Method and device for predicting sulfur hexafluoride gas pressure of composite apparatus

Abstract

The embodiment of the invention provides a combined electric appliance sulfur hexafluoride gas pressure prediction method and device, and the method comprises the steps: inputting to-be-measured data into a pre-trained Transform gas prediction model, and obtaining a corresponding prediction result, the Transform gas prediction model is obtained by inputting time sequence data of sulfur hexafluoride gas pressure within a preset time length into the Transform gas prediction model for training and performing parameter optimization through a flower pollination algorithm; under the condition that the error between the prediction result and the actual test set result does not meet the error requirement, repeating the pre-trained Transformer gas prediction model to obtain the corresponding prediction result, and under the condition that the error between the prediction result and the actual test set result meets the error requirement, recording the prediction result; according to the method, the whole sequence can be processed in parallel by utilizing the self-attention mechanism of Transform, the training efficiency is remarkably improved, meanwhile, the optimal parameter configuration of the model can be quickly found by adopting the flower pollination algorithm, and the prediction precision and the model performance are improved.

Description

Technical Field

[0001] The present application relates to the field of power equipment detection, and particularly to a method and device for predicting the sulfur hexafluoride gas pressure of a combined electrical apparatus. Background Art

[0002] Sulfur hexafluoride (SF6) gas has been widely used in the power system due to its excellent insulation and arc extinguishing performance, especially in gas-insulated metal-enclosed switchgear (GIS). GIS uses SF6 gas as the insulating medium and seals all high-voltage electrical components in a grounded metal enclosure. However, SF6 gas will decompose into various toxic gases under the influence of factors such as arc, corona, partial discharge, and high temperature, posing a serious threat to the operation and maintenance personnel and equipment safety of GIS. Therefore, accurately predicting the SF6 gas pressure to facilitate timely maintenance measures is of great significance for the safe operation of the power system.

[0003] Research on time-series data in the power system has a history of decades, mainly focusing on the fields of load forecasting and electricity price forecasting. However, there is less research on SF6 gas pressure prediction. Although existing methods have made some progress in certain aspects, there are still deficiencies. For example, some studies have used a combined model of convolutional neural network (CNN) and bidirectional long short-term memory network (BiLSTM) to improve the accuracy of short-term load forecasting; some studies have also used a convolutional neural network to extract load features and combined with a gated recurrent unit (GRU) for prediction, showing better performance than traditional statistical methods. In addition, short-term load forecasting of ship power grids based on artificial neural networks and improved BP neural network prediction methods have also achieved high prediction accuracy in specific application scenarios. For SF6 gas pressure prediction, some studies have tried to use deep learning methods such as long short-term memory network (LSTM), but the effect is still limited.

[0004] Due to the many deficiencies of the above methods, although LSTM and GRU have to some extent solved the problem of long sequence dependence by introducing a memory mechanism, they still face the problem of gradient disappearance or gradient explosion when dealing with ultra-long sequences. In addition, their sequential calculation characteristics limit the parallelization ability, resulting in a slow model training speed, especially inefficient when dealing with large-scale time-series data. And the traditional BP neural network is more suitable for dealing with static data, with weak ability to capture time dependence in time-series data and difficult to meet the requirements of complex time-series prediction tasks. Summary of the Invention

[0005] In view of the problems in the prior art, the present application provides a method and device for predicting the sulfur hexafluoride gas pressure of a combined electrical apparatus, which can utilize the self-attention mechanism of Transformer to process the entire sequence in parallel, thereby significantly improving the training efficiency. At the same time, the flower pollination algorithm is used to quickly find the optimal parameter configuration of the Transformer gas prediction model, improving the prediction accuracy and model performance.

[0006] To solve at least one of the above problems, the present application provides the following technical solutions:

[0007] According to the first aspect of the embodiments of the present application, the present application provides a method for predicting the sulfur hexafluoride gas pressure of a combined electrical apparatus, including:

[0008] Input the data to be measured into a pre-trained Transformer gas prediction model and obtain the corresponding prediction result. Among them, the Transformer gas prediction model is obtained by processing the time series data of the sulfur hexafluoride (SF6) gas pressure within a preset time period, dividing it into a training set and a test set, inputting it into the Transformer gas prediction model for training, and optimizing the parameters through the flower pollination algorithm;

[0009] In the case where the error between the prediction result and the actual test set result does not meet the error requirement, repeat to obtain the corresponding prediction result based on the pre-trained Transformer gas prediction model. In the case where the error between the prediction result and the actual test set result meets the error requirement, record the prediction result.

[0010] According to any implementation manner of the present application, the training process of the Transformer gas prediction model includes:

[0011] Obtain the time series data of the SF6 gas pressure in a preset time period through the SF6 online monitoring device;

[0012] Process and normalize the missing data, duplicate data, abnormal data in the time series data of the SF6 gas pressure;

[0013] Divide the processed data into a training set and a test set and input it into the Transformer gas prediction model for training to obtain the Transformer gas prediction model.

[0014] According to any implementation manner of the present application, the processing and normalization of the missing data, duplicate data, and abnormal data in the time series data of the SF6 gas pressure include:

[0015] Adopt the adjacent value filling method to process the missing data, and replace the missing value with the average value of the data at the adjacent positions before and after. Among them, the abnormal numbers outside the preset range are regarded as the missing data;

[0016] Delete the duplicate data existing in the sequence numbers.

[0017] Compress the time series data to a preset normal distribution range and convert it into a dimensionless form.

[0018] According to any embodiment of the present application, the process of dividing the processed data into a training set and a test set and inputting them into the Transformer gas prediction model for training to obtain the Transformer gas prediction model includes:

[0019] Convert the time series data into embedding vectors of a fixed dimension;

[0020] Map each embedding vector element in the time series data into a one-dimensional vector containing position information to add position encoding to the embedding vector;

[0021] Based on the self-attention mechanism, generate a comprehensive representation vector reflecting the relationship between elements by calculating the weights of each element in the sequence with other elements;

[0022] Perform multiple self-attention operations based on multiple stacked self-attention layers;

[0023] Combine the target sequence according to the representation vector, and generate a corresponding prediction result by using the self-attention mechanism and the interaction mechanism;

[0024] Optimize the parameters of the Transformer gas prediction model through the flower pollination algorithm and save the optimized model parameters to obtain the Transformer gas prediction model.

[0025] According to any embodiment of the present application, the errors between the prediction result and the actual test set result include the mean absolute percentage error and the root mean square error.

[0026] According to the second aspect of the embodiments of the present application, the present application provides a sulfur hexafluoride gas pressure prediction device for a combined electrical apparatus, including:

[0027] A data input module, configured to: input the data to be measured into a pre-trained Transformer gas prediction model and obtain a corresponding prediction result, where the Transformer gas prediction model is obtained by processing the time series data of the sulfur hexafluoride (SF6) gas pressure within a preset time period, dividing it into a training set and a test set, inputting it into the Transformer gas prediction model for training, and optimizing the parameters through the flower pollination algorithm;

[0028] A prediction output module, configured to: when the error between the prediction result and the actual test set result does not meet the error requirement, repeat to obtain the corresponding prediction result based on the pre-trained Transformer gas prediction model, and when the error between the prediction result and the actual test set result meets the error requirement, record the prediction result.

[0029] According to any implementation manner of the present application, the training process of the Transformer gas prediction model includes:

[0030] A data acquisition module, configured to: obtain SF6 gas pressure time series data in a preset time period through an SF6 on-line monitoring device;

[0031] A data processing module, configured to: perform missing data, duplicate data, abnormal data processing and normalization processing on the SF6 gas pressure time series data;

[0032] A model construction module, configured to: divide the processed data into a training set and a test set and input them into the Transformer gas prediction model for training to obtain the Transformer gas prediction model.

[0033] According to any implementation manner of the present application, the data processing module includes:

[0034] A missing and abnormal processing unit, configured to: process missing data by using a neighboring value filling device, and replace the missing value with the average value of the data at the adjacent positions before and after, wherein the abnormal numbers outside the preset range are regarded as the missing data;

[0035] A duplicate value processing unit, configured to: delete the duplicate data existing in the time series data;

[0036] A compression unit, configured to: compress the time series data to a preset normal distribution range and convert it into a dimensionless form.

[0037] According to any implementation manner of the present application, the model construction module includes:

[0038] An embedding vector generation unit: convert the time series data into an embedding vector with a fixed dimension;

[0039] A position encoding unit: map each embedding vector element in the time series data into a one-dimensional vector containing position information to add position encoding to the embedding vector;

[0040] A self-attention mechanism unit: based on the self-attention mechanism, generate a comprehensive representation vector reflecting the relationship between elements by calculating the weights of each element in the sequence and other elements;

[0041] Multi-layer self-attention stacking unit: Perform multiple self-attention operations based on multiple stacked self-attention layers;

[0042] Decoding and prediction unit: Combine the target sequence according to the representation vector, and generate corresponding prediction results by using the self-attention mechanism and the interaction mechanism;

[0043] Parameter optimization and saving unit: Optimize the parameters of the Transformer gas prediction model through the flower pollination algorithm, and save the optimized model parameters to obtain the Transformer gas prediction model.

[0044] According to any embodiment of the present application, the errors between the prediction results and the actual test set results include the mean absolute percentage error and the root mean square error.

[0045] According to the third aspect of the embodiments of the present application, the present application provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps of the combined electrical apparatus sulfur hexafluoride gas pressure prediction method are implemented.

[0046] According to the fourth aspect of the embodiments of the present application, the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the combined electrical apparatus sulfur hexafluoride gas pressure prediction method are implemented.

[0047] According to the fifth aspect of the embodiments of the present application, the present application provides a computer program product, including a computer program / instructions. When the computer program / instructions are executed by a processor, the steps of the combined electrical apparatus sulfur hexafluoride gas pressure prediction method are implemented.

[0048] It can be seen from the above technical solutions that the present application provides a combined electrical apparatus sulfur hexafluoride gas pressure prediction method and device. By inputting the data to be measured into the pre-trained Transformer gas prediction model to obtain corresponding prediction results, wherein the Transformer gas prediction model is obtained by training the time series data of the sulfur hexafluoride gas pressure within a preset time period input into the Transformer gas prediction model and optimizing the parameters through the flower pollination algorithm; in the case where the error between the prediction result and the actual test set result does not meet the error requirement, repeat to obtain the corresponding prediction result based on the pre-trained Transformer gas prediction model. In the case where the error between the prediction result and the actual test set result meets the error requirement, record the prediction result. It can parallelly process the entire sequence by using the self-attention mechanism of Transformer, significantly improve the training efficiency, and at the same time, the flower pollination algorithm can quickly find the optimal parameter configuration of the model, improving the prediction accuracy and model performance. Brief Description of the Drawings

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0050] Figure 1 One of the flow diagrams of the method for predicting the sulfur hexafluoride gas pressure of the gas-insulated switchgear in the embodiments of the present application;

[0051] Figure 2 One of the flow diagrams of the method for predicting the sulfur hexafluoride gas pressure of the gas-insulated switchgear in the embodiments of the present application;

[0052] Figure 3 One of the flow diagrams of the method for predicting the sulfur hexafluoride gas pressure of the gas-insulated switchgear in the embodiments of the present application;

[0053] Figure 4 One of the flow diagrams of the method for predicting the sulfur hexafluoride gas pressure of the gas-insulated switchgear in the embodiments of the present application;

[0054] Figure 5 The flow chart of the flower pollination algorithm in the embodiments of the present application;

[0055] Figure 6 The Transformer framework based on flower pollination in the embodiments of the present application;

[0056] Figure 7 The convergence curve graph after parameter tuning of the Transformer gas prediction model in the embodiments of the present application;

[0057] Figure 8 The prediction effect graph of the Transformer gas prediction model in the embodiments of the present application;

[0058] Figure 9 One of the structural diagrams of the device for predicting the sulfur hexafluoride gas pressure of the gas-insulated switchgear in the embodiments of the present application;

[0059] Figure 10 One of the structural diagrams of the device for predicting the sulfur hexafluoride gas pressure of the gas-insulated switchgear in the embodiments of the present application;

[0060] Figure 11 One of the structural diagrams of the device for predicting the sulfur hexafluoride gas pressure of the gas-insulated switchgear in the embodiments of the present application;

[0061] Figure 12This is the fourth structural diagram of the sulfur hexafluoride gas pressure prediction device for the combined electrical apparatus in the embodiments of the present application;

[0062] Figure 13 This is the structural schematic diagram of the electronic device in the embodiments of the present application. Detailed implementation manners

[0063] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Apparently, the described embodiments are some but not all of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the scope of protection of the present application.

[0064] In the technical solutions of the present application, the acquisition, storage, use, processing, etc. of data all comply with the relevant regulations of national laws and regulations.

[0065] The present application provides a method and device for predicting the sulfur hexafluoride gas pressure of a combined electrical apparatus. By using the self-attention mechanism of Transformer to process the entire sequence in parallel, the training efficiency is significantly improved. At the same time, the flower pollination algorithm is used to quickly find the optimal parameter configuration of the model, improving the prediction accuracy and model performance.

[0066] In order to be able to use the self-attention mechanism of Transformer to process the entire sequence in parallel, significantly improve the training efficiency, and at the same time use the flower pollination algorithm to quickly find the optimal parameter configuration of the model, improving the prediction accuracy and model performance, the present application provides an embodiment of a method for predicting the sulfur hexafluoride gas pressure of a combined electrical apparatus. Refer to Figure 1 , the method for predicting the sulfur hexafluoride gas pressure of the combined electrical apparatus specifically includes the following content:

[0067] Step S101: Input the data to be measured into the pre-trained Transformer gas prediction model and obtain the corresponding prediction result. Among them, the Transformer gas prediction model is obtained by processing the time-series data of the sulfur hexafluoride (SF6) gas pressure within a preset time period, dividing it into a training set and a test set, inputting it into the Transformer gas prediction model for training, and optimizing the parameters through the flower pollination algorithm;

[0068] Step S102: In the case where the error between the prediction result and the actual test set result does not meet the error requirement, repeat to obtain the corresponding prediction result based on the pre-trained Transformer gas prediction model. In the case where the error between the prediction result and the actual test set result meets the error requirement, record the prediction result.

[0069] Among them, in this application, the data to be measured is first input into a pre-trained Transformer gas prediction model, enabling the model to perform feature extraction and relationship modeling on the input time series data based on its self-attention mechanism, and generating corresponding pressure prediction results. The Transformer gas prediction model can make full use of the time dependence and sequence features of the data, thereby achieving high-precision prediction of complex time series data. The prediction result is the gas pressure value calculated by the model according to the input data and the optimized parameters, and is provided as the output.

[0070] The generation of the Transformer gas prediction model is based on the processing of the time series data of SF6 gas pressure within a preset time period. The time series data first undergoes a series of preprocessing operations, including missing value filling, duplicate value deletion, outlier handling, and normalization, etc., to ensure the integrity, consistency, and applicability of the data. Subsequently, the processed data is divided into a training set and a test set, and the training set is input into the Transformer gas prediction model for multiple iterative trainings. At the same time, to improve the prediction accuracy and performance of the model, the flower pollination algorithm is used to optimize the hyperparameters of the Transformer gas prediction model, quickly finding the optimal parameter combination of the model, so that the model can achieve a high accuracy while maintaining high efficiency.

[0071] After obtaining the prediction result, its error is compared with the actual test set result. When the prediction error does not meet the preset requirements, the model will adjust the parameters or structure according to the error analysis and retrain until the error meets the requirements. Finally, the prediction result when the error meets the requirements is recorded as the final output of this prediction task. This process of iterative optimization and result verification ensures that the method has high prediction accuracy and stability, and can meet the strict requirements of the power system for SF6 gas pressure prediction.

[0072] From the above description, it can be seen that the method for predicting the SF6 gas pressure of a combined electrical apparatus provided by the embodiments of this application can use the self-attention mechanism of the Transformer to process the entire sequence in parallel, significantly improving the training efficiency. At the same time, the flower pollination algorithm can quickly find the best parameter configuration of the model, improving the prediction accuracy and model performance.

[0073] In an embodiment of the method for predicting the SF6 gas pressure of a combined electrical apparatus in this application, refer to Figure 2 , the training process of the Transformer gas prediction model includes:

[0074] Step S001: Obtain the time series data of SF6 gas pressure in a preset time period through the SF6 online monitoring device;

[0075] Step S002: Process the missing data, duplicate data, abnormal data and normalize the SF6 gas pressure time series data;

[0076] Step S003: Divide the processed data into a training set and a test set and input them into the Transformer gas prediction model for training to obtain the Transformer gas prediction model.

[0077] Exemplarily, the training process of the Transformer gas prediction model includes multiple stages to ensure that the model can accurately and efficiently predict the sulfur hexafluoride (SF6) gas pressure and meet the requirements of practical applications. The specific process is as follows:

[0078] First, collect the time series data of the SF6 gas pressure in the past year through the SF6 online monitoring device, which includes the gas pressure values at different time nodes and has time correlation.

[0079] Next, preprocess the collected SF6 gas pressure data to improve the integrity and reliability of the data.

[0080] Exemplarily, first, process the missing values in the data. Adopt the adjacent value filling method and replace the missing values with the mean value of the data at the adjacent positions before and after; secondly, delete the duplicate data to avoid the interference of redundant information on the model training; finally, identify and process the abnormal data. By artificially setting a reasonable range, the data beyond the range is regarded as missing values and filled.

[0081] After ensuring the data quality, further normalize the data. Among them, the purpose of normalization is to narrow the range of data feature values, make the data distribution closer to the normal distribution, thereby improving the stability of model training and accelerating the convergence speed. In addition, normalization also eliminates the dimension difference between data dimensions, enabling the model to more fairly weigh the influence of each feature.

[0082] After completing the normalization, divide the processed data into a training set and a test set. The training set is used for the parameter learning of the model, and the test set is used to evaluate the model performance.

[0083] Finally, input the training set into the Transformer gas prediction model for training. The model optimizes its internal parameters through multiple iterations of learning, and can then perform high-precision prediction on the SF6 gas pressure.

[0084] In an embodiment of the sulfur hexafluoride gas pressure prediction method for the combined electrical apparatus in the present application, refer to Figure 3 , the processing of the missing data, duplicate data, abnormal data and normalization of the SF6 gas pressure time series data includes:

[0085] Step S002A: Process the missing data using the adjacent value filling method, and replace the missing values with the average of the data at the adjacent positions before and after. Among them, the abnormal numbers outside the preset range are regarded as the missing data;

[0086] Step S002B: Delete the duplicate data existing in the time series numbers;

[0087] Step S002C: Compress the time series data to a preset normal distribution range and convert it into a dimensionless form.

[0088] Specifically, to improve the stability of model training, the standardization method is used to compress the data to make it close to the normal distribution, and at the same time eliminate the influence of the dimension. The specific conversion formula is as follows:

[0089]

[0090] Among them, x * is the normalized data, x is the actual value of the historical SF6 gas pressure, μ is the average value of the historical SF6 gas pressure, N represents the total number of SF6 gas pressure samples in the historical data, and σ is the standard deviation of the historical SF6 gas pressure.

[0091] In an embodiment of the method for predicting the SF6 gas pressure of the gas-insulated switchgear in the present application, see Figure 4 , the process of dividing the processed data into a training set and a test set and inputting them into the Transformer gas prediction model for training to obtain the Transformer gas prediction model includes:

[0092] Step S003A: Convert the time series data into an embedding vector with a fixed dimension;

[0093] Step S003B: Map each embedding vector element in the time series data into a one-dimensional vector containing position information to add position encoding to the embedding vector;

[0094] Step S003C: Based on the self-attention mechanism, generate a comprehensive representation vector reflecting the relationship between elements by calculating the weights of each element in the sequence with other elements;

[0095] Step S003D: Perform multiple self-attention operations based on multiple stacked self-attention layers;

[0096] Step S003E: Combine the representation vector with the target sequence and generate a corresponding prediction result using the self-attention mechanism and the interaction mechanism;

[0097] Step S003F: Optimize the parameters of the Transformer gas prediction model through the flower pollination algorithm, and save the optimized model parameters to obtain the Transformer gas prediction model.

[0098] Among them, in order to obtain an accurate SF6 gas pressure prediction model, the entire process from data encoding to parameter optimization is closely combined and progresses step by step, finally forming a high-precision Transformer gas prediction model.

[0099] First, through the embedding encoding technique, the pre-normalized training time series data is converted into a vector representation of a fixed dimension. These vectors express different features of the time series data, enabling the model to extract useful information from high-dimensional data, establish a comprehensive understanding of the input data, and lay a foundation for subsequent modeling.

[0100] After completing the embedding encoding, the position encoding is used to enhance the model's understanding of the element order in the sequence data. Specifically, the position encoding generates a vector with position information for each element in the sequence, compensating for the lack of order information in the embedding vector, enabling the Transformer gas prediction model to more effectively capture the characteristics of the time order in the data, thereby enhancing the modeling ability for time series data.

[0101] In order to further extract complex features in the data, the Transformer gas prediction model of this application also uses the self-attention mechanism to compare each sequence element with other elements and calculate its correlation weight. Through the weighted average of these weights, the model generates a comprehensive representation vector reflecting the relationship between the current element and other elements, allowing the model to simultaneously focus on multiple position elements in the sequence to capture potential global relationships in the data and improve the understanding ability of complex time series relationships.

[0102] Based on the above feature extraction, the Transformer uses multiple layers of self-attention stacking to further enrich the time series features. Each layer performs an attention operation on the sequence data, gradually extracting higher-order sequence information, and the design of the multi-layer structure enables the model to gradually expand from simple feature extraction to complex feature combinations, finally forming a deep understanding of the time series data.

[0103] Finally, the decoder combines the representation vector from the encoder with the target sequence and generates a prediction result through the self-attention mechanism and the interaction mechanism, ensuring the accuracy and rationality of the prediction result.

[0104] In addition, to further improve the model performance, the flower pollination algorithm is used to optimize the parameters of the Transformer gas prediction model. By simulating the pollen propagation behavior in nature, the flower pollination algorithm explores the optimal parameter configuration globally and combines local optimization to ensure the convergence speed, which not only improves the prediction accuracy of the model but also significantly enhances the training efficiency and model stability.

[0105] In an optional embodiment, the error between the prediction result and the actual test set result includes the mean absolute percentage error and the root mean square error.

[0106] Specifically, the prediction effect of the test set can be evaluated by two indicators: the mean absolute percentage error (MAPE) and the root mean square error (RMSE). The formulas are as follows:

[0107]

[0108]

[0109] where represents the predicted value of the test set, n represents the number of test samples, and y i represents the SF6 gas pressure of the i-th sample in the test set.

[0110] Among them, MAPE reflects the relative error of the predicted value relative to the actual value, which is convenient for understanding the actual application effect of the model; RMSE is used to measure the absolute error between the predicted value and the actual value, and can quantify the overall error level of the model.

[0111] To further illustrate the present solution, the present application also provides a specific application example of a method for predicting the sulfur hexafluoride gas pressure of a combined electrical appliance by using the above-mentioned sulfur hexafluoride gas pressure prediction device for combined electrical appliances, which specifically includes the following contents:

[0112] Figure 5 A flowchart of the flower pollination algorithm is shown, which specifically includes the following steps:

[0113] Initializing the population: First, a group of solutions (population) are randomly initialized for the parameter space of the Transformer gas prediction model, and each solution represents a set of parameter configurations of the Transformer gas prediction model.

[0114] Calculating the fitness value and recording the current optimal solution: By training the model and evaluating performance metrics (such as prediction error), calculate the fitness value (performance quality) of each solution and record the current optimal solution.

[0115] Parameter setting: Set the self-pollination probability parameter ppp of the flower pollination algorithm to control the ratio of subsequent self-pollination and cross-pollination.

[0116] Determine the pollen transmission type: In each iteration, generate a random number rrr. If r < p, perform self-pollination; otherwise, perform cross-pollination.

[0117] Self-pollination: Simulate the transmission of pollen on the same plant. Based on the current solution, search for new solutions within its neighborhood to explore local optima.

[0118] Cross-pollination: Simulate the transmission of pollen between different plants. Guided by the global optimal solution, update the global characteristics of the solution to explore the global optimum of the parameter space.

[0119] Evaluate the updated solution: Evaluate the newly generated solution, calculate its fitness value, and determine whether the new solution is better than the original solution.

[0120] Update the local optimal solution: According to the fitness value of the new solution, update the optimal solution of the current population and the global optimal solution.

[0121] Judge the termination condition: Check whether the algorithm meets the termination condition (such as the number of iterations reaches the upper limit or the performance improvement of the optimal solution is less than the threshold). If it is met, stop the iteration; otherwise, return to continue performing self-pollination and cross-pollination.

[0122] Output the optimal parameter configuration: The algorithm finally outputs the optimized parameter configuration of the Transformer gas prediction model for model training and prediction.

[0123] Through the above process, the flower pollination algorithm can quickly and efficiently search the parameter space, find the optimal parameter configuration, and improve the prediction accuracy and overall performance of the Transformer gas prediction model. The combination of the global exploration of cross-pollination and the local optimization of self-pollination enables the algorithm to balance global convergence and local search ability.

[0124] Furthermore, Figure 6 A Transformer framework based on flower pollination is shown, which specifically includes the following steps:

[0125] SF6 gas pressure data acquisition: Obtain the time series data of historical gas pressure from the SF6 online monitoring device to provide basic data for model training and testing. These data include gas pressure values under the time series to capture time correlations.

[0126] Data preprocessing: Preprocess the collected raw data, including handling missing values, duplicate values, and outliers, and normalize the data to ensure data integrity, reliability, and standardization. This step lays a data foundation for the efficient training and accurate prediction of the model.

[0127] Dataset Division: The preprocessed data is divided into a training set and a test set. The training set is used for optimizing and learning model parameters, and the test set is used for evaluating model performance.

[0128] Training of the Transformer Gas Prediction Model: The training set is input into the Transformer gas prediction model. Through its functions such as embedding encoding, positional encoding, self-attention mechanism, and multi-layer stacking, complex features in the time-series data are extracted. The structure of the Transformer gas prediction model can efficiently capture the dependencies in the time-series data and generate an initial prediction model.

[0129] Flower Pollination Algorithm Optimization: During the model training process, the parameters of the Transformer gas prediction model are optimized in combination with the flower pollination algorithm. The flower pollination algorithm combines global exploration and local optimization to quickly find the optimal hyperparameter configuration, improving the training efficiency and prediction accuracy of the model.

[0130] Generation of the Optimal Transformer Gas Prediction Model: After being optimized by the flower pollination algorithm, the optimal Transformer gas prediction model is generated. The parameters of this model have been strictly optimized and verified, and it can perform well on the test set.

[0131] Generation of Prediction Results: The data to be measured is input into the optimized Transformer gas prediction model to obtain the prediction results of the SF6 gas pressure. These results have been verified by the test set and have high accuracy and reliability.

[0132] Figure 7 Characterize the change of the training loss (TrainingLoss) of the Transformer gas prediction model with the number of training epochs (Epoch) after parameter optimization, that is, the convergence curve. The curve shows that the loss value drops rapidly in the initial stage (the first 50 Epochs). As the training progresses, the loss value gradually stabilizes and converges to a level close to 0 when approaching 300 Epochs, reaching a relatively good performance state.

[0133] This convergence curve reflects that the Transformer gas prediction model optimized by the flower pollination algorithm has good learning ability and convergence, can effectively reduce prediction errors, avoid overfitting or underfitting problems, and ultimately ensure that the model has high prediction accuracy and stability when processing SF6 gas pressure time-series data.

[0134] Figure 8 Characterize the actual effect of the Transformer gas prediction model on predicting the SF6 gas pressure, where the blue curve represents the actual value and the red curve represents the predicted value of the model. Figure 8It clearly reflects that the Transformer gas prediction model based on flower pollination optimization has strong time series feature extraction ability and good prediction performance, can meet the application requirements of SF6 gas pressure prediction, and at the same time provides reliable reference data support for the operation and maintenance of GIS equipment.

[0135] In order to be able to use the self-attention mechanism of Transformer to process the entire sequence in parallel, significantly improve the training efficiency, and at the same time use the flower pollination algorithm to quickly find the best parameter configuration of the model, improve the prediction accuracy and model performance, this application provides an embodiment of a combined electrical apparatus SF6 gas pressure prediction device for implementing all or part of the content of the combined electrical apparatus SF6 gas pressure prediction method, see Figure 9 , the combined electrical apparatus SF6 gas pressure prediction device specifically includes the following contents:

[0136] The data input module 1101 is used for: inputting the data to be measured into the pre-trained Transformer gas prediction model and obtaining the corresponding prediction result, wherein the Transformer gas prediction model is obtained by processing the time series data of SF6 gas pressure within a preset time period, dividing it into a training set and a test set, inputting it into the Transformer gas prediction model for training, and optimizing the parameters through the flower pollination algorithm;

[0137] The prediction output module 1102 is used for: in the case where the error between the prediction result and the actual test set result does not meet the error requirement, repeating to obtain the corresponding prediction result based on the pre-trained Transformer gas prediction model, and in the case where the error between the prediction result and the actual test set result meets the error requirement, recording the prediction result.

[0138] According to any implementation manner of this application, see Figure 10 , the training process of the Transformer gas prediction model includes:

[0139] The data acquisition module 1001 is used for: acquiring the time series data of SF6 gas pressure in a preset time period through the SF6 online monitoring device;

[0140] The data processing module 1002 is used for: processing the missing data, repeated data and abnormal data of the SF6 gas pressure time series data and performing normalization processing;

[0141] The model construction module 1003 is used for: dividing the processed data into a training set and a test set and inputting them into the Transformer gas prediction model for training to obtain the Transformer gas prediction model.

[0142] According to any embodiment of the present application, refer to Figure 11 , the data processing module includes:

[0143] A missing anomaly processing unit 1002A, configured to: process missing data by using a neighboring value filling device, and replace the missing value with the average value of the data at the adjacent positions before and after, wherein the abnormal numbers outside the preset range are regarded as the missing data;

[0144] A duplicate value processing unit 1002B, configured to: delete the duplicate data existing in the time series numbers;

[0145] A compression unit 1002C, configured to: compress the time series data to a preset normal distribution range and convert it into a dimensionless form.

[0146] According to any embodiment of the present application, refer to Figure 12 , the model construction module includes:

[0147] An embedding vector generation unit 1003A: convert the time series data into an embedding vector with a fixed dimension;

[0148] A position encoding unit 1003B: map each embedding vector element in the time series data into a one-dimensional vector containing position information to add position encoding to the embedding vector;

[0149] A self-attention mechanism unit 1003C: based on the self-attention mechanism, generate a comprehensive representation vector reflecting the relationship between elements by calculating the weights of each element in the sequence and other elements;

[0150] A multi-layer self-attention stacking unit 1003D: perform multiple self-attention operations based on multiple stacked self-attention layers;

[0151] A decoding and prediction unit 1003E: combine the representation vector with the target sequence and generate a corresponding prediction result by using the self-attention mechanism and the interaction mechanism;

[0152] A parameter optimization and saving unit 1003F: optimize the parameters of the Transformer gas prediction model by using the flower pollination algorithm, and save the optimized model parameters to obtain the Transformer gas prediction model.

[0153] According to any embodiment of the present application, the errors between the prediction result and the actual test set result include the mean absolute percentage error and the root mean square error.

[0154] As can be seen from the above description, the sulfur hexafluoride gas pressure prediction device for gas-insulated switchgear provided by the embodiments of the present application can utilize the self-attention mechanism of Transformer to process the entire sequence in parallel, significantly improving the training efficiency. At the same time, the flower pollination algorithm is used to quickly find the optimal parameter configuration of the model, enhancing the prediction accuracy and model performance.

[0155] From a hardware perspective, in order to utilize the self-attention mechanism of Transformer to process the entire sequence in parallel, significantly improving the training efficiency, and at the same time using the flower pollination algorithm to quickly find the optimal parameter configuration of the model, enhancing the prediction accuracy and model performance, the present application provides an embodiment of an electronic device for implementing all or part of the content in the sulfur hexafluoride gas pressure prediction method for gas-insulated switchgear. The electronic device specifically includes the following:

[0156] A processor, a memory, a communications interface, and a bus; wherein, the processor, the memory, and the communications interface complete communication with each other through the bus; the communications interface is used to implement information transmission between the sulfur hexafluoride gas pressure prediction device for gas-insulated switchgear and related devices such as the core business system, the user terminal, and the relevant database. The logic controller can be a desktop computer, a tablet computer, a mobile terminal, etc., and this embodiment is not limited thereto. In this embodiment, the logic controller can be implemented with reference to the embodiments of the sulfur hexafluoride gas pressure prediction method for gas-insulated switchgear and the embodiments of the sulfur hexafluoride gas pressure prediction device for gas-insulated switchgear, and the content thereof is incorporated herein, and the repeated parts will not be elaborated.

[0157] It can be understood that the user terminal may include a smart phone, a tablet electronic device, an Internet set-top box, a portable computer, a desktop computer, a personal digital assistant (PDA), a vehicle-mounted device, a smart wearable device, etc. Among them, the smart wearable device may include smart glasses, a smart watch, a smart bracelet, etc.

[0158] In practical applications, part of the sulfur hexafluoride gas pressure prediction method for gas-insulated switchgear can be executed on the electronic device side as described above, or all operations can be completed in the client device. Specifically, it can be selected according to the processing capacity of the client device and the limitations of the user usage scenario, etc. The present application does not make any limitations in this regard. If all operations are completed in the client device, the client device may further include a processor.

[0159] The above-mentioned client device may have a communication module (i.e., communication unit), which can communicate with a remote server to achieve data transmission with the server. The server may include a server on the side of the task scheduling center, and in other implementation scenarios, it may also include a server of an intermediate platform, such as a server of a third-party server platform with a communication link to the task scheduling center server. The server may include a single computer device, or a server cluster composed of multiple servers, or a server structure of a distributed device.

[0160] Figure 13 It is a schematic block diagram of the system composition of the electronic device 9600 according to an embodiment of the present application. As Figure 13 shown, the electronic device 9600 may include a central processing unit 9100 and a memory 9140; the memory 9140 is coupled to the central processing unit 9100. It should be noted that this Figure 13 is exemplary; other types of structures may also be used to supplement or replace this structure to achieve telecommunication functions or other functions.

[0161] In one embodiment, the function of the sulfur hexafluoride gas pressure prediction method for combined electrical appliances can be integrated into the central processing unit 9100. Among them, the central processing unit 9100 may be configured to perform the following controls:

[0162] Step S101: Input the data to be measured into a pre-trained Transformer gas prediction model and obtain the corresponding prediction result. Among them, the Transformer gas prediction model is obtained by processing the time series data of the sulfur hexafluoride (SF6) gas pressure within a preset time period, dividing it into a training set and a test set, inputting it into the Transformer gas prediction model for training, and optimizing the parameters through the flower pollination algorithm;

[0163] Step S102: In the case where the error between the prediction result and the actual test set result does not meet the error requirement, repeat to obtain the corresponding prediction result based on the pre-trained Transformer gas prediction model. In the case where the error between the prediction result and the actual test set result meets the error requirement, record the prediction result.

[0164] As can be seen from the above description, the electronic device provided by the embodiment of the present application uses the self-attention mechanism of Transformer to process the entire sequence in parallel, significantly improving the training efficiency. At the same time, the flower pollination algorithm can quickly find the optimal parameter configuration of the model, improving the prediction accuracy and model performance.

[0165] In another embodiment, the SF6 gas pressure prediction device of the combined electrical apparatus can be separately configured from the central processing unit 9100. For example, the SF6 gas pressure prediction device of the combined electrical apparatus can be configured as a chip connected to the central processing unit 9100, and the function of the SF6 gas pressure prediction method of the combined electrical apparatus is realized through the control of the central processing unit.

[0166] As Figure 13 shown, the electronic device 9600 may further include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It should be noted that the electronic device 9600 does not necessarily have to include Figure 13 all the components shown in Figure 13 ; in addition, the electronic device 9600 may further include

[0167] As Figure 13 shown, the central processing unit 9100 is sometimes also called a controller or an operation control, and may include a microprocessor or other processor devices and / or logic devices. The central processing unit 9100 receives inputs and controls the operations of the various components of the electronic device 9600.

[0168] Among them, the memory 9140 can be, for example, one or more of a buffer, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory, or other suitable devices. The above information related to failures can be stored, and in addition, programs for executing relevant information can also be stored. And the central processing unit 9100 can execute the programs stored in the memory 9140 to implement information storage or processing, etc.

[0169] The input unit 9120 provides inputs to the central processing unit 9100. The input unit 9120 is, for example, a key or a touch input device. The power supply 9170 is used to supply power to the electronic device 9600. The display 9160 is used to display display objects such as images and texts. The display can be, for example, an LCD display, but is not limited thereto.

[0170] The memory 9140 can be a solid-state memory. For example, a read-only memory (ROM), a random access memory (RAM), a SIM card, etc. It can also be such a memory that stores information even when powered off, can be selectively erased and has more data. Examples of such a memory are sometimes called EPROMs, etc. The memory 9140 can also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes called a buffer). The memory 9140 may include an application / function storage unit 9142, and the application / function storage unit 9142 is used to store application programs and function programs or the processes for operating the electronic device 9600 through the central processing unit 9100.

[0171] The memory 9140 may further include a data storage unit 9143 for storing data such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 9144 of the memory 9140 may include various drivers for the communication functions of the electronic device and / or for performing other functions of the electronic device (such as a messaging application, an address book application, etc.).

[0172] The communication module 9110 is a transmitter / receiver 9110 that transmits and receives signals via the antenna 9111. The communication module (transmitter / receiver) 9110 is coupled to the central processor 9100 to provide input signals and receive output signals, which may be the same as in the case of a conventional mobile communication terminal.

[0173] Based on different communication technologies, multiple communication modules 9110 may be provided in the same electronic device, such as a cellular network module, a Bluetooth module, and / or a wireless local area network module, etc. The communication module (transmitter / receiver) 9110 is also coupled to the speaker 9131 and the microphone 9132 via the audio processor 9130 to provide an audio output via the speaker 9131 and receive an audio input from the microphone 9132, thereby implementing normal telecommunication functions. The audio processor 9130 may include any suitable buffers, decoders, amplifiers, etc. In addition, the audio processor 9130 is also coupled to the central processor 9100, so that recording can be performed on the local machine through the microphone 9132, and the sound stored on the local machine can be played through the speaker 9131.

[0174] Embodiments of the present application also provide a computer-readable storage medium capable of implementing all steps in the combined electrical appliance sulfur hexafluoride gas pressure prediction method where the execution subject in the above embodiments is a server or a client. A computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, all steps in the combined electrical appliance sulfur hexafluoride gas pressure prediction method where the execution subject in the above embodiments is a server or a client are implemented. For example, when the processor executes the computer program, the following steps are implemented:

[0175] Step S101: Input the data to be measured into a pre-trained Transformer gas prediction model and obtain a corresponding prediction result. Among them, the Transformer gas prediction model is obtained by processing the time series data of the sulfur hexafluoride (SF6) gas pressure within a preset time period, dividing it into a training set and a test set, inputting it into the Transformer gas prediction model for training, and optimizing the parameters through the flower pollination algorithm;

[0176] Step S102: In the case where the error between the prediction result and the actual test set result does not meet the error requirement, repeat to obtain the corresponding prediction result based on the pre-trained Transformer gas prediction model. In the case where the error between the prediction result and the actual test set result meets the error requirement, record the prediction result.

[0177] As can be seen from the above description, the computer-readable storage medium provided by the embodiments of the present application uses the self-attention mechanism of Transformer to process the entire sequence in parallel, significantly improving the training efficiency. At the same time, the flower pollination algorithm can quickly find the optimal parameter configuration of the model, improving the prediction accuracy and model performance.

[0178] The embodiments of the present application also provide a computer program product that can implement all the steps in the method for predicting the sulfur hexafluoride gas pressure of a combined electrical appliance whose execution subject in the above embodiments is a server or a client. When the computer program / instructions are executed by a processor, the steps of the method for predicting the sulfur hexafluoride gas pressure of the combined electrical appliance are implemented. For example, the computer program / instructions implement the following steps:

[0179] Step S101: Input the data to be measured into the pre-trained Transformer gas prediction model and obtain the corresponding prediction result. Among them, the Transformer gas prediction model is obtained by processing the time series data of the sulfur hexafluoride (SF6) gas pressure within a preset time period, dividing it into a training set and a test set, inputting it into the Transformer gas prediction model for training, and optimizing the parameters through the flower pollination algorithm.

[0180] Step S102: In the case where the error between the prediction result and the actual test set result does not meet the error requirement, repeat to obtain the corresponding prediction result based on the pre-trained Transformer gas prediction model. In the case where the error between the prediction result and the actual test set result meets the error requirement, record the prediction result.

[0181] As can be seen from the above description, the computer program product provided by the embodiments of the present application uses the self-attention mechanism of Transformer to process the entire sequence in parallel, significantly improving the training efficiency. At the same time, the flower pollination algorithm can quickly find the optimal parameter configuration of the model, improving the prediction accuracy and model performance.

[0182] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, apparatus, or computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0183] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (apparatuses), and computer program products according to the embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for realizing the functions specified in Figure 1 one or more of the flows Figure 1 or blocks or a combination of blocks.

[0184] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device, and the instruction device realizes the functions specified in Figure 1 one or more of the flows Figure 1 or blocks or a combination of blocks.

[0185] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in Figure 1 one or more of the flows Figure 1 or blocks or a combination of blocks.

[0186] Specific embodiments are applied in the present invention to elaborate on the principles and implementation manners of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for predicting sulfur hexafluoride gas pressure in a combined electrical appliance, characterized in that: The method comprises: The data to be tested is input into a pre-trained Transformer gas prediction model to obtain a corresponding prediction result, wherein the Transformer gas prediction model is obtained by processing the time series data of sulfur hexafluoride SF6 gas pressure within a preset time length, dividing it into a training set and a test set, inputting it into the Transformer gas prediction model for training, and optimizing the parameters through a flower pollination algorithm; When the error between the prediction result and the actual test set result does not meet the error requirement, the corresponding prediction result is obtained repeatedly based on the pre-trained Transformer gas prediction model. When the error between the prediction result and the actual test set result meets the error requirement, the prediction result is recorded.

2. The method for predicting sulfur hexafluoride gas pressure of a combined electrical appliance according to claim 1, characterized in that: The training process of the Transformer gas prediction model includes: Obtaining SF6 gas pressure time series data in a preset time period through SF6 online monitoring equipment; Processing and normalizing the missing data, duplicate data and abnormal data of the SF6 gas pressure time series data; The processed data is divided into a training set and a test set and input into a Transformer gas prediction model for training to obtain the Transformer gas prediction model.

3. The method for predicting sulfur hexafluoride gas pressure of a combined electrical appliance according to claim 2, characterized in that: The processing of missing data, duplicate data and abnormal data and normalization of the SF6 gas pressure time series data includes: Adopt the adjacent value filling method to process missing data, and replace the missing value with the average value of the adjacent position data before and after, wherein the abnormal number outside the preset range is regarded as the missing data; Deleting duplicate data in the time series number; The time series data is compressed to a preset normal distribution range and converted into a dimensionless form.

4. The method for predicting sulfur hexafluoride gas pressure of a combined electrical appliance according to claim 2, characterized in that: The processed data is divided into a training set and a test set and input into a Transformer gas prediction model for training to obtain the Transformer gas prediction model, including: Convert the time series data into an embedding vector of fixed dimension; Mapping each embedding vector element in the time series data into a one-dimensional vector containing position information to add position encoding to the embedding vector; Based on the self-attention mechanism, by calculating the weights of each element in the sequence with other elements, a comprehensive representation vector reflecting the relationship between elements is generated; Perform multiple self-attention operations based on multiple stacked self-attention layers; Combining the target sequence with the representation vector, and generating corresponding prediction results using a self-attention mechanism and an interaction mechanism; The parameters of the Transformer gas prediction model are optimized by a flower pollination algorithm, and the optimized model parameters are saved to obtain the Transformer gas prediction model.

5. The method for predicting sulfur hexafluoride gas pressure of a combined electrical appliance according to claim 1, characterized in that: The error between the predicted result and the actual test set result includes the mean absolute percentage error and the root mean square error.

6. A combined electrical appliance sulfur hexafluoride gas pressure prediction device, characterized in that: The device comprises: A data input module is used to: input the data to be tested into a pre-trained Transformer gas prediction model and obtain the corresponding prediction result, wherein the Transformer gas prediction model is obtained by processing the time series data of sulfur hexafluoride SF6 gas pressure within a preset time length, dividing it into a training set and a test set, inputting it into the Transformer gas prediction model for training, and optimizing the parameters through a flower pollination algorithm; The prediction output module is used to: when the error between the prediction result and the actual test set result does not meet the error requirement, repeatedly obtain the corresponding prediction result based on the pre-trained Transformer gas prediction model, and when the error between the prediction result and the actual test set result meets the error requirement, record the prediction result.

7. The device for predicting sulfur hexafluoride gas pressure of a combined electrical appliance according to claim 6, characterized in that: The training process of the Transformer gas prediction model includes: The data acquisition module is used to: obtain the SF6 gas pressure time series data in a preset time period through the SF6 online monitoring device; A data processing module is used to: process missing data, duplicate data and abnormal data and normalize the SF6 gas pressure time series data; The model building module is used to: divide the processed data into a training set and a test set and input them into the Transformer gas prediction model for training to obtain the Transformer gas prediction model.

8. The device for predicting sulfur hexafluoride gas pressure of a combined electrical appliance according to claim 7, characterized in that: The data processing module comprises: A missing exception processing unit is used to: process missing data using a neighboring value filling device and replace the missing value with the average value of the adjacent position data before and after, wherein the abnormal number outside the preset range is regarded as the missing data; A repeated value processing unit, used to: delete repeated data in the time series number; The compression unit is used to compress the time series data into a preset normal distribution range and convert it into a dimensionless form.

9. The device for predicting sulfur hexafluoride gas pressure of a combined electrical appliance according to claim 7, characterized in that: The model building module comprises: Embedding vector generation unit: converting the time series data into an embedding vector of fixed dimension; A position encoding unit: mapping each embedding vector element in the time series data into a one-dimensional vector containing position information, so as to add a position code to the embedding vector; Self-attention mechanism unit: Based on the self-attention mechanism, by calculating the weight of each element in the sequence with other elements, a comprehensive representation vector reflecting the relationship between elements is generated; Multi-layer self-attention stacking unit: multiple self-attention operations are performed based on multiple stacked self-attention layers; Decoding and prediction unit: combining the target sequence with the representation vector, and generating corresponding prediction results by using the self-attention mechanism and the interaction mechanism; Parameter optimization and preservation unit: optimizes the parameters of the Transformer gas prediction model through the flower pollination algorithm, and saves the optimized model parameters to obtain the Transformer gas prediction model.

10. The device for predicting sulfur hexafluoride gas pressure of a combined electrical appliance according to claim 6, characterized in that: The error between the predicted result and the actual test set result includes the mean absolute percentage error and the root mean square error.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method for predicting the sulfur hexafluoride gas pressure of a combined electrical appliance according to any one of claims 1 to 5 are implemented.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for predicting the pressure of sulfur hexafluoride gas in a combined electrical appliance as claimed in any one of claims 1 to 5 are implemented.

13. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by a processor, the steps of the method for predicting the pressure of sulfur hexafluoride gas in a combined electrical appliance as described in any one of claims 1 to 5 are implemented.

Images