Transformer insulation structure design method based on mechanical learning

Abstract

The invention relates to a transformer insulation structure design method based on mechanical learning. The method comprises the steps that parameters of a previous design scheme and transformer electric field distribution simulation data are collected; establishing a transformer insulation design characteristic project; building a pre-transformer insulation structure parameter prediction model and training; applying the trained model to simulation of a new design scheme; and determining transformer design condition parameters. According to the transformer insulation structure design method based on mechanical learning, a complex mapping relation is automatically constructed by analyzing a large amount of historical design data and simulation results, so that the efficiency and accuracy of transformer insulation structure design are improved, the insulation performance is optimized, automatic optimization and feedback adjustment are achieved, and the method is suitable for popularization and application. The limitation of a traditional design method in processing complex electric field distribution and multi-factor influence is solved, and it is ensured that the transformer design meets the high standard and high requirements of a modern electric power system.

Description

Technical Field

[0001] The present invention relates to the technical field of transformers, and in particular to a transformer insulation structure design method based on machine learning. Background Art

[0002] As a vital equipment in the power system, the reliability of the transformer is directly related to the safety and stable operation of the entire power system. The insulation structure design of the transformer is one of the key links to ensure its safe operation. Insulation problems are one of the main causes of transformer failures. Especially under long-term operation and complex working conditions, defects in the insulation structure often cause serious safety accidents, such as insulation aging, partial discharge and even transformer explosion. Therefore, ensuring the reasonable design of the transformer insulation structure can not only extend the service life of the equipment, but also prevent potential hidden dangers of failure and ensure the safe operation of the power system. With the continuous development of the power system, the operating environment of the transformer is becoming increasingly complex, and the requirements for its insulation structure design are also increasing. Traditional design methods can no longer fully meet the high standards and high requirements of modern power equipment.

[0003] The traditional transformer insulation structure design method mainly relies on the standard parameters in the design manual and the experience of the designer. However, this method has significant limitations. First, the standard parameters in the design manual are usually formulated based on typical scenarios and historical data, which is difficult to cover all possible practical application scenarios. Especially when facing complex and non-standard working conditions, the design results are often difficult to achieve the expected results. Secondly, although the experience of designers plays an important role in the design process, the accumulation of experience takes time. It is difficult for novice designers to master sufficient design skills in a short period of time. Even experienced designers may make mistakes in judgment when facing complex design tasks with multiple factors interacting, resulting in inaccurate design results. In addition, the traditional design method is inefficient when dealing with complex electric field distribution and the interaction of multiple design parameters, and it is easy to affect the stability and reliability of the design due to errors in manual calculations or empirical judgments. These problems make it difficult for traditional methods to quickly and accurately respond to the changing operating conditions and higher design requirements in modern power systems.

[0004] As a powerful data analysis and modeling tool, machine learning has shown great potential for application in transformer insulation structure design. Machine learning can automatically establish complex mapping relationships by analyzing a large amount of historical design data and simulation results, thereby more accurately predicting the optimal design parameters of the insulation structure. Compared with traditional methods, the design method based on machine learning has significant advantages. First, machine learning can greatly improve design efficiency, reduce human errors in the design process, and ensure the stability and reliability of the design results. Secondly, when faced with complex multi-factor influences, the machine learning model can better capture and process the relationship between these factors, thereby optimizing the design of the insulation structure. In addition, machine learning can also realize automatic optimization and feedback adjustment of the design process. When the design results do not meet expectations, the model can be quickly adjusted to ensure that the final design meets all requirements. In summary, the transformer insulation structure design method based on machine learning can not only cope with the complexity of modern power systems, but also has a strong application prospect. It is an important development direction in the field of transformer design in the future.

[0005] Patent CN 103310963B proposes a design method for a 35kV distribution three-winding transformer, which includes determining the 0.4kV side capacity, transformer body design, material selection, operating loss calculation and economic analysis. This method is suitable for 35kV substations in remote areas. By optimizing the design and selectively equipping on-load tap-changing switches, it reduces costs and losses and achieves economic operation. Although patent CN 103310963B proposes a design method for a 35kV distribution three-winding transformer suitable for remote areas, it relies on traditional design processes and manual standards, fails to cope with special design requirements under complex operating conditions, and does not adequately consider the mutual influence of multiple factors, resulting in insufficient design flexibility and accuracy.

[0006] Patent CN 117150791B proposes a transformer main insulation structure optimization design method, which involves determining the objective function, setting the insulation structure parameters as decision variables and obtaining constraints. The target solution set is obtained by solving the objective function, and a comprehensive evaluation is performed on it to finally determine the optimization parameters of the main insulation structure. This method makes up for the shortcomings of existing design methods and improves the reliability of the main insulation structure. Although patent CN 117150791B optimizes the transformer main insulation structure, it still relies on traditional optimization methods and does not introduce intelligent algorithms, resulting in low efficiency in processing complex design parameters and electric field distribution, and the inability to automatically adjust the optimization results. The design process may take a long time.

[0007] Patent CN 103310963B proposes a design method for a 35kV distribution three-winding transformer, including: determining the 0.4kV side capacity according to the substation layout, setting the transformer capacity and performing the main body design, selecting appropriate materials, calculating the operating loss, and performing an economic analysis. This method is aimed at substations in remote areas. By configuring a 35 / 10 / 0.4kV three-winding transformer and optimizing the 0.4kV side capacity, the cost and loss are reduced, and the on-load tap changer is considered to achieve economic operation. Patent CN103310963B also relies on traditional design methods and manual standard parameters. It does not provide an effective solution for the complexity of the insulation structure. It is difficult to cope with the complex working conditions and high-standard design requirements in modern power systems, which may lead to inaccurate and unstable design results.

[0008] Patent CN106298215B proposes a distribution transformer design method, including configuring a design scheme, determining operating environment requirements, setting manufacturing costs and performance indicators, and comparing the life cycle cost (LCC). By establishing a cost model and sensitivity analysis, the increase in the life cycle cost is limited and the technical economy is improved. Although patent CN 106298215B improves the economy of transformer design through a life cycle cost model, its design method is mainly based on traditional sensitivity analysis and does not use modern data analysis technology to improve the accuracy of the design and optimize efficiency. Especially when facing multi-dimensional design factors, it may be difficult to reach the optimal solution.

[0009] Patent CN115310353B proposes a method for designing power transformers based on fast multi-objective optimization, including creating an optimization model and using an improved NSGA-Ⅱ algorithm for solving it. The improvement is reflected in the combination of artificial solutions and random solutions when initializing the population, so as to facilitate algorithm convergence, reduce time consumption, avoid local optimality, and maintain a wide distribution of solution space to meet the requirements of optimization effect and fast calculation. Although patent CN 115310353B uses an improved NSGA-Ⅱ algorithm for multi-objective optimization, this method still relies on some artificial solutions during initialization, which may result in limited convergence efficiency of the algorithm. In addition, when faced with more complex design scenarios, the adaptability and adjustment capabilities of the algorithm may be insufficient, and there is still a risk of falling into a local optimal solution. Summary of the invention

[0010] In view of the deficiencies in the above-mentioned prior art, the present invention provides a transformer insulation structure design method based on machine learning, which analyzes a large amount of historical design data and simulation results, constructs a complex mapping relationship, thereby improving the efficiency and accuracy of transformer insulation structure design, optimizing insulation performance, and realizing automatic optimization and feedback adjustment.

[0011] The technical solution adopted by the present invention is:

[0012] A transformer insulation structure design method based on machine learning, comprising:

[0013] Step S1, collecting parameters of previous design solutions and transformer electric field distribution simulation data;

[0014] Step S2, establishing transformer insulation design feature engineering;

[0015] Step S3, constructing a pre-transformer insulation structure parameter prediction model and training;

[0016] Step S4, applying the trained model to the simulation of the new design solution;

[0017] Step S5, determining whether the electric field distortion rate and insulation size meet the requirements, if they do, proceed to step S6, if they do not, proceed to step S4 and perform iterative calculation again;

[0018] Step S6, determining transformer design condition parameters.

[0019] Preferably, step S1, collecting parameters of previous design solutions and transformer electric field distribution simulation data includes:

[0020] Collect transformer design data from existing design solutions;

[0021] With the help of finite element analysis numerical simulation tools, the electric field distribution of the existing design scheme is simulated to obtain the electric field strength, electric field distortion rate and insulation thickness data of the transformer under actual working conditions;

[0022] The electric field distortion rate (D) is defined as:

[0023]

[0024] Where: E max and E min are the maximum and minimum values ​​of the electric field intensity respectively; E avg is the average value of the electric field strength.

[0025] Preferably, in step S2, feature engineering is to convert the original design data and simulation results into features that can be effectively understood and processed by the machine learning model, and extract key parameters that are highly correlated with the insulation structure performance of the transformer;

[0026] By describing the complex relationship between electric field distribution, material properties, and working environment, feature engineering can improve the model's predictive ability and generalization performance, thereby optimizing insulation design, improving the application effect of the model in new solutions, and ensuring that the transformer meets the efficient and safe operation requirements of modern power systems;

[0027] The process of building feature engineering includes the following steps:

[0028] First, collect multi-dimensional data related to transformer insulation design, such as material parameters, electric field distribution, environmental conditions, and historical design results;

[0029] Clean, normalize and feature screen multidimensional data to ensure data quality and consistency;

[0030] Combine electromagnetic field theory and simulation analysis to extract key features that can reflect insulation performance, such as local electric field strength and thermal field coupling characteristics;

[0031] Through dimensionality reduction technology and feature combination, a feature set suitable for machine learning models is constructed to support efficient training and optimization of the model;

[0032] Feature selection is done by analyzing the voltage level, material dielectric constant, winding spacing, oil-paper thickness, and electric field strength parameters, which directly affect the insulation performance and electric field distribution of the transformer;

[0033] In order to capture more complex design influencing factors, the interactive relationship between features is also constructed; after these features are selected, they can effectively improve the prediction and generalization capabilities of the model, so that it can maintain efficient performance in different design tasks;

[0034] Feature engineering not only helps the model better understand the input data, but also reduces the dimension of the data and improves the efficiency and accuracy of model training.

[0035] Preferably, step S3, constructing a pre-transformer insulation structure parameter prediction model and training method, and constructing a machine learning model for predicting transformer insulation structure parameters;

[0036] Building a machine learning model for predicting transformer insulation structural parameters relies on high-quality data and appropriate algorithm selection;

[0037] First, data is collected and preprocessed, including parameter normalization and outlier removal, to ensure the quality of model input data;

[0038] Then, a feature set is established, combining key features of transformer insulation design, such as material properties, electric field strength, and ambient temperature, and the model is trained on the training data;

[0039] Finally, the trained model is applied to simulation data to verify its predictive accuracy and applicability in new design solutions, and the model is continuously iterated and optimized.

[0040] The machine learning model based on the random forest algorithm uses a large amount of historical data and simulation data for training, establishes the mapping relationship between input features and target outputs, and gradually optimizes the prediction results;

[0041] The steps for processing data by the random forest algorithm are as follows:

[0042] By building multiple decision trees and integrating their prediction results, the robustness and accuracy of the model can be improved. When processing transformer insulation design data, the data set is first divided into training and test sets, and the features are standardized. During the training phase, each tree randomly extracts some samples and features from the data set for modeling to ensure the diversity of the model. Then, the outputs of all trees are integrated into the final prediction result by voting or averaging. The random forest algorithm has good generalization ability, can handle high-dimensional data and nonlinear relationships, and has strong fault tolerance to missing data and noise, which is very suitable for complex data modeling needs in transformer insulation design.

[0043] The model is trained using the collected historical data and simulation data. The model learns the mapping relationship between input features and target outputs to gradually optimize the prediction results.

[0044] Using the historical data and simulation data of previous transformer designs, we first annotate and match the input features (such as material parameters, electric field distribution, and environmental conditions) in the data with the target outputs (such as insulation performance indicators and electric field distribution results) to form training samples.

[0045] During the training process, the model learns the nonlinear mapping relationship between the input features and the target output through continuous iteration, so that the prediction error is gradually reduced;

[0046] After each iteration, the model adjusts parameters to reduce errors and optimize its prediction ability. Ultimately, the prediction results are an evaluation of the insulation structure performance in the new transformer design, such as local electric field strength distribution, insulation breakdown voltage and thermal field stability, to ensure that the new solution meets the expected insulation design requirements;

[0047] The performance of the model is continuously evaluated during the training process, using cross-validation techniques to ensure that the model does not overfit.

[0048] Preferably, in step S4, the trained model is applied to the simulation of the new design solution;

[0049] The trained model is directly applied to the new transformer design;

[0050] After the designer inputs new design conditions, the model will output the corresponding insulation structure parameters based on the trained mapping relationship;

[0051] These prediction results need to be verified by simulation tools to ensure that the design can achieve the expected performance in actual applications;

[0052] The simulation verification evaluates whether the insulation structure predicted by the model meets the design specifications by calculating key indicators such as the electric field distribution and electric field distortion rate of the new design.

[0053] The simulation verification steps are as follows:

[0054] The simulation verification uses the insulation structure parameters predicted by the input model to calculate the electric field distribution and key performance indicators of the new design, such as local electric field strength and electric field distortion rate, in the simulation tool. During the simulation process, the results are compared with the design specifications to ensure that all indicators meet the requirements. If the simulation results show that the electric field distortion rate in certain areas is too high or the electric field distribution is uneven, the input parameters need to be adjusted and re-simulated until the predicted results meet the expected performance in the simulation verification, thereby ensuring the reliability and safety of the insulation structure design.

[0055] Preferably, step S5, determining whether the electric field distortion rate, insulation size and electric field uniformity meet the requirements;

[0056] After completing the simulation verification, the design scheme predicted by the model needs to be evaluated to determine whether it meets the predetermined design requirements;

[0057] Key indicators include electric field distortion rate, insulation thickness and electric field uniformity;

[0058] If the simulation results show that these indicators are within the range of the design specifications, the design solution can be considered feasible;

[0059] If the simulation results do not meet the requirements, a feedback mechanism is initiated to analyze the differences between the model prediction results and the actual simulation, and adjust the model or feature selection based on these differences.

[0060] The beneficial effects of the present invention compared with the prior art are as follows:

[0061] The transformer insulation structure design method based on machine learning in the present invention automatically constructs complex mapping relationships by analyzing a large amount of historical design data and simulation results, thereby improving the efficiency and accuracy of transformer insulation structure design, optimizing insulation performance, and realizing automatic optimization and feedback adjustment to address the limitations of traditional design methods in dealing with complex electric field distribution and multi-factor influences, ensuring that transformer design meets the high standards and high requirements of modern power systems.

[0062] Compared with the traditional method that relies on manual standards and designer experience in the prior art, the present invention can automatically process complex design tasks by introducing machine learning technology, greatly reducing the time for manual calculation and simulation verification. This automated design process significantly improves design efficiency and greatly shortens the design cycle.

[0063] The present invention uses a large amount of historical data and simulation data for model training, which can more accurately predict the optimal insulation structure parameters. Compared with the empirical errors and manual calculation deviations that may occur in traditional methods, the design results provided by the present invention have higher accuracy and reliability, and can effectively deal with complex electric field distribution and multi-factor interactions.

[0064] Existing technologies usually rely on standard parameters in typical scenarios and are difficult to cope with special or complex operating conditions. The present invention can better capture and process the relationship between different design factors through the analysis of multi-dimensional data by machine learning models, making the design scheme more flexible and adaptable to a wider range of working conditions and usage environments.

[0065] When the design parameters do not meet the requirements, the existing technology usually needs to be manually adjusted again, which is inefficient. The present invention introduces an automatic optimization and feedback adjustment mechanism. When the insulation structure parameters predicted by the model do not meet the design requirements, feedback adjustment can be quickly performed to automatically optimize the design solution to ensure that the design meets all standards and requirements.

[0066] The existing technology relies on manual design and simulation, which is prone to errors due to the designer's experience and judgment. The present invention reduces the interference of human factors through a data-driven machine learning model, making the design process more objective and stable, and ensuring the consistency and reliability of the design results. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a flow chart of the transformer insulation structure design method based on machine learning;

[0068] Figure 2 It is a flow chart of a method for collecting parameters of previous design solutions and transformer electric field distribution simulation data;

[0069] Figure 3 It is a flowchart of constructing a pre-transformer insulation structure parameter prediction model and training method. DETAILED DESCRIPTION

[0070] The present invention is described in detail below with reference to the accompanying drawings and embodiments:

[0071] Attached Figure 1 It can be seen that a transformer insulation structure design method based on machine learning,

[0072] Step S1, collecting parameters of previous design solutions and transformer electric field distribution simulation data;

[0073] Step S2, establishing transformer insulation design feature engineering;

[0074] Step S3, constructing a pre-transformer insulation structure parameter prediction model and training;

[0075] Step S4, applying the trained model to the simulation of the new design solution;

[0076] Step S5, determining whether the electric field distortion rate and insulation size meet the requirements, if they meet the requirements ("yes"), proceeding to step S6, if they do not meet the requirements ("no"), proceeding to step S4 to recalculate iteratively;

[0077] Step S6, determining transformer design condition parameters.

[0078] like Figure 2 As shown in Figure 2, in transformer design, historical design data and simulation data are the basis for building machine learning models.

[0079] Preferably, step S1: collecting parameters of previous design solutions and transformer electric field distribution simulation data includes:

[0080] By systematically collecting key design parameters, i.e., transformer design data from existing designs; these data include voltage level, current size, geometry, and insulation material properties, all of which are relevant to transformer design and performance;

[0081] With the help of finite element analysis numerical simulation tools, key design parameters are systematically collected and electric field distribution simulation is performed on the existing design scheme, that is, the electric field strength, electric field distortion rate and insulation thickness data of the transformer under actual working conditions are obtained;

[0082] The simulation data is denoised and standardized, and can provide detailed electric field distribution information that is difficult to obtain with traditional design methods, thereby more comprehensively reflecting the impact of design parameters on transformer performance. In order to ensure the quality of the model's training data, the collected data is denoised and standardized, and biased data is removed to ensure data reliability and model prediction accuracy.

[0083] The electric field distortion rate (D) is defined as:

[0084]

[0085] Where: E max and E min are the maximum and minimum values ​​of the electric field intensity respectively; E avg is the average value of the electric field strength.

[0086] Preferably, step S2, establishing transformer insulation design feature engineering includes:

[0087] Feature engineering is the process of converting raw data into a set of features that can be effectively used by machine learning models. For transformer insulation design, it is crucial to extract key features from the original design parameters and simulation data.

[0088] Feature engineering is to convert the original design data and simulation results into features that can be effectively understood and processed by the machine learning model, and extract key parameters that are highly correlated with the performance of the transformer insulation structure;

[0089] By describing the complex relationship between electric field distribution, material properties, and working environment, feature engineering can improve the model's predictive ability and generalization performance, thereby optimizing insulation design, improving the application effect of the model in new solutions, and ensuring that the transformer meets the efficient and safe operation requirements of modern power systems;

[0090] The process of building feature engineering includes the following steps:

[0091] First, collect multi-dimensional data related to transformer insulation design, such as material parameters, electric field distribution, environmental conditions, and historical design results;

[0092] Clean, normalize and feature screen multidimensional data to ensure data quality and consistency;

[0093] The integrity and accuracy of the data are ensured by removing outliers in the design scheme and simulation data and filling missing values; then, the data of different dimensions are normalized, such as converting parameters such as electric field strength and insulation thickness to a unified scale range, so as to ensure that the influence of each parameter on the prediction results remains consistent during the model training process; finally, through feature screening, factors with low correlation with the insulation performance of the transformer are removed, and key feature data such as material parameters and electric field distribution are retained to optimize the input parameters of the model.

[0094] Combine electromagnetic field theory and simulation analysis to extract key features that can reflect insulation performance, such as local electric field strength and thermal field coupling characteristics;

[0095] Finite element analysis is used to simulate the electric field distortion rate, maximum and minimum electric field strengths, average electric field strength of the transformer under working conditions, as well as the thermal stability of the material under different working environments, and the local electric field strength, insulation thickness and thermal field coupling characteristics are extracted as key features.

[0096] Through dimensionality reduction technology and feature combination, a feature set suitable for machine learning models is constructed to support efficient training and optimization of the model;

[0097] The dimensionality reduction technology is used to extract the most representative features from the high-dimensional electric field simulation data, thereby reducing the redundancy of features; then, through feature combination, key features such as electric field strength, thermal field coupling characteristics and material parameters are combined to form new composite features, which can more accurately reflect the insulation performance.

[0098] Feature selection is done by analyzing the voltage level, material dielectric constant, winding spacing, oil-paper thickness, and electric field strength parameters, which directly affect the insulation performance and electric field distribution of the transformer;

[0099] Feature selection identifies and selects the key features that have the greatest impact on transformer insulation performance and electric field distribution by analyzing parameters such as voltage level, material dielectric constant, winding spacing, oil-paper thickness, and electric field strength, thereby simplifying the data set. On the basis of feature selection, "feature combination" appropriately combines these selected important features to generate new composite features. These composite features can more comprehensively and effectively capture the complex relationship between parameters, thereby improving the prediction performance and accuracy of the model.

[0100] In order to capture more complex design influencing factors, the interactive relationship between features is also constructed; after these features are selected, they can effectively improve the prediction and generalization capabilities of the model, so that it can maintain efficient performance in different design tasks;

[0101] Constructing the interactive relationship between features is achieved by identifying and calculating the interactions between different features and their combined effects on the insulation performance of the transformer;

[0102] By analyzing characteristics such as voltage level, material dielectric constant, winding spacing, oil-paper thickness, and electric field strength, it is possible to find out how the combination of these characteristics jointly affects the electric field distribution;

[0103] This process often uses technical methods such as polynomial expansion or interaction term models based on statistics and machine learning to capture complex nonlinear relationships;

[0104] Feature selection is done by evaluating feature importance (such as feature importance scores based on statistical indicators and machine learning algorithms) to select the most critical features for model prediction and ensure that the model maintains efficient performance in different design tasks. This method of selecting and building interactive relationships can improve the model's predictive and generalization capabilities.

[0105] Feature engineering not only helps the model better understand the input data, but also reduces the dimension of the data and improves the efficiency and accuracy of model training.

[0106] like Figure 3As shown, the random forest algorithm is selected as the machine learning model for predicting the insulation structure of the transformer; the collected historical design data and simulation data are used as the input data of the model; the final prediction model is formed, the model performance and model parameter adjustment are evaluated; and the final prediction model is formed.

[0107] Preferably, step S3, constructing a pre-transformer insulation structure parameter prediction model and training method,

[0108] Build a machine learning model for predicting transformer insulation structural parameters;

[0109] Building a machine learning model for predicting transformer insulation structural parameters relies on high-quality data and appropriate algorithm selection;

[0110] First, data is collected and preprocessed, including parameter normalization and outlier removal, to ensure the quality of model input data;

[0111] Then, a feature set is established, combining key features of transformer insulation design, such as material properties, electric field strength, and ambient temperature, and the model is trained on the training data;

[0112] Finally, the trained model is applied to simulation data to verify its predictive accuracy and applicability in new design solutions, and the model is continuously iterated and optimized.

[0113] The machine learning model based on the random forest algorithm uses a large amount of historical data and simulation data for training, establishes the mapping relationship between input features and target outputs, and gradually optimizes the prediction results;

[0114] The steps for processing data by the random forest algorithm are as follows:

[0115] By building multiple decision trees and integrating their prediction results, the robustness and accuracy of the model can be improved. When processing transformer insulation design data, the data set is first divided into training and test sets, and the features are standardized. During the training phase, each tree randomly extracts some samples and features from the data set for modeling to ensure the diversity of the model. Then, the outputs of all trees are integrated into the final prediction result by voting or averaging. The random forest algorithm has good generalization ability, can handle high-dimensional data and nonlinear relationships, and has strong fault tolerance to missing data and noise, which is very suitable for complex data modeling needs in transformer insulation design.

[0116] The model is trained using the collected historical data and simulation data. The model learns the mapping relationship between input features and target outputs to gradually optimize the prediction results.

[0117] Using the historical data and simulation data of previous transformer designs, we first annotate and match the input features (such as material parameters, electric field distribution, and environmental conditions) in the data with the target outputs (such as insulation performance indicators and electric field distribution results) to form training samples.

[0118] During the training process, the model learns the nonlinear mapping relationship between the input features and the target output through continuous iteration, so that the prediction error is gradually reduced;

[0119] After each iteration, the model adjusts parameters to reduce errors and optimize its prediction ability. Ultimately, the prediction results are an evaluation of the insulation structure performance in the new transformer design, such as local electric field strength distribution, insulation breakdown voltage and thermal field stability, to ensure that the new solution meets the expected insulation design requirements;

[0120] The performance of the model is continuously evaluated during the training process, using cross-validation techniques to ensure that the model does not overfit.

[0121] Preferably, step S4: applying the trained model to the simulation of the new design solution;

[0122] The trained model is directly applied to the new transformer design;

[0123] After the designer inputs new design conditions, the model will output the corresponding insulation structure parameters based on the trained mapping relationship;

[0124] These prediction results need to be verified by simulation tools to ensure that the design can achieve the expected performance in actual applications;

[0125] The simulation verification evaluates whether the insulation structure predicted by the model meets the design specifications by calculating key indicators such as the electric field distribution and electric field distortion rate of the new design.

[0126] The simulation verification steps are as follows:

[0127] The simulation verification uses the insulation structure parameters predicted by the input model to calculate the electric field distribution and key performance indicators of the new design, such as local electric field strength and electric field distortion rate, in the simulation tool. During the simulation process, the results are compared with the design specifications to ensure that all indicators meet the requirements. If the simulation results show that the electric field distortion rate in certain areas is too high or the electric field distribution is uneven, the input parameters need to be adjusted and re-simulated until the predicted results meet the expected performance in the simulation verification, thereby ensuring the reliability and safety of the insulation structure design.

[0128] Preferably, step S5, determining whether the electric field distortion rate and insulation size meet the requirements;

[0129] After completing the simulation verification, the design scheme predicted by the model needs to be evaluated to determine whether it meets the predetermined design requirements;

[0130] Key indicators include electric field distortion rate, insulation thickness and electric field uniformity;

[0131] If the simulation results show that these indicators are within the range of the design specifications, the design solution can be considered feasible;

[0132] If the simulation results do not meet the requirements, a feedback mechanism is initiated to analyze the differences between the model prediction results and the actual simulation, and adjust the model or feature selection based on these differences.

[0133] When the simulation results do not meet the requirements, optimize the design by adjusting multiple key parameters. First, adjust the thickness or number of insulating layers to improve the overall insulation strength and breakdown voltage. Second, change the dielectric constant and other properties of the material (such as dielectric loss) to affect the electric field distribution and thus improve the electric field distortion rate. Modify boundary conditions or apply different voltage waveforms to make the electric field more uniform. Finally, consider optimizing the model's mesh density and time step to improve the accuracy and stability of the simulation.

[0134] This feedback process will continuously optimize the model, enabling it to more accurately predict design outcomes, ultimately ensuring the reliability and effectiveness of the design solution.

[0135] Step S6, determining transformer design condition parameters;

[0136] When the simulation verification results meet all design requirements, the final transformer insulation structure design can be determined;

[0137] The design parameters predicted by the model have been verified by simulation, proving that they can perform well in real applications. The output of the design solution includes specific parameters such as the thickness of the insulation material and the spacing between the windings, which will be directly used in production.

[0138] The core innovation of this invention is to introduce machine learning technology into the design process of transformer insulation structure. By analyzing a large amount of historical design data and simulation results, complex mapping relationships are automatically constructed, thereby improving the efficiency and accuracy of the design and avoiding errors caused by experience and manual calculations in traditional design methods.

[0139] The present invention establishes transformer insulation design feature engineering to extract and optimize key features, such as voltage level, material dielectric constant, winding spacing, oil-paper thickness, and electric field strength. These features are carefully selected and interactively constructed to enhance the predictive capability of machine learning models, enabling them to maintain efficient performance in different design tasks.

[0140] The present invention proposes a machine learning model based on the random forest algorithm and uses a large amount of historical data and simulation data for training, establishes a mapping relationship between input features and target outputs, and gradually optimizes the prediction results.

[0141] The present invention proposes a set of simulation verification and feedback optimization mechanisms. The design scheme predicted by the model is verified by simulation tools to determine its rationality and effectiveness. If the design result does not meet expectations, the present invention will start the feedback mechanism, analyze the difference between the model prediction and the actual simulation, adjust the model or feature selection, and ensure the reliability of the final design scheme.

[0142] The automatic optimization and feedback adjustment mechanism in the present invention enables timely adjustment and optimization when the insulation structure parameters predicted by the model do not meet the design requirements, thereby ensuring the efficiency and reliability of the design scheme and solving the limitations of traditional design methods when faced with complex electric field distribution and multi-factor influences.

[0143] The transformer insulation structure design method based on machine learning in the present invention automatically constructs complex mapping relationships by analyzing a large amount of historical design data and simulation results, thereby improving the efficiency and accuracy of transformer insulation structure design, optimizing insulation performance, and realizing automatic optimization and feedback adjustment to address the limitations of traditional design methods in dealing with complex electric field distribution and multi-factor influences, ensuring that transformer design meets the high standards and high requirements of modern power systems.

[0144] Compared with the traditional method that relies on manual standards and designer experience in the prior art, the present invention can automatically process complex design tasks by introducing machine learning technology, greatly reducing the time for manual calculation and simulation verification. This automated design process significantly improves design efficiency and greatly shortens the design cycle.

[0145] The present invention uses a large amount of historical data and simulation data for model training, which can more accurately predict the optimal insulation structure parameters. Compared with the empirical errors and manual calculation deviations that may occur in traditional methods, the design results provided by the present invention have higher accuracy and reliability, and can effectively deal with complex electric field distribution and multi-factor interactions.

[0146] Existing technologies usually rely on standard parameters in typical scenarios and are difficult to cope with special or complex operating conditions. The present invention can better capture and process the relationship between different design factors through the analysis of multi-dimensional data by machine learning models, making the design scheme more flexible and adaptable to a wider range of working conditions and usage environments.

[0147] When the design parameters do not meet the requirements, the existing technology usually needs to be manually adjusted again, which is inefficient. The present invention introduces an automatic optimization and feedback adjustment mechanism. When the insulation structure parameters predicted by the model do not meet the design requirements, feedback adjustment can be quickly performed to automatically optimize the design solution to ensure that the design meets all standards and requirements.

[0148] The existing technology relies on manual design and simulation, which is prone to errors due to the designer's experience and judgment. The present invention reduces the interference of human factors through a data-driven machine learning model, making the design process more objective and stable, and ensuring the consistency and reliability of the design results.

[0149] The transformer insulation structure design method based on machine learning according to the present invention can be understood as a dynamic combination of some implementation details in some embodiments, especially when it comes to relevant algorithms, calculation methods, or software systems or devices. The key variables in the system or method can be set, adjusted or optimized according to the details of the construction method, so as to achieve those expected technical effects. Some details of the transformer insulation structure design method based on machine learning of the present invention can be understood as parameters in a model that can be defined in some embodiments, especially in those embodiments such as machine learning, the details of the present invention can be used to describe how to select and adjust the hyperparameters of the neural network. In some embodiments of the present invention, those details can be defined as key parameters in the algorithm, or configuration parameters between various components of a system, and the overall performance of the system or method can be optimized by these parameters. In some embodiments of the present invention, modules or devices can be used to describe those method steps or details of the present invention or their permutations and combinations, such as the technical details or features of the nth device or the nth module to implement the nth step. In certain method embodiments of the method of the present invention, the result of the final step of the arrangement and combination can be output to the manufacturing device or equipment of the corresponding transformer or transformer component to achieve automatic production, or the result can be fed back to the initial step at the same time to further form a more optimal structural parameter construction, and after multiple iterations, the result of the final step can be output to the manufacturing device or equipment of the corresponding transformer or transformer component to achieve more optimal automatic production.

[0150] The above is only a preferred embodiment of the present invention, and does not limit the structure of the present invention in any form. Any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention are within the scope of the technical solution of the present invention.

Claims

1. A transformer insulation structure design method based on machine learning, characterized in that: include: Step S1, collecting parameters of previous design solutions and transformer electric field distribution simulation data; Step S2, establishing transformer insulation design feature engineering; Step S3, constructing a pre-transformer insulation structure parameter prediction model and training; Step S4, applying the trained model to the simulation of the new design solution; Step S5, determining whether the electric field distortion rate and insulation size meet the requirements, if they do, proceed to step S6, if they do not, proceed to step S4 and perform iterative calculation again; Step S6, determining transformer design condition parameters.

2. The transformer insulation structure design method based on machine learning according to claim 1 is characterized in that: Step S1, collecting parameters of previous design solutions and transformer electric field distribution simulation data includes: Collect transformer design data from existing design solutions; With the help of finite element analysis numerical simulation tools, the electric field distribution of the existing design scheme is simulated to obtain the electric field strength, electric field distortion rate and insulation thickness data of the transformer under actual working conditions; The electric field distortion rate (D) is defined as: Where: E max and E min are the maximum and minimum values ​​of the electric field intensity respectively; E avg is the average value of the electric field strength.

3. The transformer insulation structure design method based on machine learning according to claim 1 is characterized in that: Step S2, the process of establishing feature engineering includes the following steps: Collect multi-dimensional data related to transformer insulation design; Clean, normalize and feature-screen multidimensional data; Combine electromagnetic field theory and simulation analysis to extract key features that can reflect insulation performance; Through dimensionality reduction technology and feature combination, a feature set suitable for machine learning models is constructed to support efficient training and optimization of the model; Feature selection is determined by analyzing voltage level, material dielectric constant, winding spacing, oil-paper thickness and electric field strength parameters.

4. The transformer insulation structure design method based on machine learning according to claim 1 is characterized in that: Step S3, constructing a pre-transformer insulation structure parameter prediction model and training method, and constructing a machine learning model for predicting transformer insulation structure parameters, including the following steps: Collect and preprocess data, including parameter normalization and outlier removal; Establish a feature set, combining the key features of transformer insulation design; and train the model on the training data; Apply the trained model to simulation data to verify its predictive accuracy and applicability in new design solutions, and continuously iterate and optimize the model; The machine learning model based on the random forest algorithm uses a large amount of historical data and simulation data for training, establishes the mapping relationship between input features and target outputs, and gradually optimizes the prediction results; The steps for processing data by the random forest algorithm are as follows: By constructing multiple decision trees and integrating their prediction results, when processing transformer insulation design data, the data set is first divided into a training set and a test set, and the features are standardized; During the training phase, each tree randomly extracts some samples and features from the dataset for modeling; The outputs of all trees are integrated into the final prediction result by voting or averaging; The model is trained using the collected historical data and simulation data. The model learns the mapping relationship between input features and target outputs to gradually optimize the prediction results. Using the collected historical data and simulation data of previous transformer designs, the input features in the data are labeled and matched with the target output to form training samples; During the training process, the model learns the nonlinear mapping relationship between the input features and the target output through continuous iteration, so that the prediction error is gradually reduced; After each iteration, the model adjusts parameters to reduce errors and optimize its prediction ability; ultimately, the prediction results are an evaluation of the insulation performance of the new transformer design; The performance of the model is continuously evaluated during the training process, using cross-validation to ensure that the model does not overfit.

5. The transformer insulation structure design method based on machine learning according to claim 1 is characterized in that: Step S4, applying the trained model to the simulation of the new design solution; The trained model is directly applied to the new transformer design; After inputting new design conditions, the model will output the corresponding insulation structure parameters based on the trained mapping relationship; These prediction results are verified to be reasonable through simulation tools; The simulation verification evaluates whether the insulation structure predicted by the model meets the design specifications by calculating key indicators such as the electric field distribution and electric field distortion rate of the new design. The simulation verification steps are as follows: Simulation verification calculates the electric field distribution and key performance indicators of the new design in the simulation tool by inputting the insulation structure parameters predicted by the model; During the simulation process, the results are compared with the design specifications to ensure that all indicators meet the requirements; if the simulation results show that the electric field distortion rate in some areas is too high or the electric field distribution is uneven, the input parameters are adjusted and the simulation is repeated until the predicted results meet the expected performance in the simulation verification.

6. The transformer insulation structure design method based on machine learning according to claim 1 is characterized in that: Step S5, determining whether the electric field distortion rate, insulation size and electric field uniformity meet the requirements; After completing the simulation verification, the design scheme predicted by the model is evaluated to determine whether it meets the predetermined design requirements; Key indicators include electric field distortion rate, insulation thickness and electric field uniformity; If the simulation results show that the indicators are within the range of the design specifications, the design solution is considered feasible; If the simulation results do not meet the requirements, a feedback mechanism is initiated to analyze the differences between the model prediction results and the actual simulation, and adjust the model or feature selection based on these differences.

Images