Online monitoring and fault diagnosis method for transformer on-load tap-changer

Abstract

The present invention provides a method for online monitoring and fault diagnosis of a transformer on-load tapchanger, relating to the field of electrical technology. The method comprises: obtaining arc energy and contact position information generated during a switching operation of the transformer on-load tapchanger; determining a switching performance score; obtaining operating data of the transformer on-load tapchanger; inputting the operating data into a trained fault probability prediction model to obtain predicted fault probability data during the switching operation of the transformer on-load tapchanger; determining whether the transformer on-load tapchanger meets preset target conditions; if the transformer on-load tapchanger meets the preset target conditions, determining that the transformer on-load tapchanger is in a normal state; if the transformer on-load tapchanger does not meet the preset target conditions, determining that the transformer on-load tapchanger is in a faulty state. According to the present invention, the on-load tapchanger state can be comprehensively monitored based on the switching performance and predicted fault probability.

Description

Technical Field

[0001] The present invention relates to the field of electrical technology, and in particular to an online monitoring and fault diagnosis method for a transformer on-load tap changer. Background Art

[0002] In related technologies, CN113267724A relates to an online monitoring system for on-load tap-changers of transformers, which includes: multiple signal acquisition devices and processing devices; multiple signal acquisition devices are set in a one-to-one correspondence with multiple main contacts of the on-load tap-changer, and multiple signal acquisition devices are used to correspondingly acquire optical signals generated by corresponding main contacts when performing disconnection operations and convert the optical signals into electrical signals; a processing device is used to receive multiple electrical signals sent by the multiple signal acquisition devices and output an early warning message when the amplitude of at least one of the multiple electrical signals is greater than a preset amplitude threshold. The system monitoring results provided by this solution are more accurate; and the system can monitor the on-load tap-changer without shutting down the transformer, without affecting the normal operation of the transformer, and can monitor the on-load tap-changer in real time, thereby improving the monitoring sensitivity of the on-load tap-changer and effectively monitoring faults of the on-load tap-changer.

[0003] CN113960463A discloses a method for online monitoring and fault diagnosis of an on-load tapchanger of a voltage-regulating transformer. The method includes: collecting the on-load tapchanger's vibration-acoustic fingerprint signal and the drive motor current signal; performing data analysis on the vibration-acoustic fingerprint signal and the drive motor current signal to obtain a signal analysis result; extracting characteristic parameters for on-load tapchanger fault diagnosis based on the signal analysis result; and combining the characteristic parameters with fault data, factory data, and historical data to obtain an on-load tapchanger fault diagnosis result. This solution can monitor the status and diagnose faults of the on-load tapchanger of a voltage-regulating transformer based on the vibration-acoustic fingerprint and the drive motor current in a live state. This method does not affect the normal operation of the transformer and has no electrical connection to the equipment, while offering advantages such as safety and high reliability.

[0004] Therefore, although the related art can realize the status monitoring of the transformer on-load tap changer, the related art does not consider the impact of switching performance and predicted failure probability on the fault diagnosis results, that is, it is impossible to comprehensively monitor the status of the on-load tap changer based on the switching performance and predicted failure probability. Summary of the Invention

[0005] The present invention provides a method for online monitoring and fault diagnosis of a transformer on-load tap changer, which can solve the technical problem that related technologies cannot comprehensively monitor the status of the on-load tap changer based on switching performance and predicted fault probability.

[0006] According to the present invention, a method for online monitoring and fault diagnosis of a transformer on-load tap changer is provided, comprising:

[0007] Obtain the arc energy and contact position information of the on-load tap changer generated during the transformer on-load tap changer switching action in the current monitoring period;

[0008] determining a switching performance score according to the arc energy and the contact position information of the on-load tap changer;

[0009] Acquiring operating data of the transformer on-load tap changer during a current monitoring cycle, wherein the operating data includes a vibration acoustic signal, a motor current signal, and a contact temperature generated during a switching operation of the transformer on-load tap changer;

[0010] Inputting the operating data into a trained fault probability prediction model to obtain predicted fault probability data of the transformer on-load tap changer switching action in the current monitoring period;

[0011] determining whether the transformer on-load tap changer meets a preset target condition based on the predicted fault probability data and the switching performance score;

[0012] If the transformer on-load tap changer meets the preset target condition, determining that the transformer on-load tap changer is in a normal state;

[0013] If the transformer on-load tap changer does not meet the preset target condition, it is determined that the transformer on-load tap changer is in a fault state.

[0014] Furthermore, determining a switching performance score according to the arc energy and the contact position information of the on-load tap changer includes:

[0015] Set the normal range of arc energy;

[0016] Obtaining a voltage value corresponding to the on-load tap changer contact position information;

[0017] The voltage value corresponding to the target tap position of the transformer on-load tap changer is set to the voltage value corresponding to the target position information;

[0018] A switching performance score is determined according to the arc energy, the voltage value corresponding to the on-load tap changer contact position information, the normal range of the arc energy, and the voltage value corresponding to the target position information.

[0019] Furthermore, determining a switching performance score based on the arc energy, the voltage value corresponding to the on-load tap changer contact position information, the normal range of the arc energy, and the voltage value corresponding to the target position information includes:

[0020] According to the formula

[0021] ,

[0022] Determining the handover performance score ,in, is the arc energy generated by the transformer on-load tap-changer during the i-th switching action in the current monitoring period, is the minimum arc energy, is the maximum arc energy, is the voltage value corresponding to the contact position information of the on-load tapchanger during the i-th switching action of the transformer on-load tapchanger in the current monitoring period, is the voltage value corresponding to the target position information during the i-th switching action of the transformer on-load tap-changer in the current monitoring period, N is the number of switching actions of the transformer on-load tap-changer in the monitoring period, i≤N, and both i and N are positive integers, and if is a conditional function.

[0023] Furthermore, the training step of the fault probability prediction model includes:

[0024] Acquiring historical operating data of the transformer on-load tap-changer during a historical monitoring period, wherein the historical operating data includes historical vibration and acoustic signals, historical motor current signals, and historical contact temperatures generated during a switching action of the transformer on-load tap-changer;

[0025] Obtain historical fault probability data of transformer on-load tap-changer switching actions during multiple historical monitoring periods;

[0026] The historical operation data is processed by a fault probability prediction model to obtain historical predicted fault probability data of the transformer on-load tap changer switching action during multiple historical monitoring periods;

[0027] Determining a loss function of the failure probability prediction model based on the historical operating data, the historical failure probability data, and the historical predicted failure probability data;

[0028] The fault probability prediction model is trained according to the loss function of the fault probability prediction model to obtain the trained fault probability prediction model.

[0029] Furthermore, determining a loss function of the failure probability prediction model based on the historical operating data, the historical failure probability data, and the historical predicted failure probability data includes:

[0030] According to the formula

[0031] ,

[0032] Determine the loss function of the failure probability prediction model ,in, is the historical fault probability data of the transformer on-load tap changer during the i-th switching action in the h-th historical monitoring period, is the historical predicted fault probability data of the transformer on-load tap changer at the i-th switching action in the h-th historical monitoring period, is the historical vibration acoustic signal of the transformer on-load tap-changer during the i-th switching action in the h-th historical monitoring cycle, It is the normal vibration acoustic signal of the transformer on-load tap changer during switching action. is the historical motor current signal of the transformer on-load tap-changer during the i-th switching action in the h-th historical monitoring cycle, It is the normal motor current signal when the transformer on-load tap changer is switching. is the historical contact temperature of the transformer on-load tap-changer during the i-th switching action in the h-th historical monitoring cycle, It is the normal contact temperature of the transformer on-load tap changer during switching action. is the number of historical monitoring cycles of the sth training batch, S is the number of training batches, N is the number of transformer on-load tap-changer switching actions in the monitoring cycle, i≤N, h≤ , s≤S, and i, h, s, N, and S are both positive integers.

[0033] Furthermore, determining whether the transformer on-load tap changer meets a preset target condition based on the predicted fault probability data and the switching performance score includes:

[0034] Set the failure probability threshold and switching performance score threshold;

[0035] It is determined whether the transformer on-load tap changer meets a preset target condition according to the predicted fault probability data, the switching performance score, the fault probability threshold, and the switching performance score threshold.

[0036] Furthermore, determining whether the transformer on-load tap changer meets a preset target condition based on the predicted fault probability data, the switching performance score, the fault probability threshold, and the switching performance score threshold includes:

[0037] According to the formula

[0038] ,

[0039] ,

[0040] Determine the first condition C1 and the second condition C2, where: is the predicted fault probability data of the transformer on-load tap changer during the i-th switching action in the current monitoring period, is the fault probability threshold, F is the piecewise function that compares the predicted fault probability data of the transformer on-load tap changer at the i-th switching action in the current monitoring period with the fault probability threshold, is the preset average failure probability data within a monitoring period, Score the switching performance, is the switching performance scoring threshold, N is the number of switching actions of the transformer on-load tap changer in the monitoring period, i≤N, and both i and N are positive integers, the first condition C1 and the second condition C2 are the preset target conditions;

[0041] When the first condition C1 and the second condition C2 are satisfied at the same time, it is determined that the transformer on-load tap changer meets the preset target condition.

[0042] Technical Effect: According to the present invention, the switching performance of the on-load tap changer is comprehensively monitored through arc energy and contact position information, thereby evaluating the status of the on-load tap changer. By analyzing the operating data through the trained fault probability prediction model, the failure probability of the transformer on-load tap changer can be predicted, which helps the normal operation of the transformer on-load tap changer and timely warning when a fault occurs, reduces the risk of failure, and improves the reliability and stability of the transformer on-load tap changer. When determining the switching performance score, the switching performance score can be determined by the arc energy, the voltage value corresponding to the on-load tap changer contact position information, the normal range of the arc energy, and the voltage value corresponding to the target position information. The two parameters of the arc energy and the voltage value corresponding to the contact position information respectively reflect the electrical performance and mechanical performance of the switch during the switching process, thereby improving the comprehensiveness and reliability of the switching performance score. When determining the loss function of the fault probability prediction model, the influence of the historical contact temperature on the fault probability can be used to determine the influence of the above data on the error of the historical predicted fault probability data. Based on the influence and the relative error between the historical fault probability data and the historical predicted fault probability data, and based on the conditions of the vibration acoustic signal and the historical motor current signal during the switching action of the transformer on-load tapchanger, the weight is set based on the characteristic that the more similar the conditions are to the normal vibration acoustic signal and the normal motor current signal, the greater the reference value, and the shorter the time interval with the first training batch, the lower the accuracy. In this way, the errors output by the fault probability prediction model during multiple switching actions of the transformer on-load tapchanger in each historical monitoring cycle of the sth batch are weightedly summed to obtain the loss function, so as to improve the design accuracy and objectivity of the loss function, thereby improving the training efficiency during the training process and improving the accuracy of the fault probability prediction model. When determining whether the transformer on-load tap-changer meets the preset target conditions, the first condition and the second condition can be determined based on the predicted fault probability data, the switching performance score, the fault probability threshold and the switching performance score threshold. When the first condition and the second condition are simultaneously met, it can be determined that the transformer on-load tap-changer meets the preset target conditions. The predicted fault probability data and the switching performance score are quantified, which improves the accuracy of judging the preset target conditions and realizes a comprehensive evaluation of the status of the transformer on-load tap-changer. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 The following is a flow chart showing, by way of example, a method for online monitoring and fault diagnosis of an on-load tap changer of a transformer according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following specific embodiments are used to describe the technical solution of the present invention in detail. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0045] Figure 1 A flowchart of a method for online monitoring and fault diagnosis of a transformer on-load tap changer according to an embodiment of the present invention is exemplarily shown. The method includes:

[0046] Step S101, obtaining arc energy and contact position information of the on-load tap changer generated during the on-load tap changer switching action of the transformer in the current monitoring period;

[0047] Step S102, determining a switching performance score according to the arc energy and the contact position information of the on-load tap changer;

[0048] Step S103, in the current monitoring cycle, obtaining operating data of the transformer on-load tap changer, wherein the operating data includes a vibration acoustic signal, a motor current signal, and a contact temperature generated during a switching operation of the transformer on-load tap changer;

[0049] Step S104: inputting the operating data into the trained fault probability prediction model to obtain predicted fault probability data of the transformer on-load tap changer switching action during the current monitoring period;

[0050] Step S105, determining whether the transformer on-load tap changer meets a preset target condition based on the predicted fault probability data and the switching performance score;

[0051] Step S106: if the transformer on-load tap changer meets the preset target condition, determining that the transformer on-load tap changer is in a normal state;

[0052] Step S107: If the transformer on-load tap changer does not meet the preset target condition, it is determined that the transformer on-load tap changer is in a fault state.

[0053] According to the transformer on-load tap changer online monitoring and fault diagnosis method of the embodiment of the present invention, the switching performance of the on-load tap changer is comprehensively monitored through arc energy and contact position information, thereby evaluating the status of the on-load tap changer. By analyzing the operating data through a trained fault probability prediction model, the failure probability of the transformer on-load tap changer can be predicted, which facilitates the normal operation of the transformer on-load tap changer and provides timely warnings when faults occur, reduces the risk of faults, and improves the reliability and stability of the transformer on-load tap changer.

[0054] According to one embodiment of the present invention, in step S101, each monitoring period can be set to half an hour, one hour, or the like, and the present invention is not limited thereto. Arc energy is the energy generated by arc discharge between the contacts of the on-load tap changer during switching. The on-load tap changer can be equipped with a position sensor (e.g., an encoder, potentiometer, or photoelectric sensor) to provide real-time feedback on the position of the on-load tap changer contacts. Connecting a voltage sensor and a current sensor to the on-load tap changer of the transformer can accurately record the arc voltage and arc current during arc discharge, and then integrate the product of the arc current and arc voltage to calculate the arc energy.

[0055] According to an embodiment of the present invention, in step S102 , a switching performance score is determined according to the arc energy and the contact position information of the on-load tap changer.

[0056] According to one embodiment of the present invention, step S102 includes: setting a normal range of arc energy; obtaining a voltage value corresponding to the on-load tap changer contact position information; setting a voltage value corresponding to a target tap position of the on-load tap changer of the transformer as a voltage value corresponding to the target position information; and determining a switching performance score based on the arc energy, the voltage value corresponding to the on-load tap changer contact position information, the normal range of arc energy, and the voltage value corresponding to the target position information.

[0057] According to one embodiment of the present invention, a normal range of arc energy is set. For example, for small or low-voltage switches, the normal range of transformer on-load tap-changer arc energy is 0 to 100 J. For large or high-voltage switches, the normal range of transformer on-load tap-changer arc energy is 0 to 200 J. While obtaining on-load tap-changer contact position information, the position sensor can also output a voltage value corresponding to the on-load tap-changer contact position information. For example, a potentiometer outputs a voltage signal corresponding to the position. If the position signal voltage range is 0V to 10V, the transformer on-load tap-changer tap positions are numbered from 1 to 10, and the corresponding on-load tap-changer contact position information is 7.5, then the measured output voltage (the voltage value corresponding to the on-load tap-changer contact position information) is 7.5V. Set the voltage value corresponding to the target tap position of the transformer on-load tapchanger to the voltage value corresponding to the target position information. For example, if the target tap position of the transformer on-load tapchanger is 5, the voltage value corresponding to the target position information is 5V. If the measured output voltage (the voltage value corresponding to the on-load tapchanger contact position information) is 4.8V, a position deviation will occur.

[0058] According to one embodiment of the present invention, determining a switching performance score based on the arc energy, the voltage value corresponding to the on-load tap changer contact position information, the normal range of the arc energy, and the voltage value corresponding to the target position information includes: determining the switching performance score according to formula (1): ,

[0059] (1),

[0060] in, is the arc energy generated by the transformer on-load tap-changer during the i-th switching action in the current monitoring period, is the minimum arc energy, is the maximum arc energy, is the voltage value corresponding to the contact position information of the on-load tapchanger during the i-th switching action of the transformer on-load tapchanger in the current monitoring period, is the voltage value corresponding to the target position information during the i-th switching action of the transformer on-load tap-changer in the current monitoring period, N is the number of switching actions of the transformer on-load tap-changer in the monitoring period, i≤N, and both i and N are positive integers, and if is a conditional function.

[0061] According to one embodiment of the present invention, in formula (1), When the arc energy generated during the i-th switching operation of the transformer on-load tap-changer in the current monitoring period is within the normal range of arc energy, the conditional function value is 1; otherwise, the conditional function value is 0. When the arc energy exceeds the normal range of arc energy, the contact wear is severe and the switching performance of the transformer on-load tap-changer is poor. It means that the conditional functions corresponding to the arc energy generated in each switching action of the transformer on-load tap-changer in the current monitoring period are averaged, and the ratio of switching actions in which the arc energy is within the normal range of arc energy is obtained. The larger the ratio, the more normal the arc energy values are, and the better the switching performance of the transformer on-load tap-changer. is the relative difference between the voltage value corresponding to the on-load tapchanger contact position information during the i-th switching action of the on-load tapchanger in the current monitoring period and the voltage value corresponding to the target position information during the i-th switching action of the on-load tapchanger in the current monitoring period. =1 minus the average value of the relative difference. The larger the average value, the smaller the deviation between the measured position (on-load tapchanger contact position information) and the target position information, and the better the transformer on-load tapchanger switching performance. and Multiplying them together, we can get the switching performance score. The larger the switching performance score is, the better the switching performance of the transformer on-load tap changer is.

[0062] In this way, the switching performance score can be determined based on the arc energy, the voltage value corresponding to the on-load tap changer contact position information, the normal range of arc energy, and the voltage value corresponding to the target position information. The two parameters, arc energy and the voltage value corresponding to the contact position information, respectively reflect the electrical performance and mechanical performance of the switch during the switching process, thereby improving the comprehensiveness and reliability of the switching performance score.

[0063] According to one embodiment of the present invention, in step S103, an acceleration sensor is mounted on the side outer wall of the transformer's on-load tap changer, 15-20 cm below the top and parallel to the vertical transmission rod. The vibration acoustic signal is the mechanical vibration acoustic signal generated by the transformer's on-load tap changer during switching, which can reflect the switch's mechanical performance and operating status. The motor current signal is the operating current of the transformer's on-load tap changer during switching. By monitoring changes in the motor current through a current transformer, the motor's operating status and load condition can be understood. A temperature sensor is used to collect contact temperature. Changes in contact temperature can reflect the contact contact condition and thermal load status, which is important for evaluating the electrical performance and safety of the on-load tap changer.

[0064] According to one embodiment of the present invention, in step S104, the operating data is input into a trained fault probability prediction model. The fault probability prediction model can be a neural network model. Through machine learning or deep learning technology, it is trained based on a large amount of sample data to obtain the predicted fault probability data of the transformer on-load tap changer switching action in the current monitoring period.

[0065] According to one embodiment of the present invention, the above-mentioned fault probability prediction model can be trained before use, and the training steps of the fault probability prediction model include: obtaining historical operation data of the transformer on-load tap changer in a historical monitoring period, wherein the historical operation data includes historical vibration acoustic signals, historical motor current signals and historical contact temperatures generated when the transformer on-load tap changer is switched; obtaining historical fault probability data of the transformer on-load tap changer during multiple historical monitoring periods; processing the historical operation data through the fault probability prediction model to obtain historical predicted fault probability data of the transformer on-load tap changer during multiple historical monitoring periods; determining the loss function of the fault probability prediction model based on the historical operation data, the historical fault probability data and the historical predicted fault probability data; and training the fault probability prediction model based on the loss function of the fault probability prediction model to obtain the trained fault probability prediction model.

[0066] According to one embodiment of the present invention, the transformer on-load tap changer switching actions of multiple historical monitoring cycles are divided into different training batches, and the number of historical monitoring cycles in each training batch is the same, so as to perform fault probability prediction model training for different training batches, for example, , H is the number of historical monitoring cycles, , ,…, are the number of historical monitoring cycles for the 1st, 2nd, …, Sth training batches, respectively. For each historical monitoring cycle, historical fault probability data can be obtained from the recorded transformer on-load tapchanger fault probabilities. For example, if a fault occurs during the transformer on-load tapchanger switching operation, the historical fault probability data is 1; if no fault occurs during the transformer on-load tapchanger switching operation, the historical fault probability data is 0. Faults include contact wear, synchronization issues, drive mechanism problems, brake failure, power system short circuits, fires, and so on. The greater the difference between the vibroacoustic and motor current signals generated during a transformer on-load tapchanger switching operation and the normal vibroacoustic and motor current signals generated during the on-load tapchanger switching operation, the higher the probability of faults during the transformer on-load tapchanger switching operation. For example, the waveform of the vibroacoustic signal may be distorted, with increased high-frequency components and decreased low-frequency components. The greater the difference from the normal vibroacoustic signal, the higher the probability of faults such as loose, worn, or stuck contacts. Increased or decreased amplitude and phase shifts in the motor current signal increase the probability of faults such as motor overload, short circuit, or open circuit. Higher contact temperatures increase the probability of faults during the transformer on-load tapchanger switching operation. For example, high temperatures can accelerate oxidative corrosion and mechanical deformation on the contact surface, leading to melting and spattering of the contact material and, in severe cases, contact burnout. The fault probability prediction model can be based on the relationship between the aforementioned vibroacoustic signals, motor current signals, contact temperature, and the fault probability during transformer on-load tapchanger switching operations. Based on historical operating data, it can predict historical predicted fault probability data for transformer on-load tapchanger switching operations over multiple historical monitoring periods. A loss function is determined based on the relative difference between the historical and predicted fault probability data. By applying feedback adjustment to the loss function, a trained fault probability prediction model is obtained.

[0067] According to one embodiment of the present invention, the loss function of the fault probability prediction model is determined based on the historical operation data, the historical fault probability data and the historical predicted fault probability data, including: determining the loss function of the fault probability prediction model according to formula (2): ,

[0068] (2),

[0069] in, is the historical fault probability data of the transformer on-load tap changer during the i-th switching action in the h-th historical monitoring period, is the historical predicted fault probability data of the transformer on-load tap changer at the i-th switching action in the h-th historical monitoring period, is the historical vibration acoustic signal of the transformer on-load tap-changer during the i-th switching action in the h-th historical monitoring cycle, It is the normal vibration acoustic signal of the transformer on-load tap changer during switching action. is the historical motor current signal of the transformer on-load tap-changer during the i-th switching action in the h-th historical monitoring cycle, It is the normal motor current signal when the transformer on-load tap changer is switching. is the historical contact temperature of the transformer on-load tap-changer during the i-th switching action in the h-th historical monitoring cycle, It is the normal contact temperature of the transformer on-load tap changer during switching action. is the number of historical monitoring cycles of the sth training batch, S is the number of training batches, N is the number of transformer on-load tap-changer switching actions in the monitoring cycle, i≤N, h≤ , s≤S, and i, h, s, N, and S are both positive integers.

[0070] According to one embodiment of the present invention, in formula (2), It is the relative difference between the historical fault probability data of the transformer on-load tapchanger at the i-th switching action in the h-th historical monitoring period and the historical predicted fault probability data of the transformer on-load tapchanger at the i-th switching action in the h-th historical monitoring period. It is the ratio of the historical contact temperature of the transformer on-load tapchanger during the i-th switching action in the h-th historical monitoring cycle to the normal contact temperature of the transformer on-load tapchanger during the switching action. The larger the ratio, the higher the historical contact temperature, and the greater the probability of failure during the switching action of the transformer on-load tapchanger. That is, the historical contact temperature is positively correlated with the failure probability. When the historical contact temperature is higher, the oxidation corrosion and mechanical deformation of the contact surface will be accelerated, resulting in melting and sputtering of the contact material. In severe cases, it will cause contact burning and other failures. Therefore, the larger the historical contact temperature is relative to the normal contact temperature, that is, The larger the value of , the greater the impact on the error of historical predicted failure probability data. is the similarity between the historical vibration and acoustic signal of the transformer on-load tap-changer during the i-th switching action in the h-th historical monitoring period and the normal vibration and acoustic signal of the transformer on-load tap-changer during the switching action. is the similarity between the historical motor current signal of the transformer on-load tapchanger during the i-th switching action in the h-th historical monitoring period and the normal motor current signal during the transformer on-load tapchanger switching action. To achieve a similar fault probability detection effect, the closer the historical vibration and acoustic signals, the historical motor current signals, and the normal vibration and acoustic signals, the normal motor current signals, that is, and The larger the value is, the more similar the conditions of the vibration and acoustic signals and historical motor current signals during the transformer on-load tap-changer switching action are to the normal conditions of the vibration and acoustic signals and normal motor current signals, and the greater its reference value is. Therefore, the higher its weight is. It represents the sum of the relative errors of the historical predicted fault probability data of the transformer on-load tap changer during multiple switching actions in each historical monitoring period of the sth training batch multiplied by the corresponding weight. is the weight of the sth training batch, which is used to reasonably weight the relative errors of different training batches in the loss function. For the historical monitoring period of the s+1th training batch, the accuracy of the s+1th historical predicted fault probability data output by the fault probability prediction model is usually higher than the accuracy of the historical predicted fault probability data of the sth training batch. That is, the shorter the time interval between a training batch and the first training batch, the less accurate its prediction result. In order to improve the training efficiency, the higher its weight is set, and vice versa, the more accurate the prediction result, the lower its weight is. Therefore, a higher weight can be given to items with lower accuracy, thereby improving the training intensity and training efficiency. and The product of the two terms can represent the loss function of the failure probability prediction model.

[0071] According to one embodiment of the present invention, during the process of training a fault probability prediction model, the loss function is back-propagated and some parameters within the model are adjusted to reduce the value of the loss function of the fault probability prediction model, thereby improving the accuracy of the fault probability prediction model and obtaining a trained fault probability prediction model.

[0072] In this way, the influence of the historical contact temperature on the fault probability can be used to determine the influence of the above data on the error of the historical predicted fault probability data, and thus based on the influence and the relative error between the historical fault probability data and the historical predicted fault probability data, and based on the conditions of the vibration acoustic signal and the historical motor current signal during the switching action of the on-load tapchanger of the transformer, the more similar they are to the conditions of the normal vibration acoustic signal and the normal motor current signal, the greater the reference value, and the shorter the time interval with the first training batch, the lower the accuracy, so as to set the weighted sum of the errors output by the fault probability prediction model during multiple switching actions of the on-load tapchanger of the transformer in each historical monitoring cycle of the sth batch, and obtain the loss function to improve the design accuracy and objectivity of the loss function, thereby improving the training efficiency during the training process and improving the accuracy of the fault probability prediction model.

[0073] According to one embodiment of the present invention, in step S105 , it is determined whether the transformer on-load tap changer meets a preset target condition based on the predicted fault probability data and the switching performance score.

[0074] According to one embodiment of the present invention, step S105 includes: setting a fault probability threshold and a switching performance score threshold; and determining whether the transformer on-load tap changer meets preset target conditions based on the predicted fault probability data, the switching performance score, the fault probability threshold, and the switching performance score threshold.

[0075] According to one embodiment of the present invention, a fault probability threshold and a switching performance score threshold are set. For example, the fault probability threshold may be set to 70%. If the predicted fault probability data exceeds the fault probability threshold, it indicates that a fault accident occurs during the switching operation of the transformer on-load tap changer. The switching performance score threshold may be set to 0.9. If the switching performance score exceeds the switching performance score threshold, it indicates that the switching performance of the transformer on-load tap changer during the switching operation is normal and no fault accident occurs.

[0076] According to one embodiment of the present invention, determining whether the transformer on-load tap changer meets a preset target condition based on the predicted fault probability data, the switching performance score, the fault probability threshold, and the switching performance score threshold includes: determining a first condition C1 and a second condition C2 according to formulas (3) and (4);

[0077] (3),

[0078] (4),

[0079] in, is the predicted fault probability data of the transformer on-load tap changer during the i-th switching action in the current monitoring period, is the fault probability threshold, F is the piecewise function that compares the predicted fault probability data of the transformer on-load tap changer at the i-th switching action in the current monitoring period with the fault probability threshold, is the preset average failure probability data within a monitoring period, Score the switching performance, is the switching performance scoring threshold, N is the number of switching actions of the transformer on-load tap changer in the monitoring period, i≤N, and both i and N are positive integers, the first condition C1 and the second condition C2 are the preset target conditions; when the first condition C1 and the second condition C2 are simultaneously met, it is determined that the transformer on-load tap changer meets the preset target conditions.

[0080] According to one embodiment of the present invention, in formula (3), When the predicted fault probability data of the transformer on-load tapchanger at the i-th switching action in the current monitoring period is greater than or equal to the fault probability threshold, the value is 1. When the predicted fault probability data of the transformer on-load tapchanger at the i-th switching action in the current monitoring period is less than the fault probability threshold, the value is 0. If the predicted fault probability data is relatively large, it indicates that the fault probability of the transformer on-load tapchanger at the switching action in the current monitoring period is high. In the first condition of formula (4), It means that the average value of F at all times in the current monitoring cycle is less than the preset average fault probability data in a monitoring cycle, wherein the preset average fault probability data can range from 0.55 to 0.6, that is, the fault probability in the monitoring cycle is lower than the threshold, indicating that no fault occurs when the transformer on-load tap changer switches. In the second condition, If the switching performance score is greater than or equal to the switching performance score threshold, it means that the switching performance of the transformer on-load tap changer is good and no fault occurs during the switching operation.

[0081] In this way, the first condition and the second condition can be determined based on the predicted fault probability data, the switching performance score, the fault probability threshold and the switching performance score threshold. When the first condition and the second condition are simultaneously met, it can be determined that the transformer on-load tap changer meets the preset target conditions. The predicted fault probability data and the switching performance score are quantified, which improves the accuracy of judging the preset target conditions and realizes a comprehensive evaluation of the status of the transformer on-load tap changer.

[0082] According to one embodiment of the present invention, in step S106, if the transformer on-load tap changer meets the preset target condition, it is determined that the transformer on-load tap changer is currently in a normal state, that is, the transformer on-load tap changer can maintain stable performance and has a low failure risk under the current operating environment.

[0083] According to one embodiment of the present invention, in step S107, if the transformer on-load tap changer does not meet the preset target condition, it is determined that the transformer on-load tap changer is currently in a fault state, that is, the transformer on-load tap changer has performance degradation or potential safety hazards under the current operating environment, and corresponding maintenance or repair measures need to be taken immediately.

[0084] According to an embodiment of the present invention, the online monitoring and fault diagnosis method for transformer on-load tapchangers comprehensively monitors the switching performance of the on-load tapchanger using arc energy and contact position information, thereby evaluating the on-load tapchanger status. By analyzing operating data using a trained fault probability prediction model, the fault probability of the transformer on-load tapchanger can be predicted, facilitating the normal operation of the transformer on-load tapchanger and providing timely warnings when faults occur, reducing the risk of faults, and improving the reliability and stability of the transformer on-load tapchanger. When determining the switching performance score, the arc energy, the voltage value corresponding to the on-load tapchanger contact position information, the normal range of the arc energy, and the voltage value corresponding to the target position information are used to determine the switching performance score. The arc energy and the voltage value corresponding to the contact position information respectively reflect the electrical and mechanical performance of the switch during switching, improving the comprehensiveness and reliability of the switching performance score. When determining the loss function of the fault probability prediction model, the influence of the historical contact temperature on the fault probability can be used to determine the influence of the above data on the error of the historical predicted fault probability data. Based on the influence and the relative error between the historical fault probability data and the historical predicted fault probability data, and based on the conditions of the vibration acoustic signal and the historical motor current signal during the switching action of the transformer on-load tapchanger, the weight is set based on the characteristic that the more similar the conditions are to the normal vibration acoustic signal and the normal motor current signal, the greater the reference value, and the shorter the time interval with the first training batch, the lower the accuracy. In this way, the errors output by the fault probability prediction model during multiple switching actions of the transformer on-load tapchanger in each historical monitoring cycle of the sth batch are weightedly summed to obtain the loss function, so as to improve the design accuracy and objectivity of the loss function, thereby improving the training efficiency during the training process and improving the accuracy of the fault probability prediction model. When determining whether the transformer on-load tap-changer meets the preset target conditions, the first condition and the second condition can be determined based on the predicted fault probability data, the switching performance score, the fault probability threshold and the switching performance score threshold. When the first condition and the second condition are simultaneously met, it can be determined that the transformer on-load tap-changer meets the preset target conditions. The predicted fault probability data and the switching performance score are quantified, which improves the accuracy of judging the preset target conditions and realizes a comprehensive evaluation of the status of the transformer on-load tap-changer.

[0085] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0086] Those skilled in the art will appreciate that the embodiments of the present invention described above and shown in the accompanying drawings are intended to be illustrative only and are not intended to limit the present invention. The objectives of the present invention have been fully and effectively achieved. The functional and structural principles of the present invention have been demonstrated and illustrated in the embodiments. Any variations or modifications may be made to the embodiments of the present invention without departing from the principles described.

Claims

1. A method for online monitoring and fault diagnosis of a transformer on-load tap changer, characterized in that: include: Obtain the arc energy and contact position information of the on-load tap changer generated during the transformer on-load tap changer switching action in the current monitoring period; determining a switching performance score according to the arc energy and the contact position information of the on-load tap changer; Acquiring operating data of the transformer on-load tap changer during a current monitoring cycle, wherein the operating data includes a vibration acoustic signal, a motor current signal, and a contact temperature generated during a switching operation of the transformer on-load tap changer; Inputting the operating data into a trained fault probability prediction model to obtain predicted fault probability data of the transformer on-load tap changer switching action in the current monitoring period; determining whether the transformer on-load tap changer meets a preset target condition based on the predicted fault probability data and the switching performance score; If the transformer on-load tap changer meets the preset target condition, determining that the transformer on-load tap changer is in a normal state; If the transformer on-load tap changer does not meet the preset target condition, it is determined that the transformer on-load tap changer is in a fault state.

2. The method for online monitoring and fault diagnosis of transformer on-load tap changer according to claim 1, characterized in that: Determining a switching performance score according to the arc energy and the contact position information of the on-load tap changer includes: Set the normal range of arc energy; Obtaining a voltage value corresponding to the on-load tap changer contact position information; The voltage value corresponding to the target tap position of the transformer on-load tap changer is set to the voltage value corresponding to the target position information; A switching performance score is determined according to the arc energy, the voltage value corresponding to the on-load tap changer contact position information, the normal range of the arc energy, and the voltage value corresponding to the target position information.

3. The transformer on-load tap-changer online monitoring and fault diagnosis method according to claim 2, characterized in that: Determining a switching performance score according to the arc energy, a voltage value corresponding to the on-load tap changer contact position information, a normal range of the arc energy, and a voltage value corresponding to the target position information includes: According to the formula ， Determining the handover performance score ,in, is the arc energy generated by the transformer on-load tap-changer during the i-th switching action in the current monitoring period, is the minimum arc energy, is the maximum arc energy, is the voltage value corresponding to the contact position information of the on-load tapchanger during the i-th switching action of the transformer on-load tapchanger in the current monitoring period, is the voltage value corresponding to the target position information during the i-th switching action of the transformer on-load tap-changer in the current monitoring period, N is the number of switching actions of the transformer on-load tap-changer in the monitoring period, i≤N, and both i and N are positive integers, and if is a conditional function.

4. The method for online monitoring and fault diagnosis of transformer on-load tap changer according to claim 1, characterized in that: The training steps of the fault probability prediction model include: Acquiring historical operating data of the transformer on-load tap-changer during a historical monitoring period, wherein the historical operating data includes historical vibration and acoustic signals, historical motor current signals, and historical contact temperatures generated during a switching action of the transformer on-load tap-changer; Obtain historical fault probability data of transformer on-load tap-changer switching actions during multiple historical monitoring periods; The historical operation data is processed by a fault probability prediction model to obtain historical predicted fault probability data of the transformer on-load tap changer switching action during multiple historical monitoring periods; Determining a loss function of the failure probability prediction model based on the historical operating data, the historical failure probability data, and the historical predicted failure probability data; The fault probability prediction model is trained according to the loss function of the fault probability prediction model to obtain the trained fault probability prediction model.

5. The method for online monitoring and fault diagnosis of transformer on-load tap-changer according to claim 4, characterized in that: Determining a loss function of the failure probability prediction model based on the historical operating data, the historical failure probability data, and the historical predicted failure probability data includes: According to the formula ， Determine the loss function of the failure probability prediction model ,in, is the historical fault probability data of the transformer on-load tap changer during the i-th switching action in the h-th historical monitoring period, is the historical predicted fault probability data of the transformer on-load tap changer at the i-th switching action in the h-th historical monitoring period, is the historical vibration acoustic signal of the transformer on-load tap-changer during the i-th switching action in the h-th historical monitoring cycle, It is the normal vibration acoustic signal of the transformer on-load tap changer during switching action. is the historical motor current signal of the transformer on-load tap-changer during the i-th switching action in the h-th historical monitoring cycle, It is the normal motor current signal when the transformer on-load tap changer is switching. is the historical contact temperature of the transformer on-load tap-changer during the i-th switching action in the h-th historical monitoring cycle, It is the normal contact temperature of the transformer on-load tap changer during switching action. is the number of historical monitoring cycles of the sth training batch, S is the number of training batches, N is the number of transformer on-load tap-changer switching actions in the monitoring cycle, i≤N, h≤ , s≤S, and i, h, s, N, and S are both positive integers.

6. The method for online monitoring and fault diagnosis of transformer on-load tap changer according to claim 1, characterized in that: Determining whether the transformer on-load tap changer meets a preset target condition based on the predicted fault probability data and the switching performance score includes: Set the failure probability threshold and switching performance score threshold; It is determined whether the transformer on-load tap changer meets a preset target condition according to the predicted fault probability data, the switching performance score, the fault probability threshold, and the switching performance score threshold.

7. The transformer on-load tap-changer online monitoring and fault diagnosis method according to claim 6, characterized in that: Determining whether the transformer on-load tap changer meets a preset target condition based on the predicted fault probability data, the switching performance score, the fault probability threshold, and the switching performance score threshold includes: According to the formula ， ， Determine the first condition C1 and the second condition C2, where: is the predicted fault probability data of the transformer on-load tap changer during the i-th switching action in the current monitoring period, is the fault probability threshold, F is the piecewise function that compares the predicted fault probability data of the transformer on-load tap changer at the i-th switching action in the current monitoring period with the fault probability threshold, is the preset average failure probability data within a monitoring period, Score the switching performance, is the switching performance scoring threshold, N is the number of switching actions of the transformer on-load tap changer in the monitoring period, i≤N, and both i and N are positive integers, the first condition C1 and the second condition C2 are the preset target conditions; When the first condition C1 and the second condition C2 are satisfied at the same time, it is determined that the transformer on-load tap changer meets the preset target condition.

Images