Prediction method and system for rod-plate gap direct current discharge voltage at extreme temperature

Abstract

The invention discloses a method and a system for predicting rod-plate gap direct current discharge voltage at an extreme temperature, and belongs to the technical field. The method comprises the following steps: taking the structural features and the meteorological features as feature input data; dividing the processed data into a training set and a test set according to a preset proportion, and optimizing key hyper-parameters of an AdaBoost-SVR model on the training set by adopting a five-fold cross validation grid search method so as to select an optimal hyper-parameter combination to construct a model; and inputting the feature input data into the AdaBoost-SVR model for prediction to obtain the direct current discharge voltage of the rod-plate gap at the extreme temperature. The method is suitable for data with small samples, reliable prediction values can be output only by depending on basic structure features and meteorological feature input of the rod-plate gap under the direct-current voltage without complex calculation, and the technical method has a continuous learning mechanism and can optimize the prediction precision along with test data accumulation.

Description

Technical Field

[0001] This invention relates to the field of high voltage and insulation technology, and more specifically, to a method and system for predicting the DC discharge voltage of a rod-plate gap under extreme temperatures. Background Technology

[0002] With the continuous deepening of my country's "West-to-East Power Transmission" energy strategy and the large-scale construction of ultra-high voltage direct current (UHVDC) projects, long-distance, large-capacity power transmission has become the core direction of the State Grid's development. UHVDC transmission technology, with its advantages of long transmission distance, large capacity, and low loss, plays an irreplaceable role in cross-regional energy allocation. The DC withstand level of the external insulation air gap of key DC equipment is of great significance for the design of the equipment's external insulation structure and its safe operation. As the most typical electrode structure, the accurate prediction of the DC discharge voltage of the rod-plate gap can effectively guide the design of the external insulation.

[0003] The DC discharge voltage of the rod-plate gap is influenced by a combination of factors, including gap distance, atmospheric pressure, dry temperature, humid temperature, and humidity. These factors exhibit a complex nonlinear relationship with the discharge voltage. Currently, the g-parameter method recommended in IEC 60060-1 is commonly used in engineering for insulation coordination design. However, this method is derived from limited experimental data, and its applicability, especially under conditions such as high altitude and extreme temperature and humidity, lacks sufficient verification. Furthermore, the limited number of DC discharge test samples under various temperature and humidity conditions makes it difficult to establish high-precision prediction models using traditional regression and fitting methods, severely restricting the efficiency and reliability of engineering design.

[0004] In recent years, domestic and international research institutions, including the China Electric Power Research Institute, have conducted numerous rod-plate gap DC discharge experiments in extreme natural temperature environments ranging from -30℃ to 45℃, accumulating experimental data under different meteorological conditions. This has provided a solid foundation for establishing data-driven prediction models. How to fully utilize existing data to construct a discharge voltage prediction model that accurately reflects the coupling effects of multiple factors has become an urgent need in the field of high-voltage engineering. Summary of the Invention

[0005] To address the above problems, this invention proposes a method for predicting the DC discharge voltage of a rod-plate gap under extreme temperatures, comprising:

[0006] Based on pre-acquired multiple sets of rod-plate gap DC discharge voltage data, structural features and meteorological features are selected, and the structural features and meteorological features are used as feature input data;

[0007] The Min-Max normalization method is used to normalize multiple sets of rod-plate gap DC discharge voltage data to obtain processed data. Then, the processed data is divided into training set and test set according to a preset ratio, and a grid search method with 5-fold cross-validation is used on the training set.

[0008] Using the mean squared error (MSE) as the objective function for training the model, key hyperparameters are selected for tuning to obtain the optimal key hyperparameters of the model. The model is then trained using the optimal key hyperparameters and the training set to construct the AdaBoost-SVR model.

[0009] The AdaBoost-SVR model is tested using the test set to evaluate its prediction accuracy. If the prediction accuracy meets the requirements, the feature input data is input into the AdaBoost-SVR model for prediction to obtain the DC discharge voltage of the rod-plate gap at extreme temperatures.

[0010] Optional structural features include: gap distance.

[0011] Optional meteorological features include: dry temperature, wet temperature, relative humidity, and air pressure.

[0012] Optional, the default ratio is 20%.

[0013] Optional, key hyperparameters include: learning rate, number of base learners, SVR regularization parameter, SVR kernel function parameter, and insensitive loss parameter.

[0014] Optionally, key hyperparameters can be selected for tuning, including: pre-setting the value range of the key hyperparameters, and using a grid search method to traverse all combinations of key hyperparameters, filtering out the parameter combination with the smallest MSE during cross-validation within the value range, as the optimal key hyperparameters.

[0015] Optionally, the model is trained on the training set to build the AdaBoost-SVR model, including:

[0016] Using the optimal key hyperparameters, an initial AdaBoost-SVR model is constructed. The structural features and meteorological features in the training set are used as feature inputs to train the initial AdaBoost-SVR model. The AdaBoost algorithm is used to perform weighted ensemble of multiple SVR base learners to obtain the AdaBoost-SVR model.

[0017] Furthermore, this invention also proposes a prediction system for the DC discharge voltage of a rod-plate gap under extreme temperatures, comprising:

[0018] The initial unit is used to select structural features and meteorological features based on multiple sets of pre-acquired rod-plate gap DC discharge voltage data, and to use the structural features and meteorological features as feature input data.

[0019] The data processing unit is used to normalize multiple sets of rod-plate gap DC discharge voltage data using the Min-Max normalization method to obtain processed data. Then, the processed data is divided into training set and test set according to a preset ratio, and a grid search method with 5-fold cross-validation is used on the training set.

[0020] The modeling unit is used to select key hyperparameters for tuning, using the mean squared error (MSE) as the optimization objective function of the training model, to obtain the optimal key hyperparameters of the model, and to train the model with the optimal key hyperparameters and the training set to construct the AdaBoost-SVR model.

[0021] The output unit is used to test the AdaBoost-SVR model using the test set, evaluate the prediction accuracy of the AdaBoost-SVR model, and if the prediction accuracy meets the requirements, input the feature input data into the AdaBoost-SVR model for prediction to obtain the DC discharge voltage of the rod-plate gap at extreme temperatures.

[0022] Optional structural features include: gap distance.

[0023] Optional meteorological features include: dry temperature, wet temperature, relative humidity, and air pressure.

[0024] Optional, the default ratio is 20%.

[0025] Optional, key hyperparameters include: learning rate, number of base learners, SVR regularization parameter, SVR kernel function parameter, and insensitive loss parameter.

[0026] Optionally, key hyperparameters can be selected for tuning, including: pre-setting the value range of the key hyperparameters, and using a grid search method to traverse all combinations of key hyperparameters, filtering out the parameter combination with the smallest MSE during cross-validation within the value range, as the optimal key hyperparameters.

[0027] Optionally, the training set is used to train the training model to build the AdaBoost-SVR model, including:

[0028] Using the optimal key hyperparameters, an initial AdaBoost-SVR model is constructed. The structural features and meteorological features in the training set are used as feature inputs to train the initial AdaBoost-SVR model. The AdaBoost algorithm is used to perform weighted ensemble of multiple SVR base learners to obtain the AdaBoost-SVR model.

[0029] In another aspect, the present invention also provides a computing device, comprising: one or more processors;

[0030] A processor is used to execute one or more programs;

[0031] When the one or more programs are executed by the one or more processors, the method described above is implemented.

[0032] In another aspect, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed, implements the method described above.

[0033] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0034] This invention provides a method for predicting the DC discharge voltage of a rod-plate gap under extreme temperatures, comprising: selecting structural features and meteorological features based on multiple sets of pre-acquired rod-plate gap DC discharge voltage data, and using the structural features and meteorological features as feature input data; normalizing the multiple sets of rod-plate gap DC discharge voltage data using the Min-Max normalization method to obtain processed data, then dividing the processed data into a training set and a test set according to a preset ratio, and using a grid search method with 5-fold cross-validation on the training set; using the mean square error (MSE) as the objective function for optimizing the training model, selecting key hyperparameters for tuning, obtaining the optimal key hyperparameters of the model, training the model with the optimal key hyperparameters and the training set to construct an AdaBoost-SVR model; testing the AdaBoost-SVR model using the test set to evaluate the prediction accuracy of the AdaBoost-SVR model; if the prediction accuracy meets the requirements, then inputting the feature input data into the AdaBoost-SVR model for prediction to obtain the DC discharge voltage of the rod-plate gap under extreme temperatures. This invention is suitable for data with small sample sizes. It only relies on the basic structural features of the rod-plate gap and meteorological features as inputs. It can output reliable prediction values ​​without complex calculations. The technical method of this invention has a continuous learning mechanism, which can optimize the prediction accuracy as experimental data accumulates. It has significant long-term application value and provides reliable design guidance for power transmission and transformation engineering design. Attached Figure Description

[0035] Figure 1 This is a flowchart of the method of the present invention;

[0036] Figure 2 This is a flowchart of an embodiment of the method of the present invention;

[0037] Figure 3 This is a flowchart illustrating the modeling process of an embodiment of the method of the present invention;

[0038] Figure 4 This describes the structure of the system of the present invention. Detailed Implementation

[0039] Exemplary embodiments of the invention will now be described with reference to the accompanying drawings. However, the invention may be embodied in many different forms and is not limited to the embodiments described herein. These embodiments are provided to fully and completely disclose the invention and to fully convey its scope to those skilled in the art. The terminology used in the exemplary embodiments illustrated in the drawings is not intended to limit the invention. In the drawings, the same units / elements are referred to by the same reference numerals.

[0040] Unless otherwise stated, the terms used herein (including technical terms) have their common meaning as understood by one of ordinary skill in the art. Furthermore, it is understood that terms defined in commonly used dictionaries should be understood to have a meaning consistent with the context of their relevant field, and not to be interpreted as having an idealized or overly formal meaning.

[0041] Example 1:

[0042] This invention proposes a method for predicting the DC discharge voltage of a rod-plate gap under extreme temperatures, such as... Figure 1 As shown, it includes:

[0043] Step 1: Based on the pre-acquired multiple sets of rod-plate gap DC discharge voltage data, select structural features and meteorological features, and use the structural features and meteorological features as feature input data;

[0044] Step 2: The Min-Max normalization method is used to normalize the DC discharge voltage data of multiple rod-plate gaps to obtain processed data. Then, the processed data is divided into training set and test set according to a preset ratio, and a grid search method with 5-fold cross-validation is used on the training set.

[0045] Step 3: Using the mean squared error (MSE) as the objective function for optimizing the training model, select key hyperparameters for tuning, obtain the optimal key hyperparameters of the model, and train the training model with the optimal key hyperparameters, as well as the training subset and validation subset, to construct the AdaBoost-SVR model.

[0046] Step 4: Use the test set to test the AdaBoost-SVR model and evaluate the prediction accuracy of the AdaBoost-SVR model. If the prediction accuracy meets the requirements, input the feature input data into the AdaBoost-SVR model for prediction to obtain the DC discharge voltage of the rod-plate gap at extreme temperatures.

[0047] Among them, structural features include: gap distance.

[0048] Meteorological characteristics include: dry temperature, wet temperature, relative humidity, and air pressure.

[0049] The preset ratio of the training set to the test set is 8:2.

[0050] Key hyperparameters include: learning rate, number of base learners, SVR regularization parameter, SVR kernel function parameter, and insensitive loss parameter.

[0051] The process of selecting key hyperparameters for optimization includes: pre-setting the value range of the key hyperparameters, and using a grid search method to traverse all combinations of key hyperparameters to select the parameter combination with the smallest MSE during cross-validation within the value range as the optimal key hyperparameter.

[0052] The entire training set is used to train the training model to build the AdaBoost-SVR model, including:

[0053] Using the optimal key hyperparameters, an initial AdaBoost-SVR model is constructed. The structural features and meteorological features in the training set are used as feature inputs to train the initial AdaBoost-SVR model. The AdaBoost algorithm is used to perform weighted ensemble of multiple SVR base learners to obtain the AdaBoost-SVR model.

[0054] The invention will be further illustrated below with specific examples:

[0055] Case process, such as Figure 2 As shown, it includes:

[0056] Step 1: Based on the experimental data of 36 sets of DC discharge voltage data for the rod-plate gap, the gap distance was selected as the structural feature, and dry temperature, wet temperature, relative humidity and air pressure were selected as meteorological features. The structural features and meteorological features were used as the feature inputs of the model. The Min-Max normalization method was used to process the data. 8 sets of data were randomly selected from the 36 sets of data at a ratio of 20% as the test set for subsequent model performance verification.

[0057] Step 2: For the remaining 28 sets of rod-plate gap DC discharge voltage data, a grid search method with 5-fold cross-validation is used. The mean squared error (MSE) is used as the objective function for model optimization. Key hyperparameters are selected for tuning, including the learning rate lr, the number of base learners n_estimators, the SVR regularization parameter C, the SVR kernel parameter γ, and the insensitive loss parameter ∈. A reasonable range of hyperparameter values ​​is first preset, and then all hyperparameter combinations are traversed using the grid search method to select the parameter combination with the minimum MSE during cross-validation, which is then used as the optimal hyperparameters for the AdaBoost-SVR model.

[0058] Step 3: Based on the optimal hyperparameters obtained in Step 2, construct the AdaBoost-SVR model. The specific process is as follows: Figure 3 As shown, the 28 sets of data divided in step one are selected as the complete training set. The gap distance is used as the structural feature, and dry temperature, wet temperature, relative humidity and air pressure are used as meteorological features. The two types of features are input into the AdaBoost-SVR model for training. The AdaBoost algorithm is used to weight and integrate multiple SVR base learners to finally obtain the AdaBoost-SVR model with discharge voltage prediction capability.

[0059] Step 4: Using the 8 test sets of data divided in Step 1, evaluate the prediction accuracy of the AdaBoost-SVR model trained in Step 3. If the prediction accuracy is met, select rod-plate gap discharge voltage data from the new dataset to be predicted and input it into the AdaBoost-SVR discharge voltage prediction model to obtain the predicted value of the discharge voltage.

[0060] in:

[0061] Step 1: Data Preprocessing and Feature Selection

[0062] 1. Data Acquisition and Feature Selection:

[0063] A total of 36 data points of rod-plate gap discharge voltage were obtained. The gap distance was selected as the structural feature, and dry temperature, wet temperature, relative humidity and air pressure were selected as meteorological features. The structural features and meteorological features were used as the feature inputs for subsequent training models.

[0064] 2. Data normalization:

[0065] For each group of data, the Min-Max normalization method is used to process the data. The formula is as follows:

[0066]

[0067] x i x represents the original eigenvalues. min and x maxThe minimum and maximum values ​​of this feature. Normalization is used to scale all feature values ​​to the range of 0 to 1.

[0068] 3. Divide the dataset into training and testing sets:

[0069] Eight sets of data were randomly selected as the test set for the AdaBoost-SVR model to evaluate its predictive performance; the remaining 28 sets of data were used as the training set for model training and hyperparameter tuning.

[0070] Step 2: Hyperparameter tuning of the AdaBoost-SVR model:

[0071] 1. Select the hyperparameter training set for hyperparameter tuning:

[0072] The five-fold cross-validation method is adopted, in which the training set is randomly divided into five mutually exclusive subsets. In each round, four subsets are used as the training subset and the remaining subset is used as the validation subset for model training and evaluation. Finally, the evaluation results of the five rounds are combined.

[0073] 2. Select the hyperparameters that need to be tuned for AdaBoost-SVR:

[0074] Choose the learning rate lr, the number of base learners n_estimators, the SVR regularization parameter C, the SVR kernel function parameter γ, and the insensitive loss parameter ∈ as the hyperparameters that need to be tuned.

[0075] 3. Establish the objective function for optimization:

[0076] Based on the requirements of the model prediction task, the mean squared error (MSE) is set as the optimization objective function:

[0077]

[0078] Among them, y i This is the actual value. Let be the predicted value, and n be the number of samples. Using a grid search method, the combination of hyperparameters that minimizes the MSE is selected as the optimal hyperparameters during cross-validation.

[0079] Step 3: Model Training

[0080] 1. Set the AdaBoost-SVR hyperparameters and build the AdaBoost-SVR model:

[0081] Based on the optimal hyperparameters obtained in step two, an AdaBoost-SVR model is established. The specific hyperparameters include the learning rate lr, the number of base learners n_estimators, the SVR regularization parameter C, the SVR kernel function parameter γ, and the insensitive loss parameter ∈.

[0082] 2. Select training set data:

[0083] The 28 datasets selected in step one were used as the training set, with their structural and meteorological features as input features. The final AdaBoost-SVR model was then retrained on this complete training set using the optimal hyperparameters.

[0084] 3. Train the AdaBoost-SVR model:

[0085] Using mean squared error (MSE) as the objective function, AdaBoost-SVR improves model performance by integrating multiple SVR base learners. For the rod-plate gap discharge voltage prediction task, AdaBoost-SVR iteratively trains each SVR base learner and adjusts sample weights based on the performance of the base learners. The specific process is as follows:

[0086] 3.1 Sample weight initialization:

[0087] Initially, each training sample is assigned an equal weight. Let the number of training samples be n. Then:

[0088]

[0089] 3.2 Iterative Base Learner:

[0090] For the t-th iteration (t = 1, 2, ..., n_estimators):

[0091] Using the current sample weight ω t,i Training an SVR base learner h t (x);

[0092] Calculate the weighted error of the base learner on the training set:

[0093]

[0094] Where h t (x i ) represents the predicted value, y i This is the actual discharge voltage value;

[0095] 3.3 Computational Base Learner:

[0096] The weights of the base learner in the ensemble model are calculated based on the weighted error.

[0097]

[0098] 3.4 Update sample weights:

[0099] Adjust the weights of each sample, increasing the weight of samples with larger prediction errors:

[0100] ω t+1,i =ω t,i *exp(-α t *h t (x i )*y i ), (6)

[0101] The weights are then normalized.

[0102]

[0103] 3.5 Integrated Base Learners:

[0104] The final AdaBoost-SVR model is a weighted combination of all base learners, meaning that for a new sample x, the predicted value is:

[0105]

[0106] Where T = n_estimators.

[0107] Step 4: Discharge voltage prediction based on the AdaBoost-SVR model:

[0108] Prepare forecast data:

[0109] Based on the experimental design or practical application requirements, determine the specific structural characteristics of the rod-plate gap for which the discharge voltage needs to be predicted, including the gap distance. Identify relevant meteorological characteristics, such as dry temperature, wet temperature, relative humidity, and air pressure, as feature inputs to the model and perform the same normalization process.

[0110] Perform voltage prediction:

[0111] The data is input into the trained Adaboost-SVR model obtained in step two to obtain the predicted value of the discharge voltage. The specific formula is as follows:

[0112]

[0113] Where x contains the structural and meteorological characteristics of rod-plate gap discharge data;

[0114] Based on the AdaBoost-SVR prediction model for rod-plate gap discharge voltage obtained from the above steps, the discharge voltage can be predicted by inputting the gap distance, dry temperature, wet temperature, relative humidity, and air pressure.

[0115] 3. Model Evaluation:

[0116] The model performance was evaluated on eight sets of test data, and metrics such as mean absolute percentage error (MAPE) were calculated. In this example, the MAPE of the AdaBoost-SVR method was 2.69%, indicating that the method can accurately predict the discharge voltage of the rod-plate gap.

[0117] The AdaBoost-SVR discharge voltage prediction model in this invention has significant advantages: it is highly user-friendly with a simple operation process, especially suitable for small sample data scenarios; it only requires input of the basic structural features and meteorological characteristics of the rod-plate gap, without the need for manual intervention in complex calculations, and can output reliable discharge voltage prediction results; at the same time, it has a continuous learning mechanism, and with the continuous accumulation of experimental data, it can re-adjust hyperparameters and retrain the model to gradually optimize the prediction accuracy, demonstrating outstanding long-term application value; it can provide accurate data support for gap distance design and insulation coordination scheme formulation under DC voltage in power transmission and transformation projects, and provide reliable guidance for improving engineering safety and efficiency.

[0118] Example 2:

[0119] This invention also proposes a prediction system 200 for the DC discharge voltage of the rod-plate gap under extreme temperatures, such as... Figure 4 As shown, it includes:

[0120] Initial unit 201 is used to select structural features and meteorological features based on multiple sets of pre-acquired rod-plate gap DC discharge voltage data, and use the structural features and meteorological features as feature input data;

[0121] Data processing unit 202 is used to process multiple sets of rod-plate gap DC discharge voltage data using the Min-Max normalization method to obtain processed data, divide the processed data into training set and test set according to a preset ratio, and use a grid search method with 5-fold cross-validation on the training set.

[0122] Modeling unit 203 is used to optimize the training model with mean squared error (MSE) as the objective function, select key hyperparameters for tuning, obtain the optimal key hyperparameters of the model, and train the model with the optimal key hyperparameters and the training set to construct the AdaBoost-SVR model.

[0123] Output unit 204 is used to test the AdaBoost-SVR model using the test set, evaluate the prediction accuracy of the AdaBoost-SVR model, and if the prediction accuracy meets the requirements, input the feature input data into the AdaBoost-SVR model for prediction to obtain the DC discharge voltage of the rod-plate gap at extreme temperatures.

[0124] Among them, structural features include: gap distance.

[0125] Meteorological characteristics include: dry temperature, wet temperature, relative humidity, and air pressure.

[0126] The preset ratio is 20%.

[0127] Key hyperparameters include: learning rate, number of base learners, SVR regularization parameter, SVR kernel function parameter, and insensitive loss parameter.

[0128] The process of selecting key hyperparameters for optimization includes: pre-setting the value range of the key hyperparameters, and using a grid search method to traverse all combinations of key hyperparameters to select the parameter combination with the smallest MSE during cross-validation within the value range as the optimal key hyperparameter.

[0129] The process of training the model using the training set to construct the AdaBoost-SVR model includes:

[0130] Using the optimal key hyperparameters, an initial AdaBoost-SVR model is constructed. The structural features and meteorological features in the training set are used as feature inputs to train the initial AdaBoost-SVR model. The AdaBoost algorithm is used to perform weighted ensemble of multiple SVR base learners to obtain the AdaBoost-SVR model.

[0131] This invention is suitable for data with small sample sizes. It only relies on the basic structural features of the rod-plate gap and meteorological features as inputs. It can output reliable prediction values ​​without complex calculations. The technical method of this invention has a continuous learning mechanism, which can optimize the prediction accuracy as experimental data accumulates. It has significant long-term application value and provides reliable design guidance for power transmission and transformation engineering design.

[0132] Example 3:

[0133] Based on the same inventive concept, this invention also provides a computer device, which includes a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the computer storage medium to implement corresponding method flows or corresponding functions, thereby implementing the steps of the methods in the above embodiments.

[0134] Example 4:

[0135] Based on the same inventive concept, this invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the steps of the method in the above embodiments.

[0136] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, 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.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0137] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0138] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0139] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0140] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0141] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for predicting the DC discharge voltage of a rod-plate gap under extreme temperatures, characterized in that, include: Based on pre-acquired multiple sets of rod-plate gap DC discharge voltage data, structural features and meteorological features are selected, and the structural features and meteorological features are used as feature input data; The Min-Max normalization method was used to normalize multiple sets of rod-plate gap DC discharge voltage data to obtain processed data. Then, the processed data was divided into training set and test set according to a preset ratio. On the training set, a grid search method with 5-fold cross-validation was used, with mean square error (MSE) as the optimization objective function of the training model. The key hyperparameters of the AdaBoost-SVR model were tuned to obtain the optimal key hyperparameters of the model and to construct the final prediction model. The AdaBoost-SVR model is tested using the test set to evaluate its prediction accuracy. If the prediction accuracy meets the requirements, the feature input data is input into the AdaBoost-SVR model for prediction to obtain the DC discharge voltage of the rod-plate gap at extreme temperatures.

2. The prediction method according to claim 1, characterized in that, The structural features include: gap distance.

3. The prediction method according to claim 1, characterized in that, The meteorological characteristics include: dry temperature, wet temperature, relative humidity, and air pressure.

4. The prediction method according to claim 1, characterized in that, The preset ratio of the training set to the test set is 8:

2.

5. The prediction method according to claim 1, characterized in that, The key hyperparameters include: learning rate, number of base learners, SVR regularization parameter, SVR kernel function parameter, and insensitive loss parameter.

6. The prediction method according to claim 1, characterized in that, The selection and tuning of key hyperparameters includes: pre-setting the value range of the key hyperparameters, and using a grid search method to traverse all combinations of key hyperparameters to select the parameter combination with the smallest MSE during cross-validation within the value range as the optimal key hyperparameter.

7. The prediction method according to claim 1, characterized in that, The method involves training the model using the training set to construct the AdaBoost-SVR model, including: Using the optimal key hyperparameters, an initial AdaBoost-SVR model is constructed. The structural features and meteorological features in the training set are used as feature inputs to train the initial AdaBoost-SVR model. The AdaBoost algorithm is used to perform weighted ensemble of multiple SVR base learners to obtain the AdaBoost-SVR model.

8. A system for predicting the DC discharge voltage of a rod-plate gap under extreme temperatures, characterized in that, include: The initial unit is used to select structural features and meteorological features based on multiple sets of pre-acquired rod-plate gap DC discharge voltage data, and to use the structural features and meteorological features as feature input data. The data processing unit is used to process multiple sets of rod-plate gap DC discharge voltage data using the Min-Max normalization method to obtain processed data, divide the processed data into training set and test set according to a preset ratio, and use a grid search method with 5-fold cross-validation on the training set. The modeling unit is used to optimize the training model using the mean squared error (MSE) as the objective function, select key hyperparameters for tuning, obtain the optimal key hyperparameters of the model, and train the model with the optimal key hyperparameters and the training set to construct the AdaBoost-SVR model. The output unit is used to test the AdaBoost-SVR model using the test set, evaluate the prediction accuracy of the AdaBoost-SVR model, and if the prediction accuracy meets the requirements, input the feature input data into the AdaBoost-SVR model for prediction to obtain the DC discharge voltage of the rod-plate gap at extreme temperatures.

9. The prediction system according to claim 8, characterized in that, The structural features include: gap distance.

10. The prediction system according to claim 8, characterized in that, The meteorological characteristics include: dry temperature, wet temperature, relative humidity, and air pressure.

11. The prediction system according to claim 8, characterized in that, The preset ratio of the training set to the test set is 8:

2.

12. The prediction system according to claim 8, characterized in that, The key hyperparameters include: learning rate, number of base learners, SVR regularization parameter, SVR kernel function parameter, and insensitive loss parameter.

13. The prediction system according to claim 8, characterized in that, The selection and tuning of key hyperparameters includes: pre-setting the value range of the key hyperparameters, and using a grid search method to traverse all combinations of key hyperparameters to select the parameter combination with the smallest MSE during cross-validation within the value range as the optimal key hyperparameter.

14. The prediction method according to claim 8, characterized in that, The method involves training the model using the training set to construct the AdaBoost-SVR model, including: Using the optimal key hyperparameters, an initial AdaBoost-SVR model is constructed. The structural features and meteorological features in the training set are used as feature inputs to train the initial AdaBoost-SVR model. The AdaBoost algorithm is used to perform weighted ensemble of multiple SVR base learners to obtain the AdaBoost-SVR model.

15. A computer device, characterized in that, include: One or more processors; A processor is used to execute one or more programs; When the one or more programs are executed by the one or more processors, the method described in any one of claims 1-7 is implemented.

16. A computer-readable storage medium, characterized in that, It contains a computer program, which, when executed, implements the method as described in any one of claims 1-7.

Images