feature being selected as the splitting feature and the change range of the abnormal score caused by its splitting. The identified key parameters will help determine the fault type and cause, and form a complete evaluation result of the health status of the control loop.

[0040] In a specific embodiment, the process of executing step S104 may specifically include the following steps: Perform a grading process on the evaluation result of the health status of the control loop, determine the control parameter adjustment strategy according to the abnormal level and type, determine different adjustment weights for different types of abnormalities, and obtain the control parameter adjustment plan; Perform iterative learning control on the control parameter adjustment plan, design a learning gain matrix according to the control characteristics of the high-pressure gas relay, analyze and process the historical control data, and obtain the optimized control trajectory data; Specifically, batch process the historical operation data of the high-pressure gas relay. Each batch contains control data under similar working conditions to obtain a control data learning set; extract the control error trajectory features based on the control data learning set, calculate the shape features of the time-domain error curve and the frequency-domain error distribution features of each control cycle to obtain an error pattern feature library; perform similarity analysis on the error pattern feature library, use the dynamic time warping algorithm to calculate the similarity between the current error trajectory and the historical error trajectories, identify the historical error pattern with the highest similarity, and obtain the error pattern matching result; construct a learning gain matrix according to the error pattern matching result. The learning gain value is dynamically adjusted according to the health status of the control loop. The higher the degree of abnormality, the smaller the learning gain, and obtain an adaptive learning gain matrix; perform iterative calculations on the historical control input based on the adaptive learning gain matrix, correct the control input of the next cycle through the error propagation correction formula, perform ten iterative operations, and obtain an optimized control input sequence; fuse the optimized control input sequence with the current control strategy, determine the fusion weight according to the evaluation result of the health status of the control loop, and smooth the control trajectory to generate the optimized control trajectory data.

[0041] Apply the control trajectory optimization data to the PID controller, predict the deviation trend of the current control cycle by analyzing the error trajectory of the previous control cycle, adjust the control output, and obtain the compensated control signal; Perform automatic adjustment calculation of PID parameters according to the compensated control signal, and dynamically optimize the proportional coefficient, integral coefficient, and differential coefficient by using the error convergence speed and overshoot index through the gradient descent method to obtain the optimized PID parameter set; Conduct robustness verification on the optimized PID parameter set, perform simulation tests on multiple working points within the range of SF6 gas pressure and temperature changes, evaluate the stability of control performance, and obtain the PID controller parameters with robustness; Form a complete control strategy according to the PID controller parameters with robustness and the characteristics of the gas relay, including the normal mode parameter set, the slight abnormal mode parameter set, the moderate abnormal mode parameter set, and the safety mode parameter set, and generate the relay control parameter set.

[0042] Specifically, classify and process the evaluation results of the health status of the control loop, and determine the control parameter adjustment strategy according to the abnormal levels (slight, moderate, severe, and critical) and abnormal types (pressure abnormality, density abnormality, temperature abnormality, etc.) obtained by the isolation forest algorithm. For slight abnormalities, make small-scale parameter fine-tuning; for moderate abnormalities, make medium-scale parameter adjustments; for severe abnormalities, implement large-scale parameter reorganization; for critical abnormalities, switch to the safety mode parameter set. Determine different adjustment weights for different types of abnormalities. For example, mainly adjust the parameters related to the pressure control loop for pressure abnormalities, and focus on adjusting the temperature control parameters for temperature abnormalities. Obtain a targeted control parameter adjustment plan through this method of classifying and implementing policies. Iterative learning control is a control strategy for repetitive processes. By learning the error of the previous control cycle, it optimizes the next control input. Design a learning gain matrix for the control characteristics of the high-pressure gas relay. The learning gain matrix determines the degree to which the control system absorbs historical experience. Too large will cause the system to be unstable, and too small will result in insignificant learning effects. Analyze and process the historical control data to obtain the control trajectory optimization data. This process includes six detailed sub-steps: Batch process the historical operation data of the high-voltage gas relay, group the continuous control data according to the similarity of operating conditions, and each batch contains the control data under similar operating conditions. Similar operating conditions refer to the operating states where factors such as external environmental temperature, load conditions, and control objectives are similar. For example, classify the control data with the environmental temperature in the range of 5°C to 15°C and the load rate between 70% and 80% into one batch, and obtain the control data learning set after such classification. Extract the control error trajectory features based on the control data learning set. The control error trajectory refers to the curve of the change of the difference between the control target value and the actual value over time. By calculating the shape features of the time-domain error curve (such as rise time, peak error, steady-state error, etc.) and the frequency-domain error distribution features (obtaining the distribution of errors in different frequency components through Fourier transform) for each control cycle, a systematic error pattern feature library is formed. Perform similarity analysis on the error pattern feature library, and use the dynamic time warping algorithm to calculate the similarity between the current error trajectory and the historical error trajectories. The dynamic time warping algorithm is a method for measuring the similarity of two time series, which can handle the cases of unequal sequence lengths and different rhythms, and is especially suitable for analyzing control error trajectories. By calculating the similarity between the error trajectory of the current control cycle and all the error trajectories in the historical library, identify the historical error pattern with the highest similarity, and obtain the error pattern matching result. Construct a learning gain matrix according to the error pattern matching result. The learning gain value is dynamically adjusted according to the health status of the control loop. The higher the degree of abnormality, the smaller the learning gain. This design is because when the system has a high degree of abnormality, the reference value of historical experience decreases, and over-reliance may exacerbate the problem. Allocate different learning gains according to the importance and sensitivity of different control variables to form an adaptive learning gain matrix. Based on the adaptive learning gain matrix, perform iterative calculations on the historical control inputs, and correct the control input of the next cycle through the error propagation correction formula. The error propagation correction formula multiplies the control error of the current cycle by the learning gain and adds it to the control input of the next cycle to achieve feedforward compensation. This process usually performs ten iterative operations, and each iteration is further optimized based on the previous result, and finally an optimized control input sequence is obtained. Integrate the optimized control input sequence with the current control strategy, and determine the integration weight according to the evaluation result of the health status of the control loop. For the control loop with good health status, it is biased towards the current control strategy; for the control loop with poor health status, more of the optimized input sequence is adopted. Smooth the control trajectory to eliminate possible jump points and generate smooth control trajectory optimization data.

[0043] Apply the control trajectory optimization data to the PID controller, which is the most commonly used type of controller in industrial control systems and consists of three parts: proportional, integral, and derivative. Predict the deviation trend of the current control cycle by analyzing the error trajectory of the previous control cycle, adjust the control output, and obtain the compensated control signal. Perform automatic adjustment calculations of PID parameters based on the compensated control signal, and evaluate the control performance using the error convergence speed and overshoot indicators. The error convergence speed reflects the speed of system response, and the overshoot reflects the system stability. Dynamically optimize the proportional coefficient, integral coefficient, and derivative coefficient through the gradient descent method. The gradient descent method adjusts the parameters along the direction where the performance index decreases fastest, iteratively solves the optimal value, and obtains the optimized PID parameter set. Conduct robustness verification on the optimized PID parameter set, and perform simulation tests for multiple operating points within the range of SF6 gas pressure and temperature changes. The selection of operating points usually covers extreme conditions of equipment operation, such as combinations under the highest temperature, lowest temperature, highest pressure, lowest pressure, etc. Screen out the PID controller parameters with good robustness by evaluating the control performance stability under different operating points, such as step response, disturbance rejection ability, etc. Form a complete control strategy based on the robust PID controller parameters and the characteristics of the gas relay, including normal mode parameter sets, slightly abnormal mode parameter sets, moderately abnormal mode parameter sets, and safety mode parameter sets, and generate a relay control parameter set.

[0044] The method for monitoring the high-voltage gas relay in the embodiment of the present application has been described above. Next, the system for monitoring the high-voltage gas relay in the embodiment of the present application will be described. Please refer to Figure 2 , an embodiment of the system for monitoring the high-voltage gas relay in the embodiment of the present application includes: An acquisition module 201, configured to perform data acquisition on the SF6 gas pressure, density, and temperature parameters of the high-voltage gas relay based on a closed-loop control system to obtain relay control loop operation data; An analysis module 202, configured to perform control deviation calculation and residual vector analysis on the relay control loop operation data to obtain relay control performance indicators; An identification module 203, configured to input the relay control performance indicators into an isolation forest algorithm for abnormal mode identification, determine the degree of abnormality by calculating the sample isolation path length, and obtain an evaluation result of the health status of the control loop; An execution module 204, configured to perform adaptive control parameter adjustment according to the evaluation result of the health status of the control loop, optimize the PID parameters using an iterative learning control algorithm, and generate a relay control parameter set.

[0045] Above Figure 2The system for monitoring a high-voltage gas relay in the embodiments of the present invention will be described in detail from the perspective of modular functional entities. Next, the device for monitoring a high-voltage gas relay in the embodiments of the present invention will be described in detail from the perspective of hardware processing.

[0046] Figure 3 FIG. 4 is a schematic structural diagram of a device for monitoring a high-voltage gas relay provided by an embodiment of the present invention. The device 300 for monitoring a high-voltage gas relay may vary greatly due to different configurations or performances, and may include one or more processors (central processing units, CPUs) 310 (for example, one or more processors) and a memory 320, and one or more storage media 330 for storing application programs 333 or data 332 (for example, one or more mass storage device terminals). Among them, the memory 320 and the storage media 330 may be transient storage or persistent storage. The program stored in the storage media 330 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations on the device 300 for monitoring a high-voltage gas relay. Further, the processor 310 may be configured to communicate with the storage media 330 and execute a series of instruction operations in the storage media 330 on the device 300 for monitoring a high-voltage gas relay to implement the steps of the method for monitoring a high-voltage gas relay described above.

[0047] The device 300 for monitoring a high-voltage gas relay may further include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input / output interfaces 360, and / or one or more operating systems 331, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, and so on. Those skilled in the art can understand that Figure 3 the shown structural diagram of the device for monitoring a high-voltage gas relay does not limit the device for monitoring a high-voltage gas relay provided by the present invention, and may include more or fewer components than shown in the figure, or combine some components, or have different component arrangements.

[0048] The present invention also provides a computer-readable storage medium. The computer-readable storage medium may be a non-volatile computer-readable storage medium, or may also be a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions run on a computer, the computer is caused to execute the steps of the method for monitoring a high-voltage gas relay.

[0049] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the foregoing method embodiments, and will not be described herein again.

[0050] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a device for monitoring a high-voltage gas relay (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs that can store program codes.

[0051] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for monitoring a high-voltage gas relay, characterized in that, The method includes: Performing data acquisition on the SF6 gas pressure, density, and temperature parameters of the high-voltage gas relay based on a closed-loop control system to obtain the operation data of the relay control circuit; Calculating the control deviation and performing residual vector analysis on the operation data of the relay control circuit to obtain the relay control performance index; Inputting the relay control performance index into the isolation forest algorithm for abnormal pattern recognition, determining the degree of abnormality by calculating the isolation path length of the sample, and obtaining the evaluation result of the health state of the control circuit; Performing adaptive control parameter adjustment according to the evaluation result of the health state of the control circuit, optimizing the PID parameters using the iterative learning control algorithm, and generating a set of relay control parameters.

2. The method for monitoring a high-voltage gas relay according to claim 1, characterized in that, The performing data acquisition on the SF6 gas pressure, density, and temperature parameters of the high-voltage gas relay based on a closed-loop control system to obtain the operation data of the relay control circuit includes: Collecting the physical parameters of the SF6 gas through the pressure sensor, density sensor, and temperature sensor installed in the gas chamber of the high-voltage gas relay, and marking the collected data with time stamps to obtain the original parameter time series data; Performing outlier detection on the original parameter time series data according to the three-standard-deviation principle, replacing the single-point outlier with the average value of the front and back time points, and repairing the continuous outliers using the linear interpolation method to obtain the cleaned parameter data; Inputting the cleaned parameter data into the Kalman filter, and performing state estimation based on the process noise covariance matrix and the measurement noise covariance matrix to obtain the filtered parameter data; Performing standardization processing on the filtered parameter data in the closed-loop control system, converting the data with different dimensions into standard distribution data, and obtaining the standardized control circuit parameter characteristics; Calculating the statistics within a ten-minute sliding window based on the standardized control circuit parameter characteristics, including mean, variance, maximum value, minimum value, kurtosis, and skewness, to obtain a multi-dimensional control circuit feature vector; Calculating the deviation between the controller output signal and the actual system response according to the multi-dimensional control circuit feature vector and the action time data recorded by the relay action mechanism sensor, and generating the operation data of the relay control circuit.

3. The method for monitoring a high-voltage gas relay according to claim 1, characterized in that, The calculating the control deviation and performing residual vector analysis on the operation data of the relay control circuit to obtain the relay control performance index includes: Performing segmented processing on the operation data of the relay control circuit, and dividing the data into multiple pressure adjustment cycles according to the change points of the SF6 gas pressure control instruction to obtain the segmented data of gas pressure control; Calculating the difference between the pressure set value and the actual pressure value for each adjustment cycle in the segmented data of gas pressure control to form a gas pressure control deviation sequence, and obtaining the gas pressure control deviation data; Calculating the integral absolute error and the integral time absolute error of the gas relay pressure control circuit according to the gas pressure control deviation data, performing weighted summation on the error data, and obtaining the evaluation value of the gas pressure control accuracy; Comparing the evaluation value of the gas pressure control accuracy with the historical reference value of the relay, calculating the gas pressure control performance degradation rate, and combining the gas insulation strength analysis to obtain the relay insulation performance attenuation index; Construct a residual vector of the gas relay based on the air pressure control deviation data. By calculating the difference between the theoretical gas pressure response and the actual gas pressure response, perform a structural analysis on the residual vector to obtain the relay leakage characteristic distribution; According to the relay leakage characteristic distribution and the relay insulation performance degradation index, comprehensively evaluate the state of the relay gas system through a weighted fusion algorithm to generate the relay control performance index.

4. The method for monitoring a high-voltage gas relay according to claim 1, wherein Input the relay control performance index into the isolation forest algorithm for abnormal pattern recognition. Determine the degree of abnormality by calculating the sample isolation path length to obtain the evaluation result of the control loop health status, including: Perform data standardization processing on the relay control performance index and perform feature space transformation to obtain a standardized control performance feature set; Construct an isolation forest model based on the standardized control performance feature set. Construct multiple decision trees by randomly selecting sample subsets and randomly selecting features. Randomly select a feature and a splitting threshold at each node for data splitting until the samples are completely isolated or reach the preset maximum depth to obtain the abnormal sample detection structure; Input the standardized control performance feature set into the abnormal sample detection structure, record the path length of each feature sample point from the root node to the leaf node, and take the average value of the path lengths of each decision tree as the isolation path length to obtain the abnormal separation data of the feature samples; Perform normalization processing according to the abnormal separation data of the feature samples, convert the average path length into a standardized abnormal index between zero and one, and the closer the value is to one, the higher the probability of abnormality, to obtain the relay control abnormal probability value; Set multi-level abnormal thresholds based on the relay control abnormal probability value, divide the abnormality into four levels: minor abnormality, moderate abnormality, severe abnormality, and critical abnormality, and calibrate the thresholds according to the characteristics of the high-pressure gas relay to obtain the relay abnormal level result; Analyze the contribution degree of each control performance index to the abnormality according to the relay abnormal level result, determine the key parameters affecting the relay performance through feature importance calculation, and generate the evaluation result of the control loop health status.

5. The method for monitoring a high-voltage gas relay according to claim 4, characterized in that, Construct an isolation forest model based on the standardized control performance feature set. Construct multiple decision trees by randomly selecting sample subsets and randomly selecting features. Randomly select a feature and a splitting threshold at each node for data splitting until the samples are completely isolated or reach the preset maximum depth to obtain the abnormal sample detection structure, including: Perform sample subset division on the standardized control performance feature set, use the random sampling method to extract multiple sample subsets from the feature set, and keep the number of samples in each subset equal to obtain a set of training sample subsets; Construct a decision tree forest based on the set of training sample subsets, construct a completely random decision tree for each sample subset, and set the total number of decision trees to the logarithm of the total sample number to obtain the initial forest structure; Perform construction process control on each decision tree in the initial forest structure, set the maximum tree depth to the logarithm of the feature dimension to prevent overfitting, and obtain a set of decision trees with limited depth; Perform node splitting operations based on the set of depth - limited decision trees. Randomly select a feature dimension for each internal node, and randomly select a splitting threshold within the value range of this feature to divide the data into left and right sub - trees, obtaining a randomly split decision tree structure; Calculate sample paths according to the randomly split decision tree structure, record the average path length of each normal sample in the training set from the root node to the leaf node, establish the path length distribution of the normal operating state of the high - voltage gas relay, and obtain a normal - state reference model; Calibrate the parameters of the normal - state reference model, adjust the abnormal determination boundary according to the characteristics of the gas relay, and set the conversion coefficient of the abnormal degree in combination with expert knowledge to form the abnormal sample detection structure.

6. The method for monitoring a high-voltage gas relay according to claim 1, wherein The above - mentioned performing adaptive control parameter adjustment according to the control loop health status evaluation result, optimizing the PID parameters using the iterative learning control algorithm to generate a relay control parameter set, including: Perform hierarchical processing on the control loop health status evaluation result, determine the control parameter adjustment strategy according to the abnormal level and type, and determine different adjustment weights for different types of abnormalities to obtain a control parameter adjustment plan; Perform iterative learning control on the control parameter adjustment plan, design a learning gain matrix according to the control characteristics of the high - voltage gas relay, analyze and process the historical control data to obtain control trajectory optimization data; Apply the control trajectory optimization data to the PID controller, predict the deviation trend of the current control cycle by analyzing the error trajectory of the previous control cycle, and adjust the control output to obtain a compensated control signal; Perform automatic adjustment calculation of PID parameters according to the compensated control signal, use the error convergence speed and overshoot index, and dynamically optimize the proportional coefficient, integral coefficient, and differential coefficient through the gradient descent method to obtain an optimized PID parameter group; Verify the robustness of the optimized PID parameter group, perform simulation tests for multiple operating points within the range of SF6 gas pressure and temperature changes, evaluate the stability of control performance, and obtain PID controller parameters with robustness; Form a complete control strategy according to the PID controller parameters with robustness and the characteristics of the gas relay, including a normal - mode parameter group, a minor - abnormality mode parameter group, a moderate - abnormality mode parameter group, and a safety - mode parameter group, and generate the relay control parameter set.

7. The method for monitoring a high-voltage gas relay according to claim 6, characterized in that, The above - mentioned performing iterative learning control on the control parameter adjustment plan, designing a learning gain matrix according to the control characteristics of the high - voltage gas relay, analyzing and processing the historical control data to obtain control trajectory optimization data, including: Batch - process the historical operation data of the high - voltage gas relay, with each batch containing control data under similar working conditions, to obtain a control data learning set; Extract control error trajectory features based on the control data learning set, calculate the shape features of the time - domain error curve and the frequency - domain error distribution features of each control cycle to obtain an error pattern feature library; Perform similarity analysis on the error pattern feature library, calculate the similarity between the current error trajectory and the historical error trajectories using the dynamic time warping algorithm, identify the historical error pattern with the highest similarity, and obtain the error pattern matching result; Construct a learning gain matrix based on the error pattern matching result, where the learning gain value is dynamically adjusted according to the health status of the control loop, and the higher the degree of abnormality, the smaller the learning gain, to obtain an adaptive learning gain matrix; Perform iterative calculations on the historical control inputs based on the adaptive learning gain matrix, correct the control inputs for the next cycle through the error propagation correction formula, and perform ten iterative operations to obtain an optimized control input sequence; Fuse the optimized control input sequence with the current control strategy, determine the fusion weight according to the evaluation result of the health status of the control loop, and perform smoothing processing on the control trajectory to generate the optimized data for the control trajectory.

8. A system for monitoring a high-voltage gas relay, characterized in that, For implementing the method for monitoring a high-voltage gas relay as described in any one of claims 1-7, the system for monitoring a high-voltage gas relay includes: An acquisition module for performing data acquisition on the SF6 gas pressure, density, and temperature parameters of the high-voltage gas relay based on a closed-loop control system to obtain the operation data of the relay control loop; An analysis module for performing control deviation calculation and residual vector analysis on the operation data of the relay control loop to obtain the relay control performance indicators; An identification module for inputting the relay control performance indicators into an isolation forest algorithm for abnormal pattern identification, determining the degree of abnormality by calculating the isolation path length of the samples, and obtaining the evaluation result of the health status of the control loop; An execution module for performing adaptive control parameter adjustment according to the evaluation result of the health status of the control loop, optimizing the PID parameters using an iterative learning control algorithm, and generating a relay control parameter set.

9. An apparatus for monitoring a high-voltage gas relay, characterized in that, Comprising a memory and a processor, the memory stores a computer program that can run on the processor, and when the processor executes the computer program, it implements the method for monitoring a high-voltage gas relay as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is run by the processor, it causes the processor to execute the method for monitoring a high-voltage gas relay as described in any one of claims 1 to 7.

Images