Method for predicting residual service life of low-voltage parallel capacitor based on artificial intelligence

Abstract

The invention relates to a low-voltage parallel capacitor residual service life prediction method based on artificial intelligence. The invention relates to the technical field of capacitor residual service life prediction, and the method comprehensively considers various environmental factors, and adopts an improved bidirectional long-short-term memory network (BiLSTM) combining a convolutional neural network (CNN) and a whale optimization algorithm (WOA) to improve the prediction precision, speed and robustness. Compared with a traditional LSTM (Long Short-Term Memory) neural network model, an experimental result shows that the root mean square error (RMSE) and the mean absolute error (MAE) are remarkably reduced, and the stability and the reliability of a prediction result are improved. The method can effectively reduce the loss of the power system caused by prediction errors, improves the operation efficiency of equipment, prolongs the service life of the equipment, and has a higher practical application value.

Description

Technical Field

[0001] The present invention relates to the technical field of predicting the remaining service life of capacitors, and is a method for predicting the remaining service life of low-voltage shunt capacitors based on artificial intelligence. Background Art

[0002] Low-voltage shunt capacitors are widely used in power systems, mainly for improving power factor, stabilizing voltage, reducing line losses, etc. However, with the increase in operation time, capacitors will gradually age due to factors such as overload, temperature fluctuations, and voltage instability, resulting in performance degradation or failure, affecting the stability of the power system. To avoid system outages or equipment damage caused by capacitor failures, it is of great significance to accurately predict its remaining service life (RUL). Traditional prediction methods mainly rely on physical models and statistical analysis, and are calculated based on empirical laws and simplified assumptions. However, these methods have problems such as poor adaptability and low prediction accuracy when dealing with complex environments. In recent years, the application of deep learning algorithms has gradually deepened in various engineering fields. Convolutional neural networks (CNNs) and bidirectional long short-term memory networks (BiLSTMs) have shown strong performance by effectively extracting features and modeling time series data. Combining these two models can better handle the non-linear features and time-dependent relationships in capacitor operation data, thereby improving the accuracy of life prediction. To further improve the prediction effect, the whale optimization algorithm (WOA), as a new type of intelligent optimization algorithm, optimizes the neural network weight parameters by simulating the predation behavior of whales, has strong global search ability, and can effectively avoid the problem of local optimal solutions. Combining WOA with CNN-BiLSTM can further enhance the prediction accuracy and robustness of the model, fully mine the potential information in capacitor operation data, and thus provide a more accurate prediction of the remaining service life. This method can not only improve the accuracy of capacitor life prediction, but also provide a scientific basis for the maintenance decision-making of the power system, supporting the safe, stable and efficient operation of the system. Summary of the Invention

[0003] Aiming at the deficiencies of the existing technology, the present invention provides a method for predicting the remaining service life of low-voltage shunt capacitors based on artificial intelligence to solve the problem of being unable to accurately predict the remaining service life of low-voltage shunt capacitance values, and to avoid the national economic losses and the suspension of daily production and life caused by inaccurate prediction.

[0004] The present invention provides the following technical solutions:

[0005] A method for predicting the remaining service life of low-voltage shunt capacitors based on artificial intelligence, the method comprising the following steps:

[0006] Step 1: Calculate the rated capacitance value by using the rated data on the nameplate of the low-voltage shunt capacitor adopted.

[0007] Step 2: Based on the low-voltage shunt capacitor cyclic aging experiment test platform, by accelerating the number of cyclic charge and discharge times N, read and record the working parameters during the switching of capacitors every time interval t, to simulate the aging process of the equivalent capacitance value during operation and obtain the remaining capacitance value data calculated from the voltage and current data at different times under different working conditions;

[0008] Step 3: Form three groups of data sets from the obtained data, respectively verify the influence of different influencing factors on the remaining service life of low-voltage shunt capacitors, and establish the mapping relationship between each influencing factor and the remaining capacitance value;

[0009] Step 4: Combine the obtained data sets into a normalized data set, clean and normalize the data, and divide it into a training set and a test set;

[0010] Step 5: Build a CNN model from the training set, extract local features from the data, process them through the convolutional layer and the pooling layer, and then build a BiLSTM model. Use the BiLSTM network to capture the dependencies of the time series and output the predicted value of the capacitor life;

[0011] Step 6: Use the output optimal BiLSTM network parameters and model to test the test set, and use RMSE and MAE to measure the accuracy of the experiment;

[0012] Step 7: Deploy the output optimal BiLSTM network parameters and model to the actual system for real-time prediction of the remaining life of the capacitor.

[0013] The specific content of step 2 is as follows:

[0014] Through the low-voltage shunt capacitor cyclic aging experiment test platform, the cyclic charge and discharge times N are equivalently replaced by the actual working time of the low-voltage shunt capacitor; in the cyclic aging test platform, the charging is provided by a constant current source, the discharging is carried out through an electronic load, and the relay is responsible for switching the charge and discharge states;

[0015] When the constant current source is connected, the low-voltage shunt capacitor is in the charging state; when the electronic load is connected, the discharging is carried out;

[0016] The controller realizes these switching operations through the connected relay. At the same time, the voltage detection module monitors the terminal voltage of the low-voltage shunt capacitor in real time and compares it with the preset charge and discharge cut-off voltages;

[0017] The thermostat provides specific temperature conditions for the experiment. Through the above real-time data calculation, when the remaining capacitance value of the capacitor reaches 20% of its rated value, the experiment is stopped and the experimental cycle number N is recorded.

[0018] Preferably, step 2 further includes:

[0019] There are three variables: the working voltage U, the compensation current I, and the working temperature °C. Each variable is set to three different values with a deviation not exceeding ±5% of the rated working value. When verifying the influence of one variable on the capacitance value, the consistency is strictly controlled.

[0020] Preferably, the obtained data sets are merged into a normalized data set, the data is cleaned and normalized, the data set is randomly grouped, and 80% of it is divided into a training set and 20% into a test set.

[0021] Preferably, step 5 is specifically as follows:

[0022] Construct a CNN model from the training set obtained in step four, extract local features from the data, process them through the convolutional layer and the pooling layer, and then construct a BiLSTM model. Use the BiLSTM network to capture the dependencies of the time series and output the predicted value of the capacitor life.

[0023] Enter the fitness evaluation, and use the prediction error of the BiLSTM model as the effect of the fitness evaluation hyperparameter.

[0024] Through position update, select the shrinking encirclement or spiral path for update according to the whale optimization algorithm.

[0025] Continue iterative update until the predetermined condition is reached.

[0026] Preferably, the step 5 further includes the continuous iterative process, which adopts the early stopping method and cross-validation to prevent overfitting, and introduces the learning rate adaptive adjustment algorithm to dynamically adjust the learning rate during the training process in order to find the optimal solution.

[0027] By continuously adjusting the algorithm iteration, the number of training times, the number of searches, and the number of hidden layers in the convolutional neural network, the optimal parameters of BiLSTM are obtained.

[0028] Preferably, the root mean square error RMSE and the mean absolute error MAE are used to measure the accuracy. MAE is used to calculate the average of the absolute values of the differences between the predicted value and the true value. RMSE determines the degree of dispersion of the samples, and the smaller the RMSE obtained by calculating the root mean square error of the data verified by non-linear fitting of the data, the better.

[0029] A remaining service life prediction system for low-voltage shunt capacitors based on artificial intelligence, the system includes:

[0030] A calculation module, which calculates the rated capacitance value by using the rated data of the nameplate of the low-voltage shunt capacitor adopted.

[0031] An experimental module, which is based on a low-voltage shunt capacitor cyclic aging experimental test platform. By accelerating the number of cyclic charge and discharge times N, the working parameters during the switching of capacitors are read and recorded every time after a time t, so as to simulate the aging process of the equivalent capacitance value during operation and obtain the remaining capacitance value data calculated from the voltage and current data at different times under different working conditions.

[0032] A data module, which forms three groups of data sets from the obtained data, respectively verifies the influence of different influencing factors on the remaining service life of low-voltage shunt capacitors, and establishes the mapping relationship between each influencing factor and the remaining capacitance value.

[0033] An integration module, which combines the obtained data sets into a normalized data set, cleans and normalizes the data, and divides it into a training set and a test set.

[0034] A model training module, which constructs a CNN model from the training set, extracts local features from the data, processes them through convolutional layers and pooling layers, and then constructs a BiLSTM model. The BiLSTM network is used to capture the dependencies of the time series and output the predicted value of the capacitor life.

[0035] A test module, which uses the output optimal BiLSTM network parameters and the model to test the test set, and uses RMSE and MAE to measure the accuracy of the experiment.

[0036] A verification module, which deploys the output optimal BiLSTM network parameters and the model to an actual system for real-time prediction of the remaining life of the capacitor.

[0037] A computer-readable storage medium, on which a computer program is stored, and the program is executed by a processor to implement a method for predicting the remaining service life of low-voltage shunt capacitors based on artificial intelligence.

[0038] A computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, it implements a method for predicting the remaining service life of low-voltage shunt capacitors based on artificial intelligence.

[0039] The present invention has the following beneficial effects:

[0040] Compared with the prior art, the present invention:

[0041] The present invention focuses on the important functions of low-voltage shunt capacitors in the power system, which are widely used in various industrial, commercial, and residential electricity applications to improve power factor, stabilize voltage, and reduce line losses. Compared with traditional remaining service life prediction methods, predicting the remaining service life of capacitors through artificial intelligence algorithms can effectively identify potential faults in advance, avoid outages caused by sudden equipment failures and high maintenance costs, thereby improving the operational stability and economic benefits of the power system. Accurate life prediction can not only optimize the maintenance and replacement plans of capacitors, reduce unnecessary resource waste, but also improve the operating efficiency and return on investment of equipment. With the continuous growth of power demand, ensuring the efficient operation of power equipment is crucial for ensuring sustainable economic development. The prediction method based on the WOA-CNN-BiLSTM neural network can provide more accurate predictions for the power system, help reduce the impact of sudden failures on production and life, thereby improving the overall reliability of the system and promoting the rational use of energy resources. Through this intelligent prediction and optimal scheduling, it not only helps to reduce social and economic costs, but also provides support for sustainable development goals and promotes the realization of a green and low-carbon economy. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0043] Figure 1 It shows the flowchart of the method of the present invention;

[0044] Figure 2 It shows the structural diagram of the low-voltage shunt capacitor cyclic aging experiment platform of the present invention;

[0045] Figure 3 It shows the scatter plot of the influence of temperature on the capacitance value of the low-voltage shunt capacitor of the present invention;

[0046] Figure 4 It shows the scatter plot of the influence of voltage on the capacitance value of the low-voltage shunt capacitor of the present invention;

[0047] Figure 5 It shows the scatter plot of the influence of current on the capacitance value of the low-voltage shunt capacitor of the present invention;

[0048] Figure 6 It shows the result diagram of the prediction of the low-voltage shunt capacitor based on the WOA-CNN-BiLSTM artificial intelligence algorithm of the present invention;

[0049] Figure 7 It shows the error graph of the prediction result of the low - voltage shunt capacitor based on the WOA - CNN - BiLSTM artificial intelligence algorithm of the present invention. Detailed implementation manners

[0050] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0051] In the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be understood as a limitation of the present invention. In addition, the terms "first", "second", "third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0052] In the description of the present invention, it should be noted that unless otherwise clearly defined and limited, the terms "installed", "connected", "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0053] The present invention is described in detail below in combination with specific embodiments. Specific embodiment one:

[0055] According to Figures 1 to 7 As shown, the specific optimized technical solution adopted by the present invention to solve the above - mentioned technical problems is: The present invention relates to a method for predicting the remaining service life of low - voltage shunt capacitors based on artificial intelligence.

[0056] The present invention provides a method for predicting the remaining service life of low - voltage shunt capacitors based on artificial intelligence, and the method includes the following steps:

[0057] Step 1: Calculate the rated capacitance value by using the rated data on the nameplate of the low - voltage shunt capacitor.

[0058] Step 2: Based on the low-voltage shunt capacitor cyclic aging experimental test platform, by accelerating the number of cyclic charge and discharge N, read and record the working parameters during the switching of capacitors every time interval t, to simulate the aging process of the equivalent capacitance value during operation and obtain the remaining capacitance value data calculated from the voltage and current data at different times under different working conditions;

[0059] Step 3: Form three groups of data sets from the obtained data, respectively verify the influence of different influencing factors on the remaining service life of low-voltage shunt capacitors, and establish the mapping relationship between each influencing factor and the remaining capacitance value;

[0060] Step 4: Merge the obtained data sets into a normalized data set, clean and normalize the data, and divide it into a training set and a test set;

[0061] Step 5: Construct a CNN model from the training set, extract local features from the data, process them through the convolutional layer and the pooling layer, and then construct a BiLSTM model. Use the BiLSTM network to capture the dependencies of the time series and output the predicted value of the capacitor life;

[0062] Step 6: Use the output optimal BiLSTM network parameters and model to test the test set, and use RMSE and MAE to measure the accuracy of the experiment;

[0063] Step 7: Deploy the output optimal BiLSTM network parameters and model to the actual system for real-time prediction of the remaining life of the capacitor.

[0064] The research object of the present invention is low-voltage shunt capacitors, which are mainly used for reactive power compensation in power systems, focusing on improving voltage stability and power factor. Its aging mechanism is mainly affected by multiple factors such as temperature, voltage, and current, resulting in dielectric aging and thermal stress. The life termination criterion is that the capacitance value decays to 80% of the initial value. The purpose of the present invention is to accurately predict the remaining service life of low-voltage shunt capacitors, thereby greatly avoiding the occurrence of serious electrical accidents and personal injuries, while ensuring the power quality of the power system and extending the service life of other electrical equipment in the power system.

[0065] The present invention adopts a WOA-CNN-BiLSTM hybrid model, where CNN is used to extract local features, BiLSTM processes bidirectional time series dependencies, and WOA (Whale Optimization Algorithm) is used to optimize hyperparameters. This model directly fuses spatio-temporal features through the cascade of CNN and BiLSTM.

[0066] The core evaluation metrics of the present invention are RMSE (0.0193) and MAE (0.0087), and the comparison benchmarks include traditional LSTM, NSGA-II-SVR, Kalman filtering, etc. In addition, the present invention has also achieved real-time prediction (8ms-level response) on the STM32H743 microcontroller, meeting the standards of engineering real-time monitoring. The core evaluation metrics of the invention patent are RMSE (0.030752), MAE (0.030138), and the correlation coefficient R2 (0.99315), and the comparison benchmark is a model optimized by single LSTM, SVR, and traditional PSO.

[0067] The core innovations of the present invention include the WOA-CNN-BiLSTM hybrid architecture, two-way time series modeling to suppress noise interference, and hardware real-time deployment verification. Its theoretical value mainly lies in proposing a new WOA-CNN-BiLSTM algorithm model, providing a new idea for predicting the remaining useful life of complex electrical equipment. Its engineering value mainly lies in avoiding the decline of power quality, malignant electrical accidents, and personal injury accidents caused by the prediction error of the remaining useful life of reactive power compensation equipment in the power system. The core innovations of its invention patent include IPSO fusion with GA to improve parameter search, the LSTM-SVR weight allocation strategy, and the life test design for pulse scenarios. Its theoretical contribution lies in verifying the superiority of the combined model in small-sample and non-linear scenarios, and its engineering value lies in improving the reliability of NB network equipment and reducing the risk of communication interruption.

[0068] The present invention focuses on low-voltage shunt capacitors, adopts the WOA-CNN-BiLSTM hybrid model, emphasizes the overall architecture and hardware adaptation, and pays more attention to industrial practicability. Specific Embodiment 2:

[0070] The difference between Embodiment 2 and Embodiment 1 of this application is only that:

[0071] Step 2 is specifically:

[0072] Through the low-voltage shunt capacitor cyclic aging experiment test platform, the number of cyclic charge and discharge times N is equivalently replaced by the actual working time of the low-voltage shunt capacitor; in the cyclic aging test platform, the charging is provided by a constant current source, the discharging is carried out through an electronic load, and the relay is responsible for switching the charge and discharge states;

[0073] When the constant current source is turned on, the low-voltage shunt capacitor is in the charging state; when the electronic load is turned on, the discharging is carried out;

[0074] The controller realizes these switching operations through the connected relay. At the same time, the voltage detection module monitors the terminal voltage of the low-voltage shunt capacitor in real time and compares it with the preset charge and discharge cut-off voltages;

[0075] The incubator provides specific temperature conditions for the experiment. Through real-time calculation of the above data, when the remaining capacitance value of the capacitor reaches 20% of its rated value, the experiment is stopped, and the number of experimental cycles N is recorded. Specific Embodiment Three:

[0077] The difference between Embodiment Three and Embodiment Two of this application lies only in:

[0078] Step 2 further includes:

[0079] There are three variables: working voltage U, compensation current I, and working temperature °C. Each variable is set to three different values, and the deviation does not exceed ±5% of the rated working value. When verifying the influence of one variable on the capacitance value, strict consistency control is carried out. Specific Embodiment Four:

[0081] The difference between Embodiment Four and Embodiment Three of this application lies only in:

[0082] The obtained data sets are merged into a normalized data set, the data is cleaned and normalized, and the data set is randomly grouped, with 80% divided into the training set and 20% divided into the test set. Specific Embodiment Five:

[0084] The difference between Embodiment Five and Embodiment Four of this invention lies only in:

[0085] Step 5 is specifically:

[0086] Build a CNN model with the training set obtained in Step Four, extract local features from the data, process them through the convolutional layer and the pooling layer, and then build a BiLSTM model. Use the BiLSTM network to capture the dependencies of the time series and output the predicted value of the capacitor life;

[0087] Enter the fitness evaluation, and use the prediction error of the BiLSTM model as the effect of the fitness evaluation hyperparameter;

[0088] Through position update, select the shrinking encirclement or spiral path for update according to the whale optimization algorithm;

[0089] Continue iterative update until the predetermined condition is reached. Specific Embodiment Six:

[0091] The difference between Embodiment Six and Embodiment Five of this invention lies only in:

[0092] Step 5 further includes the continuous iterative process, which adopts the early stopping method and cross-validation to prevent overfitting, and introduces the learning rate adaptive adjustment algorithm to dynamically adjust the learning rate during the training process in order to find the optimal solution;

[0093] By continuously adjusting the algorithm iteration, the number of training times, the number of searches, and the number of hidden layers in the convolutional neural network, the optimal parameters of the BiLSTM are obtained. Specific Embodiment Seven:

[0095] The difference between Embodiment Seven and Embodiment Six of the present invention lies only in:

[0096] The accuracy is measured by the root mean square error RMSE and the mean absolute error MAE. MAE is used to calculate the average of the absolute values of the differences between the predicted values and the true values. RMSE determines the degree of dispersion of the samples. The smaller the RMSE obtained by calculating the root mean square error of the data verified by non-linear fitting of the data, the better. Specific Embodiment Eight:

[0098] The difference between Embodiment Eight and Embodiment Seven of the present invention lies only in:

[0099] The present invention provides a remaining service life prediction system for low-voltage shunt capacitors based on artificial intelligence. The system includes:

[0100] A calculation module that calculates the rated capacitance value by using the rated data on the nameplate of the low-voltage shunt capacitor.

[0101] An experiment module that, based on a low-voltage shunt capacitor cyclic aging experiment test platform, reads and records the working parameters during the switching of the capacitor every time after accelerating the cyclic charge and discharge times N, to simulate the aging process of the equivalent capacitance value during operation and obtain the remaining capacitance value data calculated from the voltage and current data at different times under different working conditions.

[0102] A data module that forms three groups of data sets from the obtained data to respectively verify the influence of different influencing factors on the remaining service life of the low-voltage shunt capacitor, and establish a mapping relationship between each influencing factor and the remaining capacitance value.

[0103] An integration module that merges the obtained data sets into a normalized data set, cleans and normalizes the data, and divides it into a training set and a test set.

[0104] A model training module that constructs a CNN model from the training set, extracts local features from the data, processes them through convolutional layers and pooling layers, and then constructs a BiLSTM model. The BiLSTM network is used to capture the dependencies of the time series and output the predicted value of the capacitor life.

[0105] A test module that tests the output of the best BiLSTM network parameters with the model using the test set, and measures the accuracy of the experiment by using RMSE and MAE.

[0106] The verification module deploys the output optimal BiLSTM network parameters and the model into an actual system for real-time prediction of the remaining life of the capacitor.

[0107] The present invention provides a method for predicting the remaining service life of a low-voltage shunt capacitor. This method comprehensively considers various environmental factors and uses an improved bidirectional long short-term memory network (BiLSTM) that combines a convolutional neural network (CNN) and a whale optimization algorithm (WOA) to improve the prediction accuracy, speed, and robustness. By comparing with a traditional long short-term memory neural network (LSTM) model, the experimental results show that the present invention significantly reduces both the root mean square error (RMSE) and the mean absolute error (MAE), improving the stability and reliability of the prediction results. The present invention can effectively reduce the power system losses caused by prediction errors, improve the equipment operation efficiency, extend the equipment service life, and has high practical application value. Specific Embodiment Nine:

[0109] The difference between the ninth embodiment and the eighth embodiment of the present invention lies only in:

[0110] The present invention provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it is used to implement a method for predicting the remaining service life of a low-voltage shunt capacitor based on artificial intelligence. Specific Embodiment Ten:

[0112] The difference between the tenth embodiment and the ninth embodiment of the present invention lies only in:

[0113] The present invention provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements a method for predicting the remaining service life of a low-voltage shunt capacitor based on artificial intelligence. Specific Embodiment Eleven:

[0115] The difference between the eleventh embodiment and the tenth embodiment of the present invention lies only in:

[0116] The object of the present invention is to solve the problem of being unable to accurately predict the remaining service life of a low-voltage shunt capacitance value and avoid the national economic losses and the suspension of daily production and life caused by inaccurate prediction. Therefore, the present invention provides a method for predicting the remaining service life of a low-voltage shunt capacitor based on artificial intelligence.

[0117] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0118] Step 1: Calculate the rated capacitance value using the rated data on the nameplate of the low-voltage shunt capacitor. Based on the known operating characteristics of the capacitor at different capacitance values, use the remaining capacitance value as the basis for judging the lifespan, thus forming the basic conditions for the lifespan prediction theory;

[0119] Step 2: Through the low-voltage shunt capacitor cyclic aging experiment test platform, accelerate the cyclic charge and discharge times N. Read and record the operating parameters during the switching of the capacitor every time t has passed. Obtain the remaining capacitance value data calculated from the aging process of the equivalent capacitance value during operation and the voltage and current data at different times under different operating conditions. There are three variables in this experiment: operating voltage U, compensation current I, and operating temperature °C. Set three different values for each variable, with a deviation not exceeding ±5% of the rated operating value. When verifying the influence of one variable on the capacitance value, strictly control the consistency of other experimental variables;

[0120] Step 3: Form three groups of data sets from the experimental data obtained in Step 2, respectively verify the influence of different influencing factors on the remaining service life of the low-voltage shunt capacitor, and establish the mapping relationship between each influencing factor and the remaining capacitance value;

[0121] Step 4: Merge the data sets obtained in Step 2 into a normalized data set, clean and normalize the data, randomly divide the data set, and 80% of it is divided into the training set and 20% into the test set;

[0122] Step 5: Build a CNN model from the training set obtained in Step 4, extract local features from the data, and process them through the convolutional layer and pooling layer. Then build a BiLSTM model, use the BiLSTM network to capture the dependencies of the time series, and output the capacitor lifespan prediction value. Then enter the fitness evaluation, using the prediction error of the BiLSTM model as the effect of the fitness evaluation hyperparameter. Through position update, select the shrinking encirclement or spiral path for update according to the whale optimization algorithm. Continue iterative update until the predetermined condition is reached;

[0123] Step 6: After the above steps are completed, use the output optimal BiLSTM network parameters and model to test the test set, and use RMSE and MAE to measure the accuracy of the experiment. If the accuracy meets the requirements, proceed to the next step;

[0124] Step 7: After the system completes the above steps, deploy the optimal BiLSTM network parameters and model output in Step 5 to the actual system for real-time prediction of the remaining lifespan of the capacitor, completing the entire process;

[0125] Furthermore, in Step 2, by setting multiple groups of experimental parameters and ensuring the reliability of the experimental environment, a large amount of experimental data is obtained to ensure the basis for data training of the artificial intelligence algorithm;

[0126] Furthermore, in Step 2, through a specially designed experimental platform, the number of charge-discharge cycles N is equivalently replaced by the actual working time of the low-voltage shunt capacitor. In the cyclic aging test platform, charging is provided by a constant current source, discharging is carried out through an electronic load, and a relay is responsible for switching the charge-discharge state. When the constant current source is turned on, the low-voltage shunt capacitor is in the charging state; when the electronic load is turned on, discharging occurs. The controller realizes these switching operations through the connected relay. At the same time, the voltage detection module can monitor the terminal voltage of the low-voltage shunt capacitor in real time and compare it with the preset charge-discharge cut-off voltage. The thermostat provides specific temperature conditions for the experiment. Through the above real-time data calculation, the experiment is stopped when the remaining capacitance value of the capacitor reaches 20% of its rated value, and the number of experimental cycles N is recorded. Based on the above experimental principle, it is ensured that the experiment is close to the actual working mode, avoiding data errors caused by unreasonable experimental design;

[0127] Furthermore, the convolutional neural network (CNN) in Step 5 does not require manual feature design and can capture the non-linear and complex relationships hidden in the capacitor life. Time series models such as bidirectional LSTM (BiLSTM) can handle the long-term dependence problem in capacitor life prediction, not only being able to analyze the historical data existing in the dataset but also predicting future trends, improving the accuracy of life prediction;

[0128] Furthermore, in the continuous iteration process of Step 5, the early stopping method and cross-validation are adopted to prevent overfitting, and a learning rate adaptive adjustment algorithm is introduced to dynamically adjust the learning rate during the training process in order to find the optimal solution. By continuously adjusting the algorithm iteration, the number of training times, the number of searches, and the number of hidden layers in the convolutional neural network, the optimal parameters of BiLSTM are obtained;

[0129] Furthermore, in Step 7, the root mean square error (RMSE) and the mean absolute error (MAE) are used. MAE is used to calculate the average of the absolute values of the differences between the predicted values and the true values. RMSE illustrates the degree of dispersion of the samples. In this experiment, the smaller the RMSE obtained by calculating the root mean square error of the data verified by non-linear fitting is, the better;

[0130] The beneficial effects of the present invention on the prior art are as follows: The present invention pays attention to the important functions of low-voltage shunt capacitors in the power system, which are widely used in various industrial, commercial, and residential electricity applications to improve the power factor, stabilize the voltage, and reduce line losses. Compared with traditional remaining service life prediction methods, predicting the remaining service life of capacitors through artificial intelligence algorithms can effectively identify potential faults in advance, avoid outages and high maintenance costs caused by sudden equipment failures, thereby improving the operating stability and economic benefits of the power system. Accurate life prediction can not only optimize the maintenance and replacement plans of capacitors, reduce unnecessary resource waste, but also improve the operating efficiency and return on investment of equipment. With the continuous growth of power demand, ensuring the efficient operation of power equipment is crucial for ensuring sustainable economic development. The prediction method based on the WOA-CNN-BiLSTM neural network can provide more accurate predictions for the power system, help reduce the impact of sudden failures on production and life, thereby improving the overall reliability of the system and promoting the rational utilization of energy resources. Through this intelligent prediction and optimal scheduling, it not only helps to reduce social and economic costs, but also provides support for sustainable development goals and promotes the realization of a green and low-carbon economy. Specific Embodiment Twelve:

[0132] The difference between the twelfth embodiment and the eleventh embodiment of the present invention is only that:

[0133] The method of the present invention uses a low-voltage shunt capacitor cyclic aging test platform to obtain the change trend of the capacitance value of the low-voltage shunt capacitor under the continuous action of various influencing factors, and then analyzes the influence degree of each factor on it, and establishes the mapping relationship therein. Attached Figure 2 shows the structural diagram of the low-voltage shunt capacitor aging experiment platform. At the same time, the data obtained from the above experiments are prepared and input into the artificial intelligence algorithm for training and verification. During the implementation process, not only temperature, voltage, and current factors are strictly considered, but the number of cycles is also considered. Strictly controlling the experimental conditions is a necessary factor for the achievement of the present invention. Through experimental results and mathematical method verification, the experimental results predicted by the artificial intelligence algorithm adopted in this experiment are very accurate, and can effectively predict the remaining service life of low-voltage shunt capacitors, preventing national economic losses and risks threatening personal safety caused by untimely replacement and maintenance.

[0134] A method for predicting the remaining service life of low-voltage shunt capacitors based on artificial intelligence, as attached Figure 1 shown, specifically includes the following parts:

[0135] Step 1: Calculate the rated capacitance value based on the rated data of the nameplate of the low-voltage shunt capacitor used. According to the known working characteristics of the capacitor at different capacitance values, use the remaining capacitance value as the basis for judging the lifespan, thus forming the basic conditions of the lifespan prediction theory. The low-voltage shunt capacitor used in the present invention is the CLMD33 / 22.5-480-50 low-voltage shunt capacitor bank produced by ABB. The capacity of this low-voltage shunt capacitor is 22.5 kVar, the rated voltage is 480 V, the frequency is 50 Hz, and the rated capacitance value of the low-voltage shunt capacitor is approximately 311 uF. I is the compensation current value for rated operation. Here, calculate the single-phase value, and its value is approximately 46.875 A after calculation.

[0136] Step 2: Through the low-voltage shunt capacitor cyclic aging experiment test platform, by accelerating the cyclic charge and discharge times N, read and record the working parameters during the switching of the capacitor every time after time t. Based on the aging process of the equivalent capacitance value during operation and the remaining capacitance value data calculated from the voltage and current data at different times under different working conditions, there are three variables in this experiment: working voltage U, compensation current I, and working temperature °C. Set three different values for each variable, and the deviation does not exceed ±5% of the rated working value. When verifying the influence of one variable on the capacitance value, strictly control the consistency of other experimental variables. The following table specifically shows the experimental data table 1:

[0137] Table 1. Experimental data table

[0138]

[0139] Step 3: Form three groups of data sets from the experimental data obtained in Step 2 according to the above experimental groups, respectively verify the influence of different influencing factors on the remaining service life of the low-voltage shunt capacitor, and establish the mapping relationship between each influencing factor and the remaining capacitance value. Figure 3 、 4 、5 shows the mapping relationship between each factor and the remaining capacitance value;

[0140] Step 4: Merge the data sets obtained in Step 2 into a normalized data set, clean and normalize the data, randomly group the data set, and 80% of it is divided into the training set and 20% is divided into the test set. The following formula is for data standardization processing using data normalization and data denormalization:

[0141] x' scale =x p (x max -x min )+x min

[0142] Step 5: Construct a CNN model from the training set obtained in Step 4, extract local features from the data, and process them through convolutional layers and pooling layers. Then construct a BiLSTM model, use the BiLSTM network to capture the dependencies of the time series, and output the predicted value of the capacitor life. Then enter the fitness evaluation, and use the prediction error of the BiLSTM model as the effect of the fitness evaluation hyperparameters. Through position update, select the shrinking encirclement or spiral path for update according to the whale optimization algorithm. Continue iterative update until the predetermined conditions are met. The following formulas are some essential formulas for the whale optimization algorithm, BiLSTM model, and CNN model:

[0143]

[0144] Step 6: After the above steps are completed, use the output best BiLSTM network parameters and model to test the test set, and use RMSE and MAE to measure the accuracy of the experiment. The root mean square error (RMSE) and mean absolute error (MAE). MAE is used to calculate the average of the absolute values of the differences between the predicted values and the true values. RMSE illustrates the degree of dispersion of the samples. In this experiment, the smaller the RMSE obtained by calculating the root mean square error of the data verified by non-linear fitting is, the better. The following table shows the values of RMSE and MAE calculated from the verified data of this experiment, both of which meet the extremely high accuracy requirements;

[0145]

[0146] Step 7: After the system completes the above steps, deploy the best BiLSTM network parameters and model output in Step 5 to the actual system for real-time prediction of the remaining life of the capacitor, and complete the entire process. Attached Figure 6 、 7 shows the result graph and prediction result error graph of the prediction of the remaining service life of the low-voltage shunt capacitor based on the WOA-CNN-BiLSTM artificial intelligence algorithm in the present invention, both of which reach extremely high accuracy and can meet the requirements for application in the actual field to ensure the safe operation of the equipment.

[0147] In the description of this specification, the description referring to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples. Furthermore, the terms "first" and "second" are used only for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" can explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "N" is at least two, such as two, three, etc., unless otherwise specifically defined. Any process or method description represented in a flowchart or described in other ways herein can be understood to represent a module, segment, or part of code including one or more N executable instructions for implementing a customized logical function or process. The scope of the preferred embodiments of the present invention includes additional implementations, where the functions can be executed in a substantially simultaneous manner or in a reverse order according to the involved functions, rather than in the order shown or discussed, which should be understood by those skilled in the art of the embodiments of the present invention. The logic and / or steps represented in a flowchart or described in other ways herein, for example, can be considered as a sequenced list of executable instructions for implementing a logical function, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in combination with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion (electronic device) having one or N wirings, a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM).In addition, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation or other appropriate processing when necessary, and then stored in a computer memory. It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0148] Those of ordinary skill in the art can understand that all or part of the steps carried by the methods in the above embodiments can be completed by instructing relevant hardware through a program. The program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments. In addition, in each of the embodiments of the present invention, the functional units can be integrated into a processing module, or each unit can exist physically alone, or two or more units can be integrated into a module. The above integrated module can be implemented in the form of hardware or in the form of a software functional module. When the above integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0149] The above is only a preferred embodiment of a method for predicting the remaining service life of low-voltage shunt capacitors based on artificial intelligence. The protection scope of a method for predicting the remaining service life of low-voltage shunt capacitors based on artificial intelligence is not limited to the above embodiments. All technical solutions falling within this concept belong to the protection scope of the present invention. It should be noted that for those skilled in the art, several improvements and changes made without departing from the principle of the present invention should also be regarded as within the protection scope of the present invention.

Claims

1. A method for predicting the remaining service life of low-voltage shunt capacitors based on artificial intelligence, characterized in that: The method includes the following steps: Step 1: Calculate the rated capacitance value by using the rated data on the nameplate of the low-voltage shunt capacitor. Step 2: Based on the low-voltage shunt capacitor cyclic aging experiment test platform, by accelerating the cyclic charge and discharge times N, read and record the working parameters during the switching of the capacitor every time after a time t, to simulate the aging process of the equivalent capacitance value during operation and obtain the remaining capacitance value data calculated from the voltage and current data at different times under different working conditions. Step 3: Form three groups of data sets from the obtained data, respectively verify the influence of different influencing factors on the remaining service life of the low-voltage shunt capacitor, and establish the mapping relationship between each influencing factor and the remaining capacitance value. Step 4: Merge the obtained data sets into a normalized data set, clean and normalize the data, and divide it into a training set and a test set. Step 5: Build a CNN model from the training set, extract local features from the data, process them through the convolutional layer and the pooling layer, and then build a BiLSTM model. Use the BiLSTM network to capture the dependencies of the time series and output the predicted value of the capacitor life. Step 6: Use the best BiLSTM network parameters and model output for testing the test set, and use RMSE and MAE to measure the accuracy of the experiment. Step 7: Deploy the best BiLSTM network parameters and model output to the actual system for real-time prediction of the remaining life of the capacitor.

2. The method according to claim 1, characterized in that: Specifically, step 2 is as follows: Through the low-voltage shunt capacitor cyclic aging experiment test platform, the cyclic charge and discharge times N are equivalently replaced by the actual working time of the low-voltage shunt capacitor; in the cyclic aging test platform, the charging is provided by a constant current source, the discharging is carried out through an electronic load, and the relay is responsible for switching the charge and discharge states. When the constant current source is turned on, the low-voltage shunt capacitor is in the charging state; when the electronic load is turned on, discharging occurs. The controller realizes these switching operations through the connected relay. At the same time, the voltage detection module monitors the terminal voltage of the low-voltage shunt capacitor in real time and compares it with the preset charge and discharge cut-off voltages. The thermostatic chamber provides specific temperature conditions for the experiment. Through the above real-time data calculation, when the remaining capacitance value of the capacitor reaches 20% of its rated value, the experiment is stopped and the experimental cycle times N are recorded.

3. The method according to claim 2, wherein: Step 2 also includes: There are three variables: working voltage U, compensation current I, and working temperature °C. Set three different values for each variable, and the deviation does not exceed ±5% of the rated working value. When verifying the influence of one variable on the capacitance value, strict consistency is controlled.

4. The method according to claim 3, wherein: Merge the obtained data sets into a normalized data set, clean and normalize the data, randomly group the data set, and 80% of it is divided into the training set and 20% is divided into the test set.

5. The method according to claim 4, characterized in that: Specifically, step 5 is as follows: Build a CNN model from the training set obtained in step 4, extract local features from the data, process them through the convolutional layer and the pooling layer, and then build a BiLSTM model. Use the BiLSTM network to capture the dependencies of the time series and output the predicted value of the capacitor life. Enter the fitness evaluation, and use the prediction error of the BiLSTM model as the effect of the fitness evaluation hyperparameters; Through position update, select the shrinking encirclement or spiral path for update according to the whale optimization algorithm; Continue iterative update until the predetermined condition is reached.

6. The method according to claim 5, characterized in that: Step 5 also includes the continued iterative process, which adopts the early stopping method and cross-validation to prevent overfitting, and introduces the learning rate adaptive adjustment algorithm to dynamically adjust the learning rate during training in order to find the optimal solution; By continuously adjusting the algorithm iteration, the number of training times, the number of searches, and the number of hidden layers in the convolutional neural network, the optimal parameters of the BiLSTM are obtained.

7. The method according to claim 4, wherein: Use the root mean square error RMSE and the mean absolute error MAE to measure the accuracy. The MAE is used to calculate the average of the absolute values of the differences between the predicted value and the true value. The RMSE determines the degree of dispersion of the sample. The smaller the RMSE obtained by calculating the root mean square error of the data verified by non-linear fitting of the data, the better.

8. A remaining service life prediction system for low-voltage shunt capacitors based on artificial intelligence, characterized in that: The system includes: A calculation module, which calculates the rated capacitance value by using the rated data of the nameplate of the low-voltage shunt capacitor; An experimental module, which is based on the low-voltage shunt capacitor cyclic aging experimental test platform. By accelerating the cyclic charge and discharge times N, the working parameters during the switching of the capacitor are read and recorded every time t, and the remaining capacitance value data calculated from the equivalent capacitance value during the aging process during operation and the voltage and current data at different times under different working conditions are obtained; A data module, which forms three data sets from the obtained data to respectively verify the influence of different influencing factors on the remaining service life of the low-voltage shunt capacitor, and establish the mapping relationship between each influencing factor and the remaining capacitance value; An integration module, which combines the obtained data sets into a normalized data set, cleans and normalizes the data, and divides it into a training set and a test set; A model training module, which constructs a CNN model from the training set, extracts local features from the data, processes them through the convolutional layer and the pooling layer, then constructs a BiLSTM model, and uses the BiLSTM network to capture the dependencies of the time series and output the predicted value of the capacitor life; A test module, which tests the output optimal BiLSTM network parameters and the model on the test set, and uses RMSE and MAE to measure the accuracy of the experiment; A verification module, which deploys the output optimal BiLSTM network parameters and the model to the actual system for real-time prediction of the remaining life of the capacitor.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by a processor to implement the method as claimed in claims 1-7.

10. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the method as claimed in claims 1-7.

Images