Vegetable insulating oil breakdown voltage prediction method based on phase field modeling and machine learning

Through phase field modeling and machine learning methods, a breakdown voltage prediction model for plant insulating oil is constructed, which solves the problems of high computational complexity and difficult nonlinear relationship processing in the existing technology, and achieves high-precision and stable breakdown voltage prediction , improves the insulation performance of electrical equipment.

CN119026493APending Publication Date: 2024-11-26GUANGXI UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410861867.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-06-28
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

The calculation complexity of the plant insulating oil breakdown voltage prediction method based on physical models in the prior art is high, the accuracy of model parameters and experimental data affects the prediction results, and it is difficult to deal with complex nonlinear relationships.

Method used

Using a method based on phase field modeling and machine learning, the prediction model of plant insulating oil breakdown voltage is constructed by establishing total free energy equations and phase field models, combining finite element methods and gradient enhancement decision tree algorithms, and processing complex nonlinear relationships And improve prediction accuracy.

Benefits of technology

Improve the prediction accuracy of breakdown voltage, enhance the adaptability and stability of the model, can accurately predict the breakdown voltage of plant insulating oil under different conditions, and guide the oil formula optimization through characteristic importance analysis, improve the electrical equipment's Insulation performance and reliability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119026493A_ABST
    Figure CN119026493A_ABST
Patent Text Reader

Abstract

The invention discloses a method for predicting breakdown voltage of vegetable insulating oil based on phase field modeling and machine learning. The method comprises the following steps: establishing a basic model of the vegetable insulating oil; constructing a total free energy equation; establishing a plant insulating oil phase field model; solving the coupling model by using a finite element method, performing high-throughput simulation and collecting data; constructing a prediction model by adopting a gradient lifting decision tree algorithm; verifying and optimizing the prediction model; obtaining a final prediction model through final model training; through combination of a fluid mechanics equation, a heat conduction equation, a phase field model and a gradient boosting decision tree algorithm, accurate prediction of the breakdown voltage of the vegetable insulating oil is realized, a scientific basis is provided for design, optimization and application of the insulating oil, and the application performance and reliability of the insulating oil in power equipment are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of electrical insulation technology, and in particular to a method for predicting the breakdown voltage of vegetable insulating oil based on phase field modeling and machine learning. Background Art

[0002] As a new type of environmentally friendly insulating material, vegetable insulating oil is widely used in high-voltage equipment such as power transformers. Compared with traditional mineral oil, vegetable insulating oil has the advantages of biodegradability, environmental safety and excellent insulation performance. However, breakdown voltage is one of the important indicators to measure the performance of insulating oil, and its high-precision prediction is of great significance.

[0003] In traditional mineral oil insulation materials, the prediction methods of breakdown voltage are mainly based on experimental measurements and empirical formulas. Although these methods can predict the breakdown voltage to a certain extent, their applicability and accuracy are limited under complex conditions. For example, empirical formulas are difficult to adapt to changing working environments and different oil compositions.

[0004] The method based on physical model describes the flow behavior and temperature distribution of insulating oil by establishing fluid mechanics equations and heat conduction equations. Although this method improves the scientificity and accuracy of prediction, its high computational complexity, the accuracy of model parameters and the reliability of experimental data will also affect the prediction results. In addition, physical models are difficult to handle complex nonlinear relationships. Summary of the invention

[0005] In view of the fact that the prediction method based on physical models in the prior art has high computational complexity, which affects the accuracy of model parameters and the reliability of experimental data, the present invention provides a prediction method for the breakdown voltage of vegetable insulating oil based on phase field modeling and machine learning, which has high accuracy and can handle complex nonlinear relationships. The specific technical solution is as follows: Figures 1 to 3 As shown, the following steps are included:

[0006] S1: Establish a basic model for vegetable insulating oil:

[0007] S2: Construct the total free energy equation, which includes separation energy, gradient energy, electrical energy, Joule heat energy, and flow energy;

[0008] S3: A phase field model of plant insulating oil is established based on the total free energy equation, and the phase field model is used to describe the formation and evolution of the breakdown phase;

[0009] S4: The coupled basic model equations and phase field model are solved using the finite element method, and high-throughput simulation is performed to obtain the flow behavior and breakdown voltage data of vegetable insulating oil under different conditions;

[0010] S5: A prediction model for the breakdown voltage of vegetable insulating oil was constructed using the gradient boosting decision tree algorithm;

[0011] S6: Validate and optimize the prediction model;

[0012] S7: Through the final model training, a final prediction model is obtained, new feature data is input, and the breakdown voltage of the vegetable insulating oil is predicted. The expression of the final prediction model is:

[0013]

[0014] in, is the predicted breakdown voltage; M is the number of decision trees; η is the learning rate; h m (x) is the predicted value of the mth tree.

[0015] Preferably, the step S1 further comprises the following steps:

[0016] S11: Establish a fluid mechanics model of vegetable insulating oil based on its flow behavior and stress distribution under electric field and temperature changes;

[0017] S12: A thermal conduction model of vegetable insulating oil is established based on the temperature change of vegetable insulating oil and its influence on the flow behavior.

[0018] Preferably, the fluid mechanics model expression is:

[0019]

[0020] Where: ρ is the density of vegetable insulating oil; u is the velocity field of vegetable insulating oil; is the gradient field of the velocity field u; p is the pressure field in the vegetable insulating oil; is the gradient field of the pressure field p; is the viscous diffusion term, μ is the dynamic viscosity of the vegetable insulating oil; is the Laplace operator of u, representing the second-order spatial derivative of the velocity field; is the electric field convection term, which indicates the influence of the electric field on the velocity field; ε is the dielectric constant of the vegetable insulating oil; E is the electric field strength in the vegetable insulating oil;

[0021] And / or, the expression of the heat conduction model is:

[0022]

[0023] Where: ρ is the density of vegetable insulating oil; c p is the specific heat capacity of the vegetable insulating oil; u is the velocity field of the vegetable insulating oil, and T is the temperature field in the vegetable insulating oil liquid; is the gradient field of the temperature field; is the Laplace operator of T, representing the second-order spatial derivative of the temperature field; k is the thermal conductivity of the vegetable insulating oil; q· is the heat source term.

[0024] Preferably, the total free energy equation is:

[0025] F = ∫ V [f sep (η(r))+f grad (η(r))+f elec (r)+f Joule (r)+f flow (r)]dV

[0026] Where: f sep (η(r)) is the separation energy; f grad (η(r)) is the gradient energy; f elec (r) is electrical energy; f Joule (r) is Joule heat energy; f flow (r) is the flow energy.

[0027] Preferably, Where W is the interfacial energy coefficient;

[0028] Where k is the gradient energy coefficient; is the gradient of the phase field variable.

[0029] Where, ε(r) is the dielectric constant of the vegetable insulating oil at r, and E is the electric field strength in the vegetable insulating oil;

[0030] Where, σ(r) is the conductivity of the vegetable insulating oil at r; E is the electric field strength in the vegetable insulating oil.

[0031] Among them, ρ(r) is the density of vegetable insulating oil at r, and u is the velocity field of the oil;

[0032] And / or, η(r) is a phase field variable used to describe the phase state at a spatial position r; it ranges from 0 to 1, η(r) of 0 indicates an unbroken phase, and η(r) of 1 indicates a broken phase; r indicates a spatial position.

[0033] Preferably, the phase field model is:

[0034]

[0035] Where: η(r,t) is the phase field variable, η=1 indicates the breakdown phase, η=0 indicates the non-breakdown phase; L0 is the kinetic coefficient; H is the Heaviside function, which is used to ensure that the breakdown phase will grow only when the total energy density exceeds the critical value; f critical is the critical energy density; F is the total free energy.

[0036] Preferably, S6 includes the following steps: using a gradient boosting decision tree algorithm, collecting and preprocessing the simulation data and part of the experimental data obtained in step S4, selecting feature variables and performing data standardization, dividing the data into a training set and a validation set in a ratio of 9:1, using the training set data for model training, and constructing a prediction model.

[0037] Preferably, the characteristic variables include density, viscosity, content of various esters, water content, temperature, breakdown gap, electric field strength and applied voltage of the vegetable insulating oil.

[0038] Preferably, S6 comprises the following steps: using the data from step S5 to evaluate the performance of the prediction model, and calculating the prediction accuracy such as R 2 Score, mean square error (MSE), root mean square error (RMSE), etc.; according to the verification results, adjust the model parameters and the structure of the gradient boosting decision tree algorithm.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] 1. Improve prediction accuracy: By comprehensively considering fluid mechanics, heat conduction and phase field models, the complex behavior of vegetable insulating oil under electric field and temperature changes can be captured, thereby greatly improving the prediction accuracy of breakdown voltage;

[0041] 2. Processing complex nonlinear relationships: The gradient boosting decision tree algorithm can effectively process complex nonlinear relationships between multidimensional features and further improve the prediction ability of the prediction model;

[0042] 3. Strong adaptability and stability: The prediction model has been verified by high-throughput simulation and experimental data, and has strong adaptability and stability, which enhances the reliability and stability of the prediction model and can accurately predict the breakdown voltage of vegetable insulating oil under different conditions;

[0043] 4. Feature importance analysis: Through feature importance analysis, the key factors that have the greatest impact on breakdown voltage can be identified, providing a scientific basis for formula optimization and improvement of vegetable insulating oil;

[0044] 5. Guiding practical applications: The prediction model of the present invention can be used in practical applications to guide the design and optimization of vegetable insulating oils and improve the insulation performance and reliability of electrical equipment, and has important application value and economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a prediction flow chart of a prediction method for the breakdown voltage of vegetable insulating oil based on phase field modeling and machine learning of the present invention;

[0046] Figure 2It is the flow chart of numerical solution of finite element method;

[0047] Figure 3 This is the flowchart of the gradient boosting decision tree algorithm XGBoost;

[0048] Figure 4 This is a comparison chart of the experimental value and predicted value of DC breakdown voltage of soybean oil-based vegetable insulating oil;

[0049] Figure 5 This is a comparison chart between the experimental value and the predicted value of the power frequency breakdown voltage of palm oil-based vegetable insulating oil;

[0050] Figure 6 The breakdown voltage of soybean oil-based vegetable insulating oil varies with temperature;

[0051] Figure 7 It is the deviation percentage between the predicted value and the experimental value when the breakdown voltage of soybean oil-based vegetable insulating oil changes with temperature. DETAILED DESCRIPTION

[0052] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0054] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0055] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0056] Example 1

[0057] S1: Establish a basic model for vegetable insulating oil, including:

[0058] S11: Establish a fluid mechanics model of vegetable insulating oil based on its flow behavior and stress distribution under electric field and temperature changes;

[0059] S12: A thermal conduction model of vegetable insulating oil is established based on the temperature change of vegetable insulating oil and its influence on the flow behavior.

[0060] S2: Construct the total free energy equation, which includes separation energy, gradient energy, electrical energy, Joule heat energy, and flow energy;

[0061] S3: A phase field model of plant insulating oil is established based on the total free energy equation, and the phase field model is used to describe the formation and evolution of the breakdown phase;

[0062] S4: The coupled basic model equations and phase field model are solved using the finite element method, and high-throughput simulation is performed to obtain the flow behavior and breakdown voltage data of vegetable insulating oil under different conditions;

[0063] S5: A prediction model for the breakdown voltage of vegetable insulating oil was constructed using the gradient boosting decision tree algorithm;

[0064] S6: Validate and optimize the prediction model;

[0065] S7: Through the final model training, a final prediction model is obtained, new feature data is input, and the breakdown voltage of the vegetable insulating oil is predicted. The expression of the final prediction model is:

[0066]

[0067] in, is the predicted breakdown voltage; M is the number of decision trees; η is the learning rate; h m (x) is the predicted value of the mth tree.

[0068] Specifically, in the process of using the finite element method to solve the coupled basic model equations and phase field model equations, the boundary conditions are as follows: In the fluid mechanics model, at the entrance of the fluid into the solution domain, a constant or variable velocity is set: u = u inlet , set a constant or variable pressure at the outlet of the fluid leaving the solution domain: p = p outlet ; Where the fluid contacts the solid wall, the velocity is zero u = 0; In the heat conduction model: set a constant temperature on the specified boundary: T = T constant , on an adiabatic boundary, the heat flux is zero: At a boundary with heat convection, the heat flux is proportional to the temperature difference: In the process of high-throughput simulation, the flow behavior and breakdown voltage of vegetable insulating oil under different electric field strengths and temperature conditions are analyzed and data is collected; the prediction model is constructed using the gradient boosting decision tree algorithm; an appropriate gradient boosting decision tree algorithm, such as XGBoost, is selected to collect and preprocess simulation and experimental data, and feature variables are selected: density, viscosity, content of various esters, water content, temperature, breakdown gap, electric field strength and applied voltage of vegetable insulating oil, and data standardization is performed; the training set data is used for model training and a prediction model is constructed. The performance of the model is evaluated using the validation set data, and the prediction accuracy (such as R 2 Score, mean square error (MSE), root mean square error (RMSE), etc.). According to the verification results, the model parameters and structure are adjusted to improve the prediction accuracy and stability. Through the final model training, new feature data is input to predict the breakdown voltage of vegetable insulating oil, and feature importance analysis is performed to obtain the final prediction model.

[0069] In an optional embodiment, the fluid mechanics model expression is:

[0070]

[0071] Where: ρ is the density of vegetable insulating oil; u is the velocity field of vegetable insulating oil; is the gradient field of the velocity field u; p is the pressure field in the vegetable insulating oil; is the gradient field of the pressure field p; is the viscous diffusion term, μ is the dynamic viscosity of the vegetable insulating oil; is the Laplace operator of u, representing the second-order spatial derivative of the velocity field; is the electric field convection term, which indicates the influence of the electric field on the velocity field; ε is the dielectric constant of the vegetable insulating oil; E is the electric field strength in the vegetable insulating oil;

[0072] The expression of the heat conduction model is:

[0073]

[0074] Where: ρ is the density of vegetable insulating oil; c p is the specific heat capacity of the vegetable insulating oil; u is the velocity field of the vegetable insulating oil, and T is the temperature field in the vegetable insulating oil liquid; is the gradient field of the temperature field; is the Laplace operator of T, representing the second-order spatial derivative of the temperature field; k is the thermal conductivity of the vegetable insulating oil; q · is the heat source term.

[0075] In an optional embodiment, the total free energy equation is:

[0076] F = ∫ V [fsep (η(r))+f grad (η(r))+f elec (r)+f Joule (r)+f flow (r)]dV

[0077] Where: f sep (η(r)) is the separation energy; f grad (η(r)) is the gradient energy; f elec (r) is electrical energy; f Joule (r) is Joule heat energy; f flow (r) is the flow energy.

[0078] In an alternative embodiment, Where W is the interfacial energy coefficient;

[0079] Where, k is the gradient energy coefficient; is the gradient of the phase field variable.

[0080] Where, ε(r) is the dielectric constant of the vegetable insulating oil at r, and E is the electric field strength in the vegetable insulating oil;

[0081] Wherein, σ(r) is the conductivity of the vegetable insulating oil at r; E is the electric field strength in the vegetable insulating oil;

[0082] Among them, ρ(r) is the density of vegetable insulating oil at r, and u is the velocity field of the oil;

[0083] η(r) is a phase field variable used to describe the phase state at the spatial position r; it ranges from 0 to 1, η(r) of 0 indicates an unbroken phase, and η(r) of 1 indicates a broken phase; r represents the spatial position.

[0084] In an optional embodiment, S5 includes the following steps: using a gradient boosting decision tree algorithm, collecting and preprocessing the simulation data and part of the experimental data obtained in step S4, selecting feature variables and performing data standardization, dividing the data into a training set and a validation set in a ratio of 9:1, using the training set data for model training, and constructing a prediction model.

[0085] In an optional embodiment, the characteristic variables include density, viscosity, content of various esters, water content, temperature, breakdown gap, electric field strength and applied voltage of the vegetable insulating oil.

[0086] In an optional embodiment, S6 includes the following steps: using the data from step S5, i.e., the validation set data, to evaluate the performance of the prediction model and calculate the prediction accuracy, e.g., R 2Score, mean square error (MSE), root mean square error (RMSE), etc.; based on the verification results, adjust the model parameters and the structure of the gradient boosting decision tree algorithm, such as the depth of the tree, the number of trees, etc.

[0087] The final prediction model is used to predict the breakdown voltage of soybean oil-based vegetable insulating oil at room temperature and pressure when the breakdown gap is 1 mm. The water content is set to 0.1%, the triglyceride ratio is set to: oleic acid 24%, linoleic acid 52%, α-linolenic acid 7%, palmitic acid 11%, stearic acid 6%, and the boost speed is 2 kV / s. A total of 7 measurements are made. The predicted breakdown voltage value is similar to the experimental value as shown in the figure. Figure 4 As shown, Figure 4 The experimental values ​​and predicted values ​​of breakdown voltage are arranged in order from low to high. The average experimental value is 7.46 kV, and the average predicted value is 7.44 kV. The prediction accuracy meets the requirements.

[0088] Example 2

[0089] The difference between this embodiment and embodiment 1 is that the proportion of each triglyceride in the vegetable insulating oil is changed, that is, palm oil-based vegetable insulating oil is used to study the influence of different components on the prediction accuracy of the model. The temperature is set to 10°C, the breakdown gap is 2.5mm, and the voltage rise rate is 2kV / s. The water content is set to 0.1%, the triglyceride proportion is set to: oleic acid 36%, linoleic acid 15%, palmitic acid 45%, stearic acid 4%, and other conditions are the same as in embodiment 1. A total of 6 measurements were made, and the predicted breakdown voltage value is similar to the experimental value as shown in Figure 1. Figure 5 As shown, the experimental average value is 72.09kV, the predicted average value is 72.02kV, and the prediction accuracy meets the requirements.

[0090] Example 3

[0091] The difference between this embodiment and embodiment 1 is that the temperature in the environmental conditions is changed to study the influence of different temperatures on the prediction accuracy of the model. The temperature is set to a temperature gradient of 25°C-70°C, and the measurement is performed every time the temperature rises by 5°C. The other parameters are the same as those in embodiment 1, and a total of 10 measurements are performed. The influence of the breakdown voltage change with temperature on the measured value and the experimental value is shown in Figure 6 The percentage deviation between the measured value and the experimental value is shown in Figure 7 As shown. Figure 7 It can be seen that the prediction error does not exceed 3%, which is within the acceptable range and the prediction accuracy meets the requirements.

[0092] The above description is only a preferred feasible embodiment of the present invention and is not intended to limit the scope of the patent application of the present invention. All equivalent changes, equivalent substitutions or modified changes performed within the technical spirit and principles suggested by the present invention should be included in the scope of patent protection covered by the present invention.

Claims

1. A method for predicting the breakdown voltage of vegetable insulating oil based on phase field modeling and machine learning, characterized in that: The following steps are involved: S1: Establish a basic model for vegetable insulating oil: S2: Construct the total free energy equation, which includes separation energy, gradient energy, electrical energy, Joule heat energy, and flow energy; S3: A phase field model of plant insulating oil is established based on the total free energy equation, and the phase field model is used to describe the formation and evolution of the breakdown phase; S4: The coupled basic model equations and phase field model are solved using the finite element method, and high-throughput simulation is performed to obtain the flow behavior and breakdown voltage data of vegetable insulating oil under different conditions; S5: A prediction model for the breakdown voltage of vegetable insulating oil was constructed using the gradient boosting decision tree algorithm; S6: Validate and optimize the prediction model; S7: Through the final model training, a final prediction model is obtained, new feature data is input, and the breakdown voltage of the vegetable insulating oil is predicted. The expression of the final prediction model is: in, is the predicted breakdown voltage; M is the number of decision trees; η is the learning rate; h m (x) is the predicted value of the mth tree.

2. The method for predicting the breakdown voltage of vegetable insulating oil based on phase field modeling and machine learning according to claim 1, characterized in that: The step S2 further comprises the following steps: S11: Establish a fluid mechanics model of vegetable insulating oil based on its flow behavior and stress distribution under electric field and temperature changes; S12: A thermal conduction model of vegetable insulating oil is established based on the temperature change of vegetable insulating oil and its influence on the flow behavior.

3. The method for predicting the breakdown voltage of vegetable insulating oil based on phase field modeling and machine learning according to claim 2, characterized in that: The fluid mechanics model expression is: Where: ρ is the density of vegetable insulating oil; u is the velocity field of vegetable insulating oil; is the gradient field of the velocity field u; p is the pressure field in the vegetable insulating oil; is the gradient field of the pressure field p; is the viscous diffusion term, μ is the dynamic viscosity of the vegetable insulating oil; is the Laplace operator of u, representing the second-order spatial derivative of the velocity field; is the electric field convection term, which indicates the influence of the electric field on the velocity field; ε is the dielectric constant of the vegetable insulating oil; E is the electric field strength in the vegetable insulating oil; And / or, the expression of the heat conduction model is: Where: ρ is the density of vegetable insulating oil; c p is the specific heat capacity of the vegetable insulating oil; u is the velocity field of the vegetable insulating oil, and T is the temperature field in the vegetable insulating oil liquid; is the gradient field of the temperature field; is the Laplace operator of T, representing the second-order spatial derivative of the temperature field; k is the thermal conductivity of the vegetable insulating oil; q · is the heat source term.

4. The method for predicting the breakdown voltage of vegetable insulating oil based on phase field modeling and machine learning according to claim 1, characterized in that: The total free energy equation is: F=∫ V [f sep (η(r))+f grad (η(r))+f elec (r)+f Joule (r)+f flow (r)]dV Where: f sep (η(r)) is the separation energy; f grad (η(r)) is the gradient energy; f elec (r) is electrical energy; f Joule (r) is Joule heat energy; f flow (r) is the flow energy.

5. The method for predicting the breakdown voltage of vegetable insulating oil based on phase field modeling and machine learning according to claim 4, characterized in that: Where W is the interfacial energy coefficient; Where, k is the gradient energy coefficient; is the gradient of the phase field variable; Where, ε(r) is the dielectric constant of the vegetable insulating oil at r, and E is the electric field strength in the vegetable insulating oil; Wherein, σ(r) is the conductivity of the vegetable insulating oil at r; E is the electric field strength in the vegetable insulating oil; Among them, ρ(r) is the density of vegetable insulating oil at r, and u is the velocity field of the oil; And / or, η(r) is a phase field variable used to describe the phase state at a spatial position r; it ranges from 0 to 1, η(r) of 0 indicates an unbroken phase, and η(r) of 1 indicates a broken phase; r indicates a spatial position.

6. The method for predicting the breakdown voltage of vegetable insulating oil based on phase field modeling and machine learning according to claim 1, characterized in that: The phase field model is: Where: η(r,t) is the phase field variable, η=1 indicates the breakdown phase, η=0 indicates the non-breakdown phase; L0 is the kinetic coefficient; H is the Heaviside function, which is used to ensure that the breakdown phase will grow only when the total energy density exceeds the critical value; f critical is the critical energy density; F is the total free energy.

7. The method for predicting the breakdown voltage of vegetable insulating oil based on phase field modeling and machine learning according to claim 1, characterized in that: The S5 includes the following steps: using the gradient boosting decision tree algorithm, collecting and preprocessing the simulation data and part of the experimental data obtained in step S4, selecting feature variables and performing data standardization, dividing the data into a training set and a validation set in a ratio of 9:1, using the training set data for model training, and building a prediction model.

8. The method for predicting the breakdown voltage of vegetable insulating oil based on phase field modeling and machine learning according to claim 7, characterized in that: The characteristic variables include density, viscosity, content of various esters, water content, temperature, breakdown gap, electric field strength and applied voltage of the vegetable insulating oil.

9. The method for predicting the breakdown voltage of vegetable insulating oil based on phase field modeling and machine learning according to claim 1, characterized in that: S6 includes the following steps: using the data from step S5 to evaluate the performance of the prediction model and calculate the prediction accuracy as R 2 Score, mean square error (MSE), root mean square error (RMSE), etc.; according to the verification results, adjust the model parameters and the structure of the gradient boosting decision tree algorithm.