Converter Transformer Condition Detection Method and Platform Based on PD Monitoring

Abstract

The present application provides a method and platform for detecting the state of a converter transformer based on PD monitoring, which relates to the technical field of power equipment health monitoring, and includes: monitoring the electrical signal, thermal signal, acoustic signal, and optical signal of partial discharge in the converter transformer, and performing discharge scale analysis and discharge consistency analysis on the electrical signal array, thermal signal array, acoustic signal array, and optical signal array; configuring the state analysis path in the transformer state analysis channel according to the scale information and consistency information, and analyzing to obtain the insulation influence parameters; performing detection to obtain the actual insulation influence parameters, and updating the transformer state analysis channel. Through the present application, it can solve the problem that traditional detection methods usually rely on a single signal for insulation state evaluation, it is difficult to comprehensively reflect the health status of the transformer, and there are deficiencies in the comprehensiveness and accuracy of insulation state detection. By comprehensively analyzing the transformer through multi-signal monitoring, the comprehensiveness and accuracy of insulation state detection can be significantly improved.

Description

Technical Field

[0001] This application relates to the technical field of power equipment health monitoring, and particularly to a converter transformer condition detection method and platform based on PD monitoring. Background Art

[0002] Converter transformers are important equipment in the power system, especially in high-voltage DC power transmission systems, undertaking the core functions of electric energy conversion and transmission. With the long-term operation of converter transformers, the insulating materials of the equipment gradually age and deteriorate, which may lead to problems such as insulation breakdown and partial discharge, and seriously affect the safe and stable operation of the power system in severe cases.

[0003] Currently, traditional methods for detecting the insulation condition of converter transformers mainly rely on the monitoring of single signals, such as electrical signals, partial discharge monitoring, insulation resistance testing, etc. Although these methods can provide some useful insulation health information, it is difficult to comprehensively reflect the health status of the transformer, resulting in insufficient comprehensiveness and accuracy in the insulation condition detection of converter transformers. Summary of the Invention

[0004] The purpose of this application is to provide a converter transformer condition detection method and platform based on PD monitoring, so as to solve the technical problem that traditional methods for detecting the insulation condition of converter transformers usually rely only on single signals for insulation condition evaluation, are difficult to comprehensively reflect the health status of the transformer, and have insufficient comprehensiveness and accuracy in insulation condition detection.

[0005] In view of the above problems, this application provides a converter transformer condition detection method and platform based on PD monitoring.

[0006] In a first aspect, this application provides a converter transformer condition detection method based on PD monitoring, which is implemented through a converter transformer condition detection platform based on PD monitoring, and includes: monitoring the electrical signal, thermal signal, acoustic signal, and optical signal of partial discharge of the converter transformer to obtain an electrical signal array, a thermal signal array, an acoustic signal array, and an optical signal array; performing partial discharge scale analysis and partial discharge consistency analysis on the electrical signal array, thermal signal array, acoustic signal array, and optical signal array to obtain scale information and consistency information; configuring a transformer condition analysis path in the transformer condition analysis channel according to the scale information and consistency information, and analyzing to obtain insulation influence parameters; detecting the insulation influence parameters of the converter transformer to obtain actual insulation influence parameters, and updating the transformer condition analysis channel.

[0007] Preferably, the method for detecting the state of a converter transformer based on PD monitoring further includes: configuring a sensor array at a plurality of insulation positions of the insulation layer of the converter transformer to monitor and collect electrical signals, thermal signals, acoustic signals, and optical signals at the plurality of insulation positions; arranging the electrical signals, thermal signals, acoustic signals, and optical signals at the plurality of insulation positions according to the plurality of insulation positions to obtain an electrical signal array, a thermal signal array, an acoustic signal array, and an optical signal array.

[0008] Preferably, the method for detecting the state of a converter transformer based on PD monitoring further includes: performing signal abnormality analysis on the electrical signal array, the thermal signal array, the acoustic signal array, and the optical signal array to obtain an electrical abnormality array, a thermal abnormality array, an acoustic abnormality array, and an optical abnormality array; calculating the proportion of the abnormalities greater than the abnormality threshold in the electrical abnormality array, the thermal abnormality array, the acoustic abnormality array, and the optical abnormality array to obtain four abnormality ratios, calculating the mean value to obtain scale information; performing consistency analysis calculation according to the four abnormality ratios to obtain consistency information.

[0009] Preferably, the method for detecting the state of a converter transformer based on PD monitoring further includes: extracting a first electrical signal from the electrical signal array; randomly extracting a plurality of historical electrical signals from the electrical signal monitoring data within a historical time; calculating the electrical signal difference between the mean value of the plurality of historical electrical signals and the first electrical signal, and calculating the ratio of the electrical signal difference to the first electrical signal as the first electrical abnormality; continuing to calculate the electrical abnormality of all electrical signals to obtain an electrical abnormality array; performing signal abnormality analysis according to the thermal signal array, the acoustic signal array, and the optical signal array to obtain a thermal abnormality array, an acoustic abnormality array, and an optical abnormality array.

[0010] Preferably, the method for detecting the state of a converter transformer based on PD monitoring further includes: compensating and adjusting the scale information according to the consistency information to obtain adjusted scale information; training a transformer state analysis channel based on ensemble learning, where the transformer state analysis channel includes M transformer state analysis paths, and M is a positive integer; using the adjusted scale information and M to calculate and round up to obtain N, where N is a positive integer less than M; randomly selecting N transformer state analysis paths, inputting the initial insulation parameters and the electrical signal array, the thermal signal array, the acoustic signal array, and the optical signal array, identifying N path insulation influence parameters, and calculating the mean value to obtain insulation influence parameters.

[0011] Preferably, the method for detecting the state of a converter transformer based on PD monitoring further includes: according to the partial discharge monitoring data of the converter transformer within a historical time, collecting a sample electrical signal array, a sample thermal signal array, a sample acoustic signal array, and a sample optical signal array, and testing the insulation resistance of the transformer before and after collecting partial discharges to obtain a sample initial insulation parameter set and a sample insulation influence parameter set; using the sample electrical signal array, the sample thermal signal array, the sample acoustic signal array, the sample optical signal array, and the sample initial insulation parameter set as input data, using the sample insulation influence parameter set as supervised data, and randomly partitioning the input data and the supervised data with replacement to obtain M pieces of supervised training data; using the M pieces of supervised training data to train M transformer state analysis paths, and after the training is completed, combining them to obtain a transformer state analysis channel.

[0012] Preferably, the method for detecting the state of a converter transformer based on PD monitoring further includes: detecting the insulation influence parameters of the converter transformer to obtain actual insulation influence parameters; calculating the insulation influence parameter deviation between the actual insulation influence parameters and the insulation influence parameters, and determining whether it is greater than the insulation influence deviation threshold; if not, then not updating the transformer state analysis path and the transformer state analysis channel; if so, then calculating the ratio of the insulation influence deviation threshold to the insulation influence parameter deviation to obtain an error scale; multiplying the error scale by M and taking the integer to obtain O, where O is a positive integer less than M; using the electrical signal array, the thermal signal array, the acoustic signal array, the optical signal array, combined with the current initial insulation parameters and actual insulation influence parameters of the converter transformer, as updated training data, and randomly selecting O transformer state analysis paths in the transformer state analysis channel for updated training.

[0013] In a second aspect, the present application further provides a state detection platform for a converter transformer based on PD monitoring, which is used to execute the method for detecting the state of a converter transformer based on PD monitoring as described in the first aspect, and includes: a partial discharge signal monitoring module, which is used to monitor the electrical signal, thermal signal, acoustic signal, and optical signal of partial discharge of the converter transformer to obtain an electrical signal array, a thermal signal array, an acoustic signal array, and an optical signal array; a partial discharge feature analysis module, which is used to perform partial discharge scale analysis and partial discharge consistency analysis on the electrical signal array, the thermal signal array, the acoustic signal array, and the optical signal array to obtain scale information and consistency information; an insulation influence parameter analysis module, which is used to configure the transformer state analysis paths in the transformer state analysis channel according to the scale information and the consistency information, and analyze and obtain insulation influence parameters; a state analysis channel update module, which is used to detect the insulation influence parameters of the converter transformer to obtain actual insulation influence parameters, and update the transformer state analysis channel.

[0014] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0015] By monitoring the electrical, thermal, acoustic, and optical signals of partial discharge in the converter transformer, electrical signal arrays, thermal signal arrays, acoustic signal arrays, and optical signal arrays are obtained; then, partial discharge scale analysis and partial discharge consistency analysis are performed on the electrical signal arrays, thermal signal arrays, acoustic signal arrays, and optical signal arrays to obtain scale information and consistency information; then, according to the scale information and consistency information, the transformer state analysis path in the transformer state analysis channel is configured to analyze and obtain insulation influence parameters; further, the insulation influence parameters of the converter transformer are detected to obtain actual insulation influence parameters; finally, the actual insulation influence parameters and the parameter deviation of the insulation influence parameters are analyzed, and the transformer state analysis channel is optimized and updated according to the insulation influence parameter deviation. That is to say, through multi-signal monitoring and comprehensive analysis of the transformer, the health status of the transformer insulation layer can be evaluated more comprehensively and accurately, significantly improving the comprehensiveness and accuracy of insulation state detection, so that potential insulation faults can be detected in time and the probability of faults can be reduced.

[0016] The above description is only an overview of the technical solutions of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features, and advantages of this application more obvious and understandable, the following specific embodiments of this application are specifically given. It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of this application, nor is it used to limit the scope of this application. Other features of this application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in this application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only exemplary, and for those of ordinary skill in the art, other drawings can be obtained according to the provided drawings without creative efforts.

[0018] Figure 1 It is a schematic flow chart of a method for detecting the state of a converter transformer based on PD monitoring in this application;

[0019] Figure 2 It is a schematic structural diagram of a platform for detecting the state of a converter transformer based on PD monitoring in this application.

[0020] Description of the reference numerals:

[0021] Partial discharge signal monitoring module 11, partial discharge feature analysis module 12, insulation influence parameter analysis module 13, status analysis channel update module 14. Specific implementation manner

[0022] By providing a converter transformer status detection method and platform based on PD monitoring, this application solves the technical problem that traditional insulation status detection methods for converter transformers usually rely on a single signal for insulation status evaluation, making it difficult to comprehensively reflect the health status of the transformer and resulting in insufficient comprehensiveness and accuracy in insulation status detection. Through comprehensive analysis of the transformer using multi-signal monitoring, the health status of the transformer insulation layer can be evaluated more comprehensively and accurately, significantly improving the comprehensiveness and accuracy of insulation status detection, thereby enabling timely discovery of potential insulation faults and reducing the probability of faults.

[0023] Next, the technical solutions in this application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments of this application. It should be understood that this application is not limited by the example embodiments described here. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of this application. Additionally, it should be noted that for the sake of description, only parts related to this application are shown in the drawings rather than all of them.

[0024] Embodiment 1. Please refer to the attached Figure 1 , this application provides a converter transformer status detection method based on PD monitoring, which is applied to a converter transformer status detection platform based on PD monitoring and specifically includes the following steps:

[0025] S100: Monitor the electrical signal, thermal signal, acoustic signal, and optical signal of partial discharge in the converter transformer to obtain an electrical signal array, a thermal signal array, an acoustic signal array, and an optical signal array.

[0026] Furthermore, step S100 of this application further includes:

[0027] S110: Configure a sensor array at multiple insulation positions of the converter transformer insulation layer to monitor and collect the electrical signal, thermal signal, acoustic signal, and optical signal of the multiple insulation positions; S120: Arrange the electrical signal, thermal signal, acoustic signal, and optical signal of the multiple insulation positions according to the multiple insulation positions to obtain an electrical signal array, a thermal signal array, an acoustic signal array, and an optical signal array.

[0028] Specifically, a sensor array is installed on the insulation layer of the converter transformer. By monitoring the electrical, thermal, acoustic, and optical signals at multiple insulation positions of the transformer, the insulation status information of different positions can be comprehensively and real-time obtained; these signals will exhibit different characteristics during the operation of the transformer. Especially when partial discharge (PD) occurs, these signals will show significant changes. By comprehensively analyzing the signals at different positions, the insulation health status of the transformer can be detected and diagnosed more accurately, thereby improving the timeliness and accuracy of fault diagnosis.

[0029] First, at multiple insulation positions of the insulation layer of the converter transformer, a sensor array is configured, including installing high-frequency current sensors or voltage sensors at multiple key insulation positions of the transformer (such as high-voltage areas near the insulation layer) to monitor the electrical signals generated by partial discharge in real time. During partial discharge, the electrical signals will appear as high-frequency pulses, and the intensity, frequency, and duration of these pulses can reflect the location, intensity, and development trend of the discharge; using temperature sensors or infrared thermal imaging technology to monitor the temperature changes at different positions of the insulation layer. Partial discharge is often accompanied by heat release, resulting in a local temperature increase. By monitoring these thermal signals, the damaged areas of the insulation material can be identified; using acoustic sensors (such as ultrasonic sensors) to monitor the acoustic signals generated during partial discharge, especially high-frequency noise. Partial discharge is usually accompanied by acoustic signals in a specific frequency band, and these acoustic signals have obvious enhanced characteristics in the fault occurrence area; using fiber optic sensors or infrared sensors to monitor the optical signals in the insulation layer (such as the light radiation generated by partial discharge). Partial discharge will trigger arc or other discharge phenomena, and these phenomena may generate short-term optical signals that can be detected by optical sensors in a specific wavelength band.

[0030] Furthermore, when the converter transformer is operating, through the sensor array, the electrical, thermal, acoustic, and optical signals at the multiple insulation positions are monitored and collected, and according to the multiple insulation positions, the electrical, thermal, acoustic, and optical signals at the multiple insulation positions are arranged, that is, the signals at each monitoring position are arranged according to the position and function to ensure that each key area can be properly monitored and covered, obtaining an electrical signal array, a thermal signal array, an acoustic signal array, and an optical signal array. Among them, the electrical signal array contains electrical signal data from different insulation positions of the transformer, and these data can reflect the occurrence, intensity, and distribution of partial discharge; the thermal signal array contains temperature data from each position, and the temperature change can indicate the thermal effect area of partial discharge, helping to analyze the aging or damage degree of the insulation material; the acoustic signal array contains high-frequency acoustic wave data collected by acoustic sensors, which can reflect the localization characteristics of the discharge activity and help to locate the partial discharge source; the optical signal array includes the optical pulses generated during the discharge process, which can be used as an additional diagnostic index for partial discharge.

[0031] By configuring sensor arrays at multiple insulation positions of a converter transformer to collect various monitoring signals (electrical signals, thermal signals, acoustic signals, optical signals), it is possible to achieve all-round and real-time monitoring of the transformer insulation status, thereby improving the accuracy and timeliness of fault detection.

[0032] S200: Perform partial discharge scale analysis and partial discharge consistency analysis on the electrical signal array, thermal signal array, acoustic signal array, and optical signal array to obtain scale information and consistency information.

[0033] Furthermore, step S200 of this application further includes:

[0034] S210: Perform signal abnormality analysis on the electrical signal array, thermal signal array, acoustic signal array, and optical signal array to obtain an electrical abnormality array, a thermal abnormality array, an acoustic abnormality array, and an optical abnormality array.

[0035] Furthermore, step S210 of this application further includes:

[0036] S211: Extract a first electrical signal from the electrical signal array; S212: Randomly extract multiple historical electrical signals from the electrical signal monitoring data within a historical time; S213: Calculate the electrical signal difference between the mean of the multiple historical electrical signals and the first electrical signal, and calculate the ratio of the electrical signal difference to the first electrical signal as the first electrical abnormality; S214: Continue to calculate the electrical abnormality of all electrical signals to obtain an electrical abnormality array; S215: Perform signal abnormality analysis based on the thermal signal array, acoustic signal array, and optical signal array to obtain a thermal abnormality array, an acoustic abnormality array, and an optical abnormality array.

[0037] Specifically, first, randomly select an electrical signal at any insulated position within the electrical signal array as the first electrical signal; then, randomly extract multiple historical electrical signals from the electrical signal monitoring data in a historical time period (such as within the most recent week). These historical signals are usually electrical signal data collected during the operation of the transformer over a period of time in the past, reflecting the electrical characteristics under normal or stable operating conditions. The purpose of random extraction is to ensure the diversity of data and avoid bias caused by the selection of a single historical signal. Then calculate the mean value of the multiple historical electrical signals to obtain the historical electrical signal mean value, and further calculate the electrical signal difference between the historical electrical signal mean value and the first electrical signal to obtain the electrical signal difference. Here, the electrical signal difference is the absolute value of the difference between the historical electrical signal mean value and the first electrical signal; then calculate the ratio of the electrical signal difference to the first electrical signal, and take the ratio as the first electrical anomaly degree. Here, the electrical anomaly degree reflects the deviation degree between the current electrical signal and the historical normal state. The larger the value of the electrical anomaly degree, the more the current electrical signal deviates from the normal state, and there may be more serious insulation problems. Then use the same method for calculating the first electrical anomaly degree to continue calculating the electrical anomaly degrees of all electrical signals, and construct an electrical anomaly degree array. Each value in the electrical anomaly degree array represents the anomaly degree of the electrical signal at the corresponding moment. Through comparison and analysis, electrical anomalies that may exist during the operation of the transformer can be effectively identified.

[0038] Using the same method for calculating the electrical anomaly degree array, perform signal anomaly degree analysis based on the thermal signal array, acoustic signal array, and optical signal array respectively to obtain a thermal anomaly degree array, an acoustic anomaly degree array, and an optical anomaly degree array.

[0039] S220: Calculate the proportion of the anomaly degrees greater than the anomaly degree threshold within the electrical anomaly degree array, thermal anomaly degree array, acoustic anomaly degree array, and optical anomaly degree array to obtain four anomaly ratios, and calculate the mean value to obtain the scale information; S230: Perform consistency analysis calculation based on the four anomaly ratios to obtain the consistency information.

[0040] Specifically, first, configure the electrical anomaly threshold, which is used to determine whether the anomaly degree of the current electrical signal exceeds the normal range. This threshold can be set based on historical data, experimental results, or standard specifications. For example, set the electrical anomaly threshold to 30%. Then, calculate the proportion of anomalies greater than the anomaly threshold in the electrical anomaly array, thermal anomaly array, acoustic anomaly array, and optical anomaly array respectively, to obtain the electrical anomaly ratio, thermal anomaly ratio, acoustic anomaly ratio, and optical anomaly ratio. Further, calculate the mean value of the electrical anomaly ratio, thermal anomaly ratio, acoustic anomaly ratio, and optical anomaly ratio to obtain the anomaly ratio mean as the scale information. Among them, the scale information is used to reflect whether there are serious insulation problems or potential faults in the transformer in multiple dimensions. The larger the scale information, the greater the probability of insulation anomalies.

[0041] In a multi-signal monitoring system, due to factors such as sensor errors, data transmission delays, or environmental noise, the calculation of anomaly degrees of different signals may be inconsistent. This inconsistency will affect the accuracy of the analysis results. Therefore, consistency analysis must be carried out to ensure a more accurate and reliable assessment of the transformer status. Then, calculate the consistency based on the electrical anomaly ratio, thermal anomaly ratio, acoustic anomaly ratio, and optical anomaly ratio. For example, calculate the mean value of the four anomaly ratios to obtain the anomaly ratio mean. Then, calculate the absolute value of the deviation between each anomaly ratio and the anomaly ratio mean respectively to obtain four absolute deviation values, and calculate the mean value to obtain the consistency information. For example, assume that the electrical anomaly ratio, thermal anomaly ratio, acoustic anomaly ratio, and optical anomaly ratio are 0.4, 0.5, 0.45, and 0.55 respectively, then the anomaly ratio mean is 0.475, and the four absolute deviation values are 0.075, 0.025, 0.025, and 0.075 respectively, and the consistency information is 0.05. Obtain the consistency information. Among them, the larger the consistency information, the greater the difference in anomaly degrees between different signals, indicating data inconsistency, sensor errors, or other external interferences, which will reduce the accuracy of the analysis results.

[0042] S300: According to the scale information and the consistency information, configure the transformer status analysis path in the transformer status analysis channel to analyze and obtain the insulation influence parameters.

[0043] Furthermore, step S300 of this application further includes:

[0044] S310: According to the consistency information, compensate and adjust the scale information to obtain the adjusted scale information.

[0045] Specifically, the scale information is compensated and adjusted according to the consistency information. The purpose of the compensation and adjustment is to correct the scale information according to the magnitude of the consistency, so as to more accurately reflect the health state of the transformer. Especially in the case of sensor errors or data inconsistency, the scale information can be made more reliable. For example, the adjusted scale information is obtained by multiplying the sum of 1 and the consistency information by the scale information. For instance, assuming the consistency information is 0.05 and the scale information is 0.4, then the adjusted scale information is 0.42. By compensating and adjusting the scale information, the finally obtained adjusted scale information will more precisely reflect the health condition of the transformer. Especially in the face of sensor errors or inconsistent data, this compensation and adjustment method helps to improve the accuracy and reliability of the insulation state monitoring of the converter transformer, and more comprehensively evaluate the operation health condition of the equipment based on the analysis of multi-dimensional signals.

[0046] S320: Based on integrated learning, train the transformer state analysis channel, where the transformer state analysis channel includes M transformer state analysis paths, and M is a positive integer.

[0047] Furthermore, step S320 of the present application further includes:

[0048] S321: According to the partial discharge monitoring data of the converter transformer within the historical time, collect the sample electrical signal array, sample thermal signal array, sample acoustic signal array, sample optical signal array, and test the insulation resistance of the transformer before and after collecting the partial discharge to obtain the sample initial insulation parameter set and the sample insulation influence parameter set; S322: Use the sample electrical signal array, sample thermal signal array, sample acoustic signal array, sample optical signal array, and sample initial insulation parameter set as input data, use the sample insulation influence parameter set as supervision data, and randomly divide the input data and supervision data with replacement to obtain M pieces of supervised training data; S323: Use the M pieces of supervised training data to train M transformer state analysis paths, and after the training is completed, combine them to obtain the transformer state analysis channel.

[0049] Specifically, according to the partial discharge monitoring data of the converter transformer within a historical time period (such as within the most recent week), sample electrical signal arrays, sample thermal signal arrays, sample acoustic signal arrays, and sample optical signal arrays are collected; further, when conducting historical partial discharge monitoring, the insulation resistance of the transformer is tested before and after partial discharge respectively to obtain a sample initial insulation parameter set and a sample insulation influence parameter set; among them, the insulation resistance is an important indicator for measuring the performance of the insulation layer. The sample initial insulation parameter is the insulation resistance of the transformer before partial discharge, reflecting the insulation state of the transformer insulation layer during normal operation; the sample insulation influence parameter is the insulation resistance of the transformer after partial discharge. Partial discharge will cause damage to the insulation layer, resulting in a decrease in the insulation resistance. Therefore, this data can be used as a basis for diagnosing the degree of damage to the insulation layer.

[0050] Then, using the sample electrical signal array set, sample thermal signal array set, sample acoustic signal array set, sample optical signal array set, and sample initial insulation parameter set as input data, and using the sample insulation influence parameter set as supervised data, the input data and the supervised data are used as training data; then the training data is divided into M parts of data, and M selections are made with replacement from the M parts of data to obtain the first supervised training data. Using the same method, M selections are made iteratively to obtain M parts of supervised training data, where M is a positive integer, and the specific value of M can be set according to actual needs, such as setting M to 20.

[0051] Further, M transformer state analysis paths are constructed based on a BP neural network. The transformer state analysis path is a BP neural network model in machine learning that can be iteratively optimized, including an input layer, multiple hidden layers, and an output layer. Among them, the input data of the input layer is an array of electrical signals, an array of thermal signals, an array of acoustic signals, an array of optical signals, and initial insulation parameters, and the output data of the output layer is insulation impact parameters. Then, using the sample electrical signal array, sample thermal signal array, sample acoustic signal array, sample optical signal array, and sample initial insulation parameters as inputs, and the sample insulation impact parameters as supervision, the M supervision training data are used to separately supervise and train the M transformer state analysis paths. First, in each training iteration, the input layer transmits electrical signals, thermal signals, acoustic signals, optical signals, and initial insulation parameters into the network, and the final data is transmitted to the output layer, and the network generates a predicted value, that is, the insulation impact parameter. Then, the output value of the neural network is compared with the true insulation impact parameter, the error (such as the mean square error, MSE) is calculated, and according to the calculated error, the error is propagated back layer by layer from the output layer to the input layer through the backpropagation algorithm. By calculating the gradient layer by layer, the weights and biases of each layer are adjusted to gradually reduce the error. The main goal of backpropagation is to optimize the weights of the network through the gradient descent method to ensure that the network can generate an output as close as possible to the true value. Further, the gradient descent algorithm is used to adjust the weights and biases in the network. In each iteration, the weights and biases are adjusted according to the gradient of the error until the error reaches the minimum value, and the M trained transformer state analysis paths are obtained.

[0052] Finally, the M trained transformer state analysis paths are combined to obtain a transformer state analysis channel. By constructing a transformer state analysis channel based on a BP neural network, the intelligence, efficiency, and accuracy of the analysis of the insulation impact parameters of the transformer can be improved, thereby significantly enhancing the intelligence level of transformer health monitoring and improving the accuracy and efficiency of fault detection.

[0053] S330: Use the adjusted scale information and M to calculate and round to obtain N, where N is a positive integer less than M; S340: Randomly select N transformer state analysis paths, input the initial insulation parameters and the electrical signal array, thermal signal array, acoustic signal array, and optical signal array, identify and obtain N path insulation impact parameters, and calculate the mean value to obtain the insulation impact parameter.

[0054] Specifically, multiply the adjustment scale information by M and round down to obtain N, where N is a positive integer less than M. Then, obtain the insulation resistance of the transformer before multi-dimensional signal acquisition through the sensor array, which is set as the initial insulation parameter. Next, randomly select N transformer state analysis paths from the M transformer state analysis paths in the transformer state analysis channel, and input the initial insulation parameter, electrical signal array, thermal signal array, acoustic signal array, and optical signal array into the N transformer state analysis paths to output N path insulation influence parameters. Further, calculate the mean value of the N path insulation influence parameters to obtain the insulation influence parameter. By dynamically adjusting the number of paths according to the adjustment scale information, the computing resources can be reasonably allocated according to the adjustment scale information, reducing unnecessary computing overhead while maintaining the analysis accuracy and improving the operation efficiency of the system. By randomly selecting paths and calculating their mean values, the computing loads of different paths can be effectively balanced, optimizing resource allocation to ensure efficient and accurate transformer state assessment under limited computing resources.

[0055] S400: Detect the insulation influence parameter of the commutation transformer to obtain the actual insulation influence parameter, and update the transformer state analysis channel.

[0056] Furthermore, step S400 of the present application further includes:

[0057] S410: Detect the insulation influence parameter of the commutation transformer to obtain the actual insulation influence parameter; S420: Calculate the insulation influence parameter deviation between the actual insulation influence parameter and the insulation influence parameter, and determine whether it is greater than the insulation influence deviation threshold; S430: If not, do not update the transformer state analysis path and the transformer state analysis channel; S440: If so, calculate the ratio of the insulation influence deviation threshold to the insulation influence parameter deviation to obtain the error scale; S450: Multiply the error scale by M and round down to obtain O, where O is a positive integer less than M; S460: Use the electrical signal array, thermal signal array, acoustic signal array, and optical signal array, combined with the current initial insulation parameter and actual insulation influence parameter of the commutation transformer, as updated training data, and randomly select O transformer state analysis paths in the transformer state analysis channel for updated training.

[0058] Specifically, the insulation impact parameters of the commutation transformer are monitored in real time to obtain the actual insulation impact parameters, which represent the true data of the current insulation state of the transformer. Then, the insulation impact parameter deviation between the actual insulation impact parameters and the insulation impact parameters is calculated. The insulation impact parameter deviation is the absolute value of the difference between the actual insulation impact parameters and the insulation impact parameters. An insulation impact deviation threshold is configured, which is used to determine whether the deviation exceeds the acceptable error range and can be set according to the prediction accuracy requirements. The higher the prediction accuracy requirements, the smaller the insulation impact deviation threshold. Further, it is determined whether the insulation impact parameter deviation is greater than the insulation impact deviation threshold. If not, it means that the prediction result is relatively consistent with the actual situation, that is, the prediction accuracy is high, and the transformer state analysis path and the transformer state analysis channel are not updated; if so, the ratio of the insulation impact parameter deviation to the insulation impact deviation threshold is calculated to obtain the error scale. Among them, the larger the error scale, the more significant the error of the prediction result and the worse the prediction accuracy. Then, the error scale is multiplied by M and rounded to obtain O, where O is a positive integer less than M.

[0059] Finally, the electrical signal array, the thermal signal array, the acoustic signal array, and the optical signal array, combined with the current initial insulation parameters and the actual insulation impact parameters of the commutation transformer, are used as updated training data, and the updated training data is used to randomly select O transformer state analysis paths in the transformer state analysis channel for updated training to optimize the prediction ability of the transformer state analysis channel. By dynamically adjusting the number of updated training of the transformer state analysis path based on the error scale, the balance between accuracy and efficiency can be achieved. When the prediction error is large, more paths are selected for training to improve the analysis accuracy; when the error is small, the number of updated paths is reduced, thereby saving computing resources. This method can effectively improve the efficiency of the analysis process while ensuring accuracy and meet the health monitoring requirements of complex power equipment.

[0060] In summary, the state detection method for a commutation transformer based on PD monitoring provided by this application has the following technical effects:

[0061] By monitoring the electrical, thermal, acoustic, and optical signals of partial discharge in the converter transformer, an electrical signal array, a thermal signal array, an acoustic signal array, and an optical signal array are obtained; then, the partial discharge scale analysis and partial discharge consistency analysis are performed on the electrical signal array, the thermal signal array, the acoustic signal array, and the optical signal array to obtain scale information and consistency information; then, according to the scale information and consistency information, the transformer state analysis path in the transformer state analysis channel is configured, and the insulation influence parameter is analyzed and obtained; further, the insulation influence parameter detection is performed on the converter transformer to obtain the actual insulation influence parameter; finally, the parameter deviation between the actual insulation influence parameter and the insulation influence parameter is analyzed, and the transformer state analysis channel is optimized and updated according to the insulation influence parameter deviation. That is to say, by comprehensively analyzing the transformer through multi-signal monitoring, the health status of the transformer insulation layer can be evaluated more comprehensively and accurately, significantly improving the comprehensiveness and accuracy of insulation status detection, so that potential insulation faults can be detected in time and the probability of fault occurrence can be reduced.

[0062] Embodiment 2. Based on a converter transformer state detection method based on PD monitoring in the foregoing embodiment and with the same inventive concept, the present application also provides a converter transformer state detection platform based on PD monitoring. Please refer to the attached Figure 2 , including: a partial discharge signal monitoring module 11, which is used to monitor the electrical, thermal, acoustic, and optical signals of partial discharge in the converter transformer to obtain an electrical signal array, a thermal signal array, an acoustic signal array, and an optical signal array; a partial discharge feature analysis module 12, which is used to perform partial discharge scale analysis and partial discharge consistency analysis on the electrical signal array, the thermal signal array, the acoustic signal array, and the optical signal array to obtain scale information and consistency information; an insulation influence parameter analysis module 13, which is used to configure the transformer state analysis path in the transformer state analysis channel according to the scale information and consistency information, and analyze and obtain the insulation influence parameter; a state analysis channel update module 14, which is used to perform insulation influence parameter detection on the converter transformer to obtain the actual insulation influence parameter and update the transformer state analysis channel.

[0063] Further, the converter transformer state detection platform based on PD monitoring is also used for: configuring a sensor array at multiple insulation positions of the converter transformer insulation layer, monitoring and collecting the electrical, thermal, acoustic, and optical signals of the multiple insulation positions; arranging the electrical, thermal, acoustic, and optical signals of the multiple insulation positions according to the multiple insulation positions to obtain an electrical signal array, a thermal signal array, an acoustic signal array, and an optical signal array.

[0064] Further, the converter transformer condition detection platform based on PD monitoring is also used for: analyzing the signal abnormality degree of the electrical signal array, thermal signal array, acoustic signal array, and optical signal array to obtain an electrical abnormality degree array, thermal abnormality degree array, acoustic abnormality degree array, and optical abnormality degree array; calculating the proportion of the abnormality degrees greater than the abnormality threshold in the electrical abnormality degree array, thermal abnormality degree array, acoustic abnormality degree array, and optical abnormality degree array to obtain four abnormality ratios, calculating the mean value, and obtaining the scale information; performing consistency analysis and calculation according to the four abnormality ratios to obtain the consistency information.

[0065] Further, the converter transformer condition detection platform based on PD monitoring is also used for: extracting a first electrical signal from the electrical signal array; randomly extracting multiple historical electrical signals from the electrical signal monitoring data within the historical time; calculating the electrical signal difference between the mean value of the multiple historical electrical signals and the first electrical signal, and calculating the ratio of the electrical signal difference to the first electrical signal as the first electrical abnormality degree; continuing to calculate the electrical abnormality degrees of all electrical signals to obtain the electrical abnormality degree array; performing signal abnormality degree analysis according to the thermal signal array, acoustic signal array, and optical signal array to obtain the thermal abnormality degree array, acoustic abnormality degree array, and optical abnormality degree array.

[0066] Further, the converter transformer condition detection platform based on PD monitoring is also used for: compensating and adjusting the scale information according to the consistency information to obtain the adjusted scale information; training the transformer condition analysis channel based on ensemble learning, where the transformer condition analysis channel includes M transformer condition analysis paths, and M is a positive integer; calculating and rounding to obtain N using the adjusted scale information and M, where N is a positive integer less than M; randomly selecting N transformer condition analysis paths, inputting the initial insulation parameters and the electrical signal array, thermal signal array, acoustic signal array, and optical signal array, identifying N path insulation influence parameters, and calculating the mean value to obtain the insulation influence parameters.

[0067] Further, the converter transformer condition detection platform based on PD monitoring is also used for: collecting a sample electrical signal array, sample thermal signal array, sample acoustic signal array, and sample optical signal array according to the converter transformer partial discharge monitoring data within the historical time, and testing the insulation resistance of the transformer before and after local discharge to obtain a sample initial insulation parameter set and a sample insulation influence parameter set; using the sample electrical signal array, sample thermal signal array, sample acoustic signal array, sample optical signal array, and sample initial insulation parameter set as input data, using the sample insulation influence parameter set as supervised data, and randomly dividing the input data and supervised data with replacement to obtain M pieces of supervised training data; using the M pieces of supervised training data to train M transformer condition analysis paths, and combining them to obtain the transformer condition analysis channel after training.

[0068] Further, the converter transformer status detection platform based on PD monitoring is also used for: detecting insulation influence parameters of the converter transformer to obtain actual insulation influence parameters; calculating an insulation influence parameter deviation between the actual insulation influence parameters and the insulation influence parameters, and determining whether it is greater than an insulation influence deviation threshold; if not, the converter transformer status analysis path and the converter transformer status analysis channel are not updated; if so, calculating a ratio of the insulation influence deviation threshold to the insulation influence parameter deviation to obtain an error scale; multiplying the error scale by M and taking the integer to obtain O, where O is a positive integer less than M; using the electrical signal array, the thermal signal array, the acoustic signal array, the optical signal array, in combination with the current initial insulation parameters and actual insulation influence parameters of the converter transformer, as updated training data, and randomly selecting O converter transformer status analysis paths in the converter transformer status analysis channel for updated training.

[0069] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The method and specific examples of a converter transformer status detection method based on PD monitoring in the foregoing Embodiment 1 are equally applicable to a converter transformer status detection platform based on PD monitoring in this embodiment. Through the foregoing detailed description of a converter transformer status detection method based on PD monitoring, those skilled in the art can clearly know a converter transformer status detection platform based on PD monitoring in this embodiment. Therefore, for the sake of brevity of the specification, it will not be elaborated here. For the platform disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0070] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0071] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is also intended to include these changes and variations.

Claims

1. A converter transformer state detection method based on PD monitoring, characterized in that: Methods include: Monitor the electrical signal, thermal signal, acoustic signal and optical signal of partial discharge of the converter transformer to obtain an electrical signal array, a thermal signal array, an acoustic signal array and an optical signal array; Performing partial discharge scale analysis and partial discharge consistency analysis on the electrical signal array, the thermal signal array, the acoustic signal array, and the optical signal array to obtain scale information and consistency information includes: Performing signal anomaly analysis on the electrical signal array, the thermal signal array, the acoustic signal array, and the optical signal array to obtain an electrical anomaly array, a thermal anomaly array, an acoustic anomaly array, and an optical anomaly array; Calculate the proportion of abnormalities greater than the abnormality threshold in the electrical abnormality array, the thermal abnormality array, the acoustic abnormality array, and the optical abnormality array to obtain four abnormality proportions, calculate the average, and obtain scale information; According to the four abnormal proportions, consistency analysis and calculation are performed to obtain consistency information; According to the scale information and consistency information, a transformer state analysis path in the transformer state analysis channel is configured to analyze and obtain insulation influence parameters, including: According to the consistency information, compensating and adjusting the scale information to obtain adjusted scale information; Based on ensemble learning, a transformer state analysis channel is trained, wherein the transformer state analysis channel includes M transformer state analysis paths, where M is a positive integer; Using the adjustment scale information and M, calculate and round to obtain N, where N is a positive integer less than M; Randomly select N transformer state analysis paths, input the initial insulation parameters and the electrical signal array, the thermal signal array, the acoustic signal array, and the optical signal array, identify and obtain the insulation influencing parameters of the N paths, calculate the mean, and obtain the insulation influencing parameters; Insulation influencing parameters are detected on the converter transformer to obtain actual insulation influencing parameters, and the transformer status analysis channel is updated.

2. The converter transformer state detection method based on PD monitoring according to claim 1 is characterized in that: Monitor the electrical signal, thermal signal, acoustic signal and optical signal of partial discharge of the converter transformer to obtain an electrical signal array, a thermal signal array, an acoustic signal array and an optical signal array, including: At multiple insulation positions of the insulation layer of the converter transformer, a sensor array is configured to monitor and collect electrical signals, thermal signals, acoustic signals, and optical signals at the multiple insulation positions; According to the multiple insulation positions, the electrical signals, thermal signals, acoustic signals, and optical signals of the multiple insulation positions are arranged to obtain an electrical signal array, a thermal signal array, an acoustic signal array, and an optical signal array.

3. The converter transformer state detection method based on PD monitoring according to claim 1 is characterized in that: Performing signal anomaly analysis on the electrical signal array, the thermal signal array, the acoustic signal array, and the optical signal array includes: Extracting a first electrical signal in the electrical signal array; Randomly extract multiple historical electrical signals from the electrical signal monitoring data within the historical time; Calculating an electrical signal difference between an average of the plurality of historical electrical signals and the first electrical signal, and calculating a ratio of the electrical signal difference to the first electrical signal as a first electrical abnormality; Continue to calculate the electrical anomaly of all electrical signals to obtain an electrical anomaly array; According to the thermal signal array, acoustic signal array, and optical signal array, a signal anomaly analysis is performed to obtain a thermal anomaly array, an acoustic anomaly array, and an optical anomaly array.

4. The converter transformer state detection method based on PD monitoring according to claim 1 is characterized in that: Based on ensemble learning, the transformer state analysis channel is trained, including: According to the partial discharge monitoring data of the converter transformer in the historical time, the sample electrical signal array, the sample thermal signal array, the sample acoustic signal array, and the sample optical signal array are collected, and the insulation resistance of the transformer before and after the partial discharge is tested and collected to obtain the sample initial insulation parameter set and the sample insulation influencing parameter set; Using the sample electrical signal array, the sample thermal signal array, the sample acoustic signal array, the sample optical signal array, and the sample initial insulation parameter set as input data, using the sample insulation influence parameter set as supervision data, and performing random division with replacement on the input data and the supervision data to obtain M pieces of supervision training data; The M pieces of supervised training data are used to train M transformer state analysis paths, and after the training is completed, the transformer state analysis channels are obtained by combination.

5. The converter transformer state detection method based on PD monitoring according to claim 4 is characterized in that: Performing insulation influence parameter detection on the converter transformer to obtain actual insulation influence parameters and updating the transformer state analysis channel includes: Performing insulation influence parameter detection on the converter transformer to obtain actual insulation influence parameters; Calculating an insulation influence parameter deviation between the actual insulation influence parameter and the insulation influence parameter, and determining whether the deviation is greater than an insulation influence deviation threshold; If not, the transformer status analysis path and transformer status analysis channel are not updated; If yes, then calculate the ratio of the insulation influence deviation threshold to the insulation influence parameter deviation to obtain the error scale; Multiply the error scale by M and round it to obtain O, where O is a positive integer less than M; The electrical signal array, thermal signal array, acoustic signal array, and optical signal array are combined with the current initial insulation parameters and actual insulation influencing parameters of the converter transformer as update training data, and O transformer state analysis paths are randomly selected in the transformer state analysis channel for update training.

6. The converter transformer status detection platform based on PD monitoring is characterized by: The steps for implementing the converter transformer state detection method based on PD monitoring according to any one of claims 1 to 5 include: A partial discharge signal monitoring module is used to monitor the electrical signal, thermal signal, acoustic signal and optical signal of partial discharge of the converter transformer, and obtain an electrical signal array, a thermal signal array, an acoustic signal array and an optical signal array; A partial discharge characteristic analysis module is used to perform partial discharge scale analysis and partial discharge consistency analysis on the electrical signal array, thermal signal array, acoustic signal array, and optical signal array to obtain scale information and consistency information; An insulation influence parameter analysis module, configured to configure a transformer state analysis path in a transformer state analysis channel according to the scale information and consistency information, and to analyze and obtain insulation influence parameters; The status analysis channel update module is used to detect the insulation influence parameters of the converter transformer, obtain the actual insulation influence parameters, and update the transformer status analysis channel.

Images