A method, system, and storage medium for optimizing capacitor structure design

Abstract

This application relates to the field of electrical design and intelligent optimization calculation technology, and discloses a method, system, and storage medium for capacitor structure design optimization. The method includes: modeling and processing capacitor structure parameters, and partitioning the parameter space to obtain an initial model set; obtaining a multiphysics simulation model based on the geometric model and material properties; obtaining performance indicators such as capacitance, heat distribution, and stress based on the simulation model; setting the objective function and constraints, and running an optimization algorithm to obtain the optimal solution set; subsequently performing feedback control processing on the optimization results, and simultaneously analyzing the convergence path to obtain structural parameters. This application improves the execution efficiency of capacitor structure design optimization and enhances the consistency and stability of the optimization process and performance prediction results.

Description

Technical Field

[0001] This application relates to the field of electrical design and intelligent optimization calculation technology, and in particular to a capacitor structure design optimization method, system and storage medium. Background Technology

[0002] As a core energy storage and filtering component in power electronic devices, the structural design of capacitors directly affects the electrical performance stability, thermal management capability, and mechanical reliability of the devices. With the rapid development of high-frequency, high-voltage, and highly integrated electronic systems, higher requirements are being placed on capacitors in terms of electrical performance and structural adaptability, prompting their structural design to evolve from traditional two-dimensional parameter control to multi-dimensional coupling optimization.

[0003] Current capacitor design schemes primarily rely on engineering experience and finite parameter trial calculations for structural configuration selection, typically using static modeling and single-physics simulations for performance verification. This approach suffers from drawbacks such as low design efficiency, insufficient response accuracy, and unclear structure-performance mapping relationships. Particularly when facing complex material parameters, nonlinear boundary conditions, and coupled physical responses, it struggles to obtain globally optimal structural configurations. While some research has introduced computer-aided simulation methods to enhance modeling and analysis capabilities, these still primarily rely on fixed simulation paths and single-objective optimization. They lack integrated optimization methods based on multi-performance index collaborative control, convergent path feedback, and high-dimensional design space search mechanisms. This results in time-consuming optimization processes that are prone to getting trapped in local optima, severely hindering the rapid development and deployment of capacitors in high-performance applications.

[0004] Therefore, there is an urgent need for an integrated design optimization method with capabilities for structural modeling, multiphysics simulation, performance extraction, global optimization, and error feedback, in order to improve the efficiency and enhance the performance of capacitor structure design. Summary of the Invention

[0005] This application provides a capacitor structure design optimization method, system, and storage medium to improve the structural configuration efficiency and multi-performance response accuracy of capacitors under complex operating conditions, achieve comprehensive optimization goals for electrical performance, thermal stability, and mechanical strength, and solve the problems of low structural modeling efficiency, limited simulation accuracy, and slow optimization convergence in existing capacitor designs.

[0006] In a first aspect, this application provides a capacitor structure design optimization method, the capacitor structure design optimization method comprising:

[0007] The capacitor structural parameters are modeled and then discretized and combined in the model space to obtain a structural configuration model dataset. The capacitor structural parameters include geometric parameters and material properties.

[0008] Based on the geometric parameters and material properties, input conditions are set, and a calculation dataset is obtained through a preset multiphysics simulation model.

[0009] Based on the extraction of the computational dataset, performance index data is obtained, which includes capacitance value, electric field strength, thermal distribution and mechanical stress value;

[0010] Based on the performance index data, set the objective function and constraints, and then execute the optimization algorithm iterative process to obtain the global optimal solution parameter combination. The optimization algorithm iterative process is as follows: set the algorithm type and initialize the parameters, introduce the surrogate model and Bayesian auxiliary algorithm iteration until the iteration termination condition is met, and obtain the global optimal solution parameter combination.

[0011] The parameter combination and performance prediction error of the global optimal solution are analyzed and processed, and the structural parameter configuration is obtained by comparing the convergence termination judgment criteria.

[0012] Secondly, this application provides a capacitor structure design optimization system, the capacitor structure design optimization system comprising:

[0013] The structural modeling module is used to model the structural parameters of the capacitor, and to discretize and combine the structural parameters of the capacitor in the model space to obtain a structural configuration model dataset. The structural parameters of the capacitor include geometric parameters and material properties.

[0014] The simulation analysis module is used to set input conditions based on the geometric parameters and material properties, and obtain a calculation dataset through a preset multiphysics simulation model;

[0015] The performance extraction module is used to extract performance index data based on the computational dataset. The performance index data includes capacitance value, electric field strength, thermal distribution and mechanical stress value.

[0016] The optimization solution module is used to set the objective function and constraints according to the performance index data, and then execute the optimization algorithm iterative process to obtain the global optimal solution parameter combination. The optimization algorithm iterative process is as follows: set the algorithm type and initialize the parameters, introduce the surrogate model and Bayesian auxiliary algorithm iterative process until the iteration termination condition is met, and obtain the global optimal solution parameter combination.

[0017] The result evaluation module is used to analyze and process the parameter combination and performance prediction error of the global optimal solution, and to compare the convergence termination judgment criteria to obtain the structural parameter configuration.

[0018] Thirdly, a computer-readable storage medium is provided, wherein instructions are stored therein, which, when executed on a computer, cause the computer to perform the above-described capacitor structure design optimization method.

[0019] The technical solution provided in this application achieves systematization and standardization of the structural sample generation process by modeling the parametric characteristics of the capacitor structure and constructing a discretized structural variable space. This solves the problems of insufficient structural configuration schemes and incomplete variable coverage in traditional design. By introducing material properties and geometric information to establish a coupled simulation model of electric, thermal, and mechanical fields, the simulation distortion problem caused by the separation of multi-physics response modeling in existing simulation methods is solved. In the simulation solution stage, multiple performance indicators, including capacitance value, electric field strength, heat distribution, and mechanical stress, are extracted, enhancing the comprehensiveness and accuracy of structural performance evaluation. Based on this, the extracted performance data is used for objective function construction and constraint setting. A global optimization algorithm is used to achieve iterative search of the structural parameter space. At the same time, a neural network surrogate model is introduced to replace some high-cost simulation calculations. To reduce computational resource consumption and improve convergence speed during optimization, a Bayesian optimization mechanism is used to guide the search path based on uncertainty prediction, thereby enhancing global search capabilities. During optimization, a learning scheduling strategy is constructed by recording the optimization path and fitness change trends to dynamically adjust the search range and step size, solving the problems of easily getting trapped in local optima and unstable algorithm execution efficiency in high-dimensional optimization. In the optimization result output stage, an error analysis and feedback mechanism is introduced to compare the predicted performance with the actual simulation results and perform secondary judgment and iterative feedback based on the error threshold to ensure the accuracy and feasibility of structural parameters. In the overall design process, the core technical shortcomings of traditional structural optimization methods in terms of modeling efficiency, simulation accuracy, convergence stability, and result verification are effectively overcome, realizing efficient optimization design of capacitor structures and coordinated control of multiple performance objectives. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of one embodiment of the capacitor structure design optimization method in this application.

[0022] Figure 2 This is a diagram illustrating the Bayesian optimization sampling density evolution process in an embodiment of this application.

[0023] Figure 3This is a comparison chart of the convergence performance of the optimized algorithms in the embodiments of this application;

[0024] Figure 4 This is a schematic diagram of one embodiment of the capacitor structure design optimization method system in this application.

[0025] Figure 5 This is a schematic block diagram of the capacitor structure design optimization device in an embodiment of the present invention. Detailed Implementation

[0026] This application provides a method and system for optimizing capacitor structure design. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0027] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of a capacitor structure design optimization method in this application includes:

[0028] Step S1: Model the capacitor structure parameters and discretize and combine the capacitor structure parameters in the model space to obtain a structural configuration model dataset. The capacitor structure parameters include geometric parameters and material properties.

[0029] Specifically, variables such as electrode arrangement, insulation distance, and material properties of the structure are defined as sets. These structural parameters are discretized into several value levels using numerical range discretization. Multiple sets of structural parameter combinations covering the design space are generated using full combination, Latin hypercube, or orthogonal sampling methods. The corresponding 3D model is constructed based on the parameter grouping results. The model dataset uses structural parameters as indexes to record spatial layout and material allocation information, reflecting the diversity of structural samples and facilitating subsequent simulation calls, thus realizing the data flow from parameter space to structural model mapping.

[0030] Step S2: Set input conditions based on the geometric parameters and material properties, and obtain the calculation dataset through a preset multiphysics simulation model.

[0031] Specifically, when extracting geometric parameters and material properties, based on the pre-set geometric variables and material property tables such as medium and electrodes, the dimensional parameters, layer thickness information, and data such as material thermal conductivity, electrical conductivity, and elastic modulus are extracted from the structural model. These data are then uniformly organized according to the field requirements of the input template to generate a structural input configuration table and a material property table.

[0032] When setting the input conditions for the simulation model based on the extracted parameters, electrode excitation, heat source terms, and mechanical constraints are set according to the boundary requirements corresponding to the electric field, thermal field, and force field. A simulation domain is established within the spatial range, and this region is discretized into a high-density or locally refined simulation mesh file. After defining the governing equations for the electric field distribution equation, heat conduction equation, and structural mechanics equilibrium equation, a multiphysics coupling relationship is established, and interactive boundaries and variable mappings are set to construct a complete coupled simulation model. When numerically solving this coupled model, an iterative solver algorithm is used to sequentially obtain the potential distribution, temperature field distribution, and stress field response results, and response data at spatial nodes in each physical domain are extracted to generate a computational dataset.

[0033] Step S3: Based on the extraction of the calculation dataset, performance index data is obtained, including capacitance value, electric field strength, thermal distribution and mechanical stress value.

[0034] Specifically, when extracting the computational dataset, the electric field, thermal field, and mechanical field response data are categorized into independent data channels based on the physical type labels of the simulation output. A data index table is established through structural parameter mapping relationships for rapid extraction and grouping analysis of subsequent performance indicators. In analyzing the electric field distribution data, the potential difference between the anode and cathode regions is calculated by extracting their potential values. Simultaneously, charge density data is extracted at the electrode surface nodes and integrated to obtain the total electrode charge. The capacitance value under the current structural configuration is calculated using the ratio of charge to potential difference. Subsequently, gradient solving is performed on the electric field intensity data to obtain the spatially distributed intensity vector field. Extreme intensity regions are identified and marked in this distribution map for analyzing the electric field concentration trend. In the temperature field analysis, the temperature rise rate and spatial thermal distribution state are obtained by statistically calculating the temperature changes at each node in the time series, and the parameters of the heat accumulation region and the average heat diffusion path are extracted accordingly. Finally, the structural displacement and stress field results are solved jointly. The structural stress response value is obtained based on the nodal strain recovery method, and the stress maximum point is identified by combining the regional coordinates. Thus, the mechanical stress value is extracted to form a complete performance index dataset for optimization control and reliability analysis.

[0035] Step S4: Set the objective function and constraints according to the performance index data, and then execute the optimization algorithm iterative process to obtain the global optimal solution parameter combination. The optimization algorithm iterative process is as follows: set the algorithm type and initialize the parameters, introduce the surrogate model and Bayesian auxiliary algorithm iterative process until the iteration termination condition is met, and obtain the global optimal solution parameter combination.

[0036] Specifically, when setting the objective function and constraints, the objective parameters are first selected based on key parameters in the performance indicators, such as capacitance, electric field strength, temperature rise, and mechanical stress. Weight allocation ratios are then set according to optimization priority or performance requirements to construct an objective function model with a multi-objective expression. Simultaneously, a set of constraint functions, including inequalities and equality expressions, is constructed by combining information such as the physical value range of structural parameters, design tolerances, material limitations, and engineering application requirements. This constraint function restricts the solution space boundary and feasibility domain. The objective function expression and constraint function set are then input into the optimization control module, and initial search points or random populations are configured according to the optimization algorithm type to form the initial optimization state dataset.

[0037] During the optimization algorithm execution, multi-dimensional convergence control conditions, including error threshold, objective function change rate, and maximum number of iterations, are set, and the current solution set is judged in real time to determine whether it meets the termination criteria. Subsequently, the objective function of the structural parameter combination is evaluated based on the selected optimization algorithm to obtain a fitness score dataset. A surrogate model is introduced to predict and screen high-cost simulation steps, obtaining a candidate evaluation solution set to improve iteration efficiency. During the convergence process, the current optimal region is refined. A perturbation mechanism is used to perform perturbation sampling on the local parameter interval, and a Bayesian update strategy is used to reconstruct the sampling distribution based on prediction uncertainty to obtain a refined optimization candidate solution set. During the iterative execution process, the optimization path and intermediate solution state of each round are continuously recorded, and the fitness evolution trajectory and historical search strategy behavior are analyzed. A learning scheduling model is established to dynamically adjust the search parameters. Finally, the structural parameter combination with the optimal objective function, satisfied constraints, and controllable error range is selected from the solution set that meets the convergence conditions as the global optimal solution output, providing a reliable basis for structural configuration and performance prediction.

[0038] Step S5: Analyze and process the parameter combination and performance prediction error of the global optimal solution, and compare the convergence termination judgment criteria to obtain the structural parameter configuration.

[0039] Specifically, based on multiphysics simulation results, the simulation data of each structural model is processed to extract the potential distribution and charge density data of the electrode region, calculate the capacitance value, obtain the extreme value distribution of the electric field intensity, analyze the statistical parameters of the temperature field and the uniformity of thermal distribution, and identify the areas of mechanical stress concentration and their maximum values. All performance indicators are archived according to structural parameter indexes to form a structure-performance mapping database. Subsequently, multi-objective or constrained optimization calculations are performed on this database using objective functions and constraints. Parameter search is performed based on global optimization algorithms and adaptive sampling mechanisms, and the newly generated parameter sets are fed back into the simulation process. The process is iterated until the performance indicators meet the preset targets. The optimized output results are compared with the simulation performance error, and the validity of the solution is judged according to the set accuracy and convergence criteria. This achieves closed-loop control of parameter optimization, performance simulation, and convergence feedback, resulting in a capacitor structure configuration scheme that is optimal in performance, structurally reasonable, and practically feasible.

[0040] In this embodiment, a systematic structural sample set is formed by modeling the structural parameters of the capacitor and constructing a discretized structural variable space to address the problems of insufficient structural schemes and limited variable coverage in traditional design. Geometric parameters and material properties are introduced to construct a coupled simulation model of electric, thermal, and mechanical fields, avoiding response distortion caused by modeling physical fields alone. Based on this model, capacitance, electric field strength, thermal distribution, and mechanical stress indices are extracted to improve the completeness of performance evaluation. Performance data is used to construct the objective function and set constraints, and the structural parameters are iteratively solved through a global optimization algorithm. A surrogate model is introduced to replace some high-overhead simulations, reducing computational costs, and a Bayesian optimization strategy is used to guide the search path based on uncertainty, improving global search efficiency. The optimization path and fitness changes are recorded, and a learning scheduling mechanism is constructed to dynamically adjust the search strategy to improve convergence performance. In the output stage, a convergence judgment and feedback mechanism is established based on the comparison between prediction errors and simulation results to ensure that the final structural parameters meet the feasibility and accuracy requirements. This scheme improves modeling efficiency and simulation accuracy, enhances the stability and reliability of the optimization algorithm, and achieves multi-performance collaborative optimization of the capacitor structure.

[0041] In one specific embodiment, the process of performing step S1 may specifically include the following steps:

[0042] The basic configuration of the capacitor structure is analyzed, and a set of adjustable structural parameters is obtained by extracting and processing geometric parameter features.

[0043] The parameter set is ranged and classified, and the results of the constraint boundary setting are standardized to obtain the structural variable space. The constraint boundary setting results include the range of structural parameters and performance index limits.

[0044] The structure variable space is discretized, and multiple sets of parameter combination samples are generated based on the combination of each variable dimension.

[0045] Modeling is performed based on the parameter combination samples, and the three-dimensional structure generation process is executed in batches to obtain an initial structure model set;

[0046] The initial structural model set is subjected to a unified format conversion process, and the structural configuration model dataset is obtained by organizing the readable model dataset.

[0047] Specifically, the capacitor structure is decomposed into core components such as electrodes, dielectric layer, and end faces. Geometric parameters and material properties, including length, width, thickness, gap, and winding radius, are extracted for these components, forming a set of adjustable structural parameters. Reasonable value ranges are then set for each parameter based on the device's operating environment and actual engineering requirements. Parameters are categorized by function, physical effect, or process feasibility to ensure their distinctiveness and independence. Standardized expression using boundary conditions constructs a complete structural variable space, achieving standardized and unified parameter definitions.

[0048] By discretizing each parameter in the structural variable space according to intervals or step sizes, and employing multivariate combination methods such as full combination, orthogonal experiments, or Latin hypercube sampling, multiple sets of parameter combination samples covering the design space are generated, with each set of parameters corresponding to a specific structure. Using these parameter combination samples, 3D modeling tools are invoked to directly map the structural parameters to 3D structural features, achieving batch automatic generation of models. All structural models are automatically archived using parameter sets as indexes, and material properties and relevant boundary information are synchronously labeled in the model data, ensuring that each model possesses complete physical and geometric information.

[0049] The generated structural models are converted into a unified format and organized into a structural configuration model dataset with universal readability through geometric model simplification, boundary consistency processing and attribute normalization. This dataset supports the migration, reuse and data transfer of structural models between different simulation platforms.

[0050] Taking the design of metallized polypropylene film capacitors as an example, this paper conducts a detailed analysis of its components, including the winding structure, electrode layer, and polypropylene film. Key structural parameters such as metallization layer thickness, polypropylene film thickness, electrode length, electrode spacing, and winding radius are extracted, and the material properties of the metals and polypropylene used are collected. For actual electrical performance and process requirements, reasonable value ranges are set for each parameter, and parameters are classified and standardized based on material properties and structural functions to form a complete structural variable space. Interval discretization and orthogonal experimental design are used to perform multi-dimensional combinations of parameters, automatically generating a large number of structural parameter groups with different configurations. Corresponding 3D structural models are generated in batches using 3D modeling tools, and key information such as parameter indexes, material properties, and end-face structures are synchronously archived for unified model management. All structural models are formatted and standardized to form a structural configuration model dataset that is easy to use in multiphysics simulation analysis, providing efficient and automated data support for multi-performance simulation, optimization, and engineering design of metallized polypropylene film capacitors.

[0051] In one specific embodiment, the process of performing step S2 may specifically include the following steps:

[0052] The geometric parameters and material properties are extracted, and the parameters are organized according to the preset input format requirements;

[0053] The input conditions are set based on the extracted input parameters and boundary conditions, and the simulation region mesh is initialized.

[0054] The electric field distribution equation, heat conduction equation, and mechanical equilibrium equation are modeled separately and coupled to obtain the multiphysics simulation model.

[0055] The multiphysics simulation model is numerically solved, and the response data of each physics field are extracted to obtain the simulation results.

[0056] The simulation results data are formatted and the results of each performance index are labeled and summarized to obtain the computational dataset.

[0057] Specifically, during parameter extraction, parameters such as electrode size, dielectric thickness, material conductivity, dielectric constant, and thermal conductivity are categorized according to simulation requirements. Based on the preset data input template, parameter names, units, and numerical ranges are standardized to establish a list of structural and material parameters. Non-standard or missing data are supplemented and corrected, and parameters are normalized to ensure consistency and standardization in subsequent data flow and automated calls.

[0058] The processed input parameters are mapped to the boundary conditions, and typical boundary conditions such as electrode voltage, end face grounding, heat source distribution, and fixed support are assigned to corresponding structural surfaces or nodes. At the same time, a suitable simulation region mesh is generated according to the geometric characteristics and accuracy requirements of the structural model. An automatic mesh generation tool is used to subdivide the structural domain, realize the spatial discretization of the physical field distribution, and establish a one-to-one index relationship between parameters, boundaries and mesh information in the data structure, providing a high-quality spatial foundation for solving physical equations.

[0059] Based on the aforementioned parameters and mesh information, a set of physical equations for electric field distribution, heat conduction, and mechanical equilibrium are established. Structural and material characteristics are mapped to equation coefficients and initial conditions, constructing a coupled multiphysics simulation model. Numerical methods such as the finite element method are used to automatically solve the model, outputting response data for each physical field. Simulation results for electric field strength, temperature distribution, and stress response are archived according to parameter indexes. The simulation outputs are standardized, and performance indicators such as capacitance, extreme regions, maximum temperature rise, and maximum stress are extracted and labeled. This achieves automatic linkage and archiving of structural parameters, simulation models, and performance data, forming a complete computational dataset that provides data support and basis for subsequent performance analysis and optimization decisions.

[0060] Taking IGBT absorption capacitor design as an example, the capacitor structure design optimization method provided in this application can conduct multi-dimensional design to meet the high requirements for capacitance stability, thermal management capability, and mechanical structural strength under high-frequency and high-voltage operating conditions. By modeling its geometric parameters and constructing a sample space of structural variable combinations, a candidate set of structures is provided for the optimization process. A multiphysics simulation model is used to couple the electric field distribution, heat conduction path, and mechanical stress concentration area, and extract key performance indicators. By constructing an objective function that includes capacitance, electric field strength, temperature rise distribution, and stress response, and introducing a surrogate model and Bayesian optimization strategy for multi-round iterative search, the optimization convergence efficiency is improved and the simulation cost is reduced. A learning scheduling mechanism is used to dynamically adjust the search strategy to adapt to the reliability requirements of the IGBT absorption capacitor under different operating conditions. Through error feedback control, the optimal combination of structural parameters with smaller simulation prediction deviations is selected, enabling the absorption capacitor to maintain stable electrical performance and structural strength while improving energy absorption efficiency, making it suitable for the high reliability operation requirements of power modules, power conversion devices, and other applications.

[0061] In one specific embodiment, the execution process may specifically include the following steps:

[0062] Establish the electric field distribution equation and set the electrode boundary conditions;

[0063] Establish the heat conduction equation and define the material thermal conductivity and heat source term;

[0064] Establish the aforementioned mechanical equilibrium equations and define stress boundaries and displacement constraints;

[0065] Variable correlation processing is performed on the three types of control equations: electrical, thermal, and mechanical, and the interaction terms of the coupled boundary are uniformly processed.

[0066] The coupled control equations are encapsulated, and a multiphysics simulation solution interface is constructed to obtain the multiphysics simulation model.

[0067] Specifically, when constructing the electric field distribution equation, the spatial location parameters of the electrode region are extracted based on the capacitor's geometric model, and a mesh node system is generated. Fixed potential values ​​are applied to the anode and cathode using boundary marking rules, and the boundaries are set to either insulation or conduction types to form a closed electric field region. The potential distribution relationship is constructed using the Poisson equation, and the potential gradient is calculated based on the material's conductivity and charge density parameters for subsequent capacitance and electric field strength analysis. Voltage excitation conditions are set at the electrode boundaries to ensure that the simulation field has an electric field driving source, guaranteeing the integrity of the field distribution calculation and boundary consistency.

[0068] In the process of modeling the heat conduction governing equations, the thermal conductivity parameters of the materials corresponding to the geometry are extracted, and heat source terms are applied to any potential built-in heat sources. Steady-state or transient heat conduction equations are used to describe the heat diffusion process between the medium and the conductor; natural convection or isothermal boundaries are set using thermally conductive boundary conditions to simulate the effects of air cooling. Enhanced meshing is applied to high heat flux density regions to improve simulation accuracy, ensuring that the heat distribution results accurately reflect the temperature rise concentration trend, providing data support for subsequent thermal stability assessment and structural heat resistance optimization.

[0069] In the stage of establishing and coupling the mechanical equilibrium equations, external constraints are applied to the stress-bearing region of the capacitor, especially the electrode interface and the fixed boundary region. Zero displacement points are set for the structure, and free boundaries are defined to simulate the actual assembly state. Based on the fundamental equations of elasticity, a stress-strain field is established. The structural stiffness matrix is ​​constructed using the material's elastic modulus and Poisson's ratio parameters, and its variables are correlated with the temperature load of the thermal field and the electrostatic force of the electric field. Coupling terms are introduced to the intersecting boundaries to achieve unified fusion of electro-thermal-mechanical boundary data. Through the encapsulation of the control equations and the construction of a numerical solution interface, a complete multiphysics coupling model is formed to achieve precise prediction of the structural response behavior under real-world conditions.

[0070] Taking the design of DC-Link capacitors for low-profile PCBs as an example, these capacitors are typically used in compact power modules or inverter circuits, and have strict requirements for structural compactness, uniform electric field distribution, heat conduction efficiency, and stress distribution control. During the design process, an electric field distribution control equation is established to evaluate the potential gradient variation trend under different electrode structures. Fine-grained modeling of electrode spacing, electrode plate area, and conductive paths is performed to accurately predict capacitance values ​​and local electric field strength. Simultaneously, electrode boundary conditions are set to simulate actual operating voltages to ensure consistency of the simulation field. In the heat conduction modeling process, the heat exchange efficiency between the upper and lower surfaces of the capacitor and the PCB is considered. Thermal conductivity parameters between different material layers are introduced, and the mounting contact surface is set as a convection boundary to evaluate the impact of heat accumulation on capacitor performance. Transient thermal field simulation is used to simulate the temperature rise distribution after prolonged energization, guiding material selection and packaging structure optimization. In structural mechanics modeling, stress boundaries and displacement constraints are defined based on solder joint constraints and the internal support structure of the capacitor. These are used to identify potential deformation regions during low-height press-fitting and to perform a synergistic analysis of the structural response induced by thermal expansion. By unifying and coupling the aforementioned electric, thermal, and stress fields and introducing boundary interaction terms, a complete set of governing equations is constructed. This model can accurately predict the capacitive stability, thermal safety, and mechanical reliability of DC-Link capacitors under different package thicknesses and contact methods, providing optimization basis and structural support for their safe application in high-speed, high-density power electronic systems.

[0071] In one specific embodiment, the process of performing step S3 may specifically include the following steps:

[0072] The multiphysics simulation results are classified, and various physical response data are extracted.

[0073] Analyze the electric field distribution data and obtain the capacitance value based on the potential difference and charge density data between the electrodes;

[0074] Gradient calculations are performed on the electric field intensity data, and then the extreme regions are marked to obtain the electric field intensity distribution map;

[0075] Statistical analysis was performed on the temperature field simulation results, and heat distribution parameters were obtained based on the temperature rise trend data.

[0076] The displacement and stress field data are solved, and the mechanical stress value is extracted after identifying the region of maximum stress.

[0077] Specifically, when classifying multiphysics simulation results, electric field, thermal field, and mechanical response data are extracted into independent data channels, and the simulation outputs are automatically grouped based on physical attribute labels. Electric potential distribution, electric field intensity, temperature field data, and stress-strain data are assigned to their corresponding structural variable indices for rapid extraction of subsequent performance indicators. During the classification process, structured operations on multiphysics simulation data are achieved through data field matching and simulation region coordinate filtering. A unified response data interface is established to ensure the accuracy and consistency of the performance extraction process, providing a fundamental data source for performance evaluation and optimization.

[0078] When analyzing and processing the electric field distribution data, the potential values ​​at the electrode boundaries are extracted, and the anode and cathode regions are determined based on position mapping. After numerically calculating the potential difference between the two regions, the charge density data of the nodes on the electrode surface is extracted. By integrating the nodal charge, the total charge value of the electrode is calculated; the capacitance value corresponding to the target structure is obtained by the ratio of charge to potential difference. Simultaneously, this capacitance value is written into a performance index table to construct the correspondence between structural parameters and electrical response performance, and to provide a computational basis for optimizing the objective function.

[0079] When performing gradient field analysis on electric field intensity data, the spatial gradient of the potential field is solved across the entire simulation region to obtain the electric field intensity distribution vector field. Extremum analysis of the intensity gradient identifies regions of concentrated electric field and marks high-intensity distribution zones. Corresponding regions are located in the geometric structure to identify the risk of electric field focusing. Using thermal simulation data, the temperature distribution trend in the simulation time series is analyzed, and the heat accumulation and conduction efficiency are evaluated in conjunction with the temperature rise rate at each node. Coupled analysis of displacement and stress simulation results is performed to extract the location of the maximum stress value and assess the structural failure risk that stress concentration may cause. A complete set of multi-physics performance response indicators is formed for subsequent structural optimization and safety design assessment.

[0080] Taking the design of anti-interference capacitors as an example, these capacitors are typically used to suppress high-frequency noise interference in power lines. They require high stability in capacitance, uniformity of electric field, thermal stability, and structural reliability. During structural design, precise control of electrode arrangement and dielectric thickness is necessary to avoid breakdown failure due to electric field concentration. By establishing the electric field distribution control equation and setting electrode boundary conditions, simulation analysis of potential distribution and electric field intensity under different structural forms is achieved to evaluate the electrical response stability under high-frequency interference conditions. Simultaneously, the thermal resistance relationship between the package shell, leads, and potting material is considered in the thermal conduction modeling, and a heat source term under actual operating current is set to simulate the heat accumulation process. Hotspot regions are identified through temperature field analysis, and heat conduction paths are optimized to reduce the impact of local temperature rise on electrical performance. In the mechanical modeling stage, the stress boundaries that the device may experience under welding, transportation, and vibration conditions are considered, and stress concentration analysis is performed at the electrode-dielectric interface to identify risk points that may lead to microcracks or interlayer delamination. By coupling electric field, thermal field, and mechanical field data, a unified control equation is constructed to evaluate the electrothermal-mechanical integrated response behavior of anti-interference capacitors under interference conditions, providing structural optimization basis and simulation verification means to improve their reliability and service life in electromagnetic compatibility applications.

[0081] In one specific embodiment, the process of performing step S4 may specifically include the following steps:

[0082] Select and assign weights to the target parameters corresponding to the performance indicators, and establish the objective function expression;

[0083] Based on the range of structural parameters, performance requirements, and engineering constraints, a set of constraint functions is constructed.

[0084] Input the objective function expression and the set of constraint functions into the optimization control module to set the initial population or search point;

[0085] Based on the preset iteration upper limit and convergence criterion, the optimization algorithm is driven to iteratively search in the solution space until the termination condition is met;

[0086] The solution set that satisfies the convergence condition in the filtering iteration results is used to obtain the global optimal solution parameter combination.

[0087] Specifically, based on multiple sets of structural response data obtained in the performance extraction stage, target parameters with design-driven significance, such as capacitance value, electric field uniformity, maximum temperature rise, or stress distribution range, are selected from the performance index dataset. According to the design intent, multiple target parameters are weighted to construct a weighted linear or nonlinear objective function expression. The objective function is input into the optimization module in standard mathematical form to drive the search direction. At the same time, the tendency of the optimization objective is controlled in combination with the application scenario to ensure the balance between multiple objectives and the adjustability of design constraints.

[0088] By combining the domain boundaries of structural parameters, material physical limits, manufacturing capabilities, and thermal, electrical, and mechanical safety specifications, the possible value ranges of all structural variables and performance indicators are uniformly organized to construct a multi-dimensional constraint function set for engineering feasibility. This set is used during the optimization process to restrict the search path for parameter combinations and performance responses, preventing non-physical structures or unrealizable designs from entering the computational flow. The objective function and constraint functions are jointly input into the optimization control module, and population initialization parameters or search starting points are set, including control variables such as parameter dimensions, number of individuals, and initial range, establishing an initial solution space distribution for subsequent iterative calculations.

[0089] During the optimization algorithm execution phase, the iterative process is executed according to the set maximum number of iterations, convergence error threshold, and fitness change rules. Multiple rounds of structural parameter space search are carried out. For the parameter set generated in each round, the simulation module is called to perform performance calculation and objective function evaluation. The iteration history is tracked and convergence analysis is performed. High-quality solution sets that meet the convergence criteria are selected from the entire solution set. Solution points that do not meet the constraints or have large fluctuations are removed. The structural parameter combination that is convergent, stable, and has the best performance is output. This solution set is used as the recommended scheme for performance verification and engineering design.

[0090] In one specific embodiment, the optimization algorithm is driven to iteratively search the solution space according to a preset iteration upper limit and convergence criterion until a termination condition is met. The process may specifically include the following steps:

[0091] Set the optimization algorithm type and parameter configuration, initialize the initial search space of the algorithm, and obtain the optimization start state dataset;

[0092] The preset convergence control conditions are judged to obtain the set of criteria for terminating the optimization process. The preset convergence control conditions include the error threshold and the iteration upper limit.

[0093] The search process is performed based on the optimization algorithm, and the fitness score dataset is obtained by evaluating the combination of structural parameters using the objective function. At the same time, a surrogate model is introduced to assist in prediction and obtain candidate evaluation results.

[0094] The current convergence region is analyzed, and the solution space is sampled with local perturbation. Then, a set of candidate solutions for fine optimization is generated using a Bayesian update mechanism.

[0095] Record the iteration path and intermediate solutions for each round, and analyze the optimization convergence trajectory and learning scheduling history to obtain the optimized execution path dataset.

[0096] Specifically, the execution process employs a combination of an improved multi-objective particle swarm optimization algorithm and a surrogate model-assisted mechanism, with the specific objective being to minimize the multi-objective function: ,in This represents a vector of capacitor structure design parameters, such as metallization layer thickness, dielectric layer thickness, and electrode spacing. Indicates the first Several performance objective functions are defined, such as maximizing capacitance, minimizing electric field strength, and minimizing temperature rise. Each objective function is defined based on response data obtained from multiphysics simulations, maintaining consistent physical units; for example, the unit of capacitance is [missing unit]. The electric field strength is V / m, the temperature is K, and the mechanical stress is Pa.

[0097] To perform global optimization, we set the algorithm population size N, the search dimension m, and initialize the speed. With position ,in The search space is determined by the upper and lower limits of the design parameters. set up.

[0098] Setting convergence criteria includes the maximum number of iterations. The rate of change of the objective function is less than a threshold , For the first In the next iteration, the objective function with respect to the design parameters The value, Indicates the first In the next iteration, the objective function is about The value of is such that it satisfies: .

[0099] Update the velocity and position of each particle using the MOPSO iterative formula: ,

[0100] in, Inertia factor For particles In iteration number The velocity vector at the step, For particles In iteration number The updated velocity vector at each step and As a learning factor, for Random numbers within the interval This is the optimal position in the particle's history. This is the current globally optimal position. For particles In iteration number The position vector at each step directly corresponds to the capacitor design parameters, and the new structural parameters are adjusted after each iteration. Input simulation module to calculate its corresponding Construct a fitness score set.

[0101] To further reduce simulation overhead, a surrogate model is introduced, and a radial basis function neural network (RBF-NN) is used to fit the response surface of the objective function. ,in, For Gaussian kernel function, With the center point, As weight, Indicates the first Proxy estimates of performance metrics.

[0102] Select the optimal solution set of the current convergence region, and resample the solution space using the Upper Confidence Bound (UCB) criterion in Bayesian optimization: ,in, and These are the mean and standard deviation of the surrogate model's predictions for the current design point, respectively. Parameters used to control the degree of exploration. For example... Figure 2 As shown, the Bayesian optimization sampling process exhibits an evolutionary trend from broad exploration to hotspot focusing, with the sampling distribution gradually concentrating in the potential optimal region in the parameter space, thereby enhancing the intelligent exploration capability of the understanding space.

[0103] Record the path history of each iteration And construct a scheduling policy function based on the fitness trend: Through analysis rate of change adjustment Parameters are used to dynamically adjust and control the algorithm's behavior. For example... Figure 3 As shown, the improved optimization algorithm outperforms the original algorithm in terms of convergence speed and stability of the objective function, verifying the improved global search capability and iteration efficiency after the synergistic introduction of the scheduling mechanism and the proxy model.

[0104] In one specific embodiment, the process of performing step S5 may specifically include the following steps:

[0105] Extract the structural parameters and performance indices from the global optimal solution parameter combination to form an optimal solution candidate set;

[0106] The relative error value is calculated based on the prediction performance and simulation verification results of the optimal solution candidate set;

[0107] The relative error value is compared with the set accuracy threshold to determine whether the prediction accuracy meets the convergence requirement. For the solution set that does not meet the accuracy requirement, it is returned to the optimization module to re-execute the iteration process.

[0108] The optimal parameter configuration that meets the error range and performance requirements is used as the structural parameter configuration.

[0109] Specifically, the optimal combination of structural parameters output by the optimization algorithm is decoded and restored. The physical field response results are re-evaluated in conjunction with the simulation model, and key performance indicators such as capacitance, electric field strength, heat distribution and stress are extracted simultaneously. The prediction output of the surrogate model is compared, and the relative error percentage of each performance dimension is calculated and an error vector dataset is generated. The error values ​​are then compared with the pre-set accuracy threshold dimension by dimension. Candidate solutions with acceptable prediction performance are selected by Boolean decision method. Samples with high consistency between prediction results and real simulation response are marked, and a mechanism for judging the effectiveness of optimization results is established.

[0110] The structural parameter combinations that fail the error tolerance judgment are returned to the input of the optimization module. The path information and convergence trajectory of the previous round are inherited. The algorithm iteration process is restarted under the same objective function and constraints to maintain search diversity and population activity to avoid getting trapped in local optima. The sampling samples and training frequency of the surrogate model are dynamically adjusted to strengthen the learning density of error sample regions. The algorithm is driven to focus on re-optimizing error-sensitive regions to improve the matching stability between the prediction model and the physical simulation.

[0111] For structural parameter sets that pass the convergence accuracy test and meet the performance target constraints, the results are calibrated and recorded, and automatically archived into the structural optimization result set. Simultaneously, a standardized design configuration document and an input interface file that can be directly called by the modeling platform are generated. This file associates the objective function value, performance index details, error vector, and optimization path identifiers, supporting visualization, simulation reproduction, and result traceability. Ultimately, the final version of the structural parameters, balancing functionality, accuracy, and feasibility, is generated at the structural configuration output end.

[0112] Taking the design of an axial IGBT absorption capacitor as an example, the convergence verification and output configuration of structural optimization parameters are achieved through the aforementioned S5 steps, providing a clear implementation path in engineering. For this type of capacitor, the operating environment involves high-frequency, high-voltage pulse conditions, and capacitance, electric field uniformity, thermal stability, and mechanical reliability are key performance indicators. After optimization, the extracted optimal solution includes multiple parameter combinations such as electrode spacing, metallization layer thickness, and dielectric film stacking structure. By comparing with multiphysics simulation results, the actual value and surrogate predicted value corresponding to the objective function are obtained respectively. The relative error percentage is calculated for each objective dimension, for example, the capacitance error is controlled within 2%, the maximum point error of electric field strength does not exceed 1.5%, and the temperature rise prediction deviation is maintained within 3%.

[0113] When the error of some solutions exceeds the preset convergence threshold, the system sends the set of parameters back to the optimization algorithm module for re-evaluation, and calls new samples from the approximate region to supplement the training of the surrogate model, thereby improving the surrogate accuracy and avoiding "biased convergence". At the same time, the local perturbation mechanism is enabled to generate a new candidate solution set in a small area around the current convergence region for simulation evaluation, ensuring that the optimization process can continue to converge toward a solution with higher accuracy, and continuously collecting information from error hotspot regions to feed back to the model training layer.

[0114] For the optimal structural combination that meets the error requirements, such as a configuration in a certain solution with a metallized electrode thickness of 0.75 μm, a dielectric film thickness of 4.2 μm, and an electrode spacing of 1.1 mm, simulation verification shows that it meets the high-frequency dielectric stability, electric field distribution uniformity, and stress release capability required for IGBT applications. The system marks this configuration as the final optimization result output and associates all corresponding simulation data, optimization path information, and error records, archiving it as reproducible design input. This input can be directly exported to CAD modeling or process development platforms, realizing an integrated closed-loop process of design-verification-mass production. This process improves design efficiency while ensuring the physical feasibility and high reliability of the design solution.

[0115] The capacitor structure design optimization method in the embodiments of this application has been described above. The following is a systematic description of the capacitor structure design optimization method in the embodiments of this application. Please refer to [link / reference]. Figure 4 One embodiment of the capacitor structure design optimization method system in this application includes:

[0116] The structural modeling module 201 is used to model the structural parameters of the capacitor, and to discretize and combine the structural parameters of the capacitor in the space of the model to obtain a structural configuration model dataset. The structural parameters of the capacitor include geometric parameters and material properties.

[0117] The simulation analysis module 202 is used to set input conditions based on the geometric parameters and material properties, and obtain a calculation dataset through a preset multiphysics simulation model.

[0118] The performance extraction module 203 is used to extract performance index data based on the calculation dataset, the performance index data including capacitance value, electric field strength, thermal distribution and mechanical stress value.

[0119] The optimization solution module 204 is used to set the objective function and constraints according to the performance index data, and then execute the optimization algorithm iterative process to obtain the global optimal solution parameter combination. The optimization algorithm iterative process is as follows: set the algorithm type and initialize the parameters, introduce the surrogate model and Bayesian auxiliary algorithm iteratively until the iteration termination condition is met, and obtain the global optimal solution parameter combination.

[0120] The result evaluation module 205 is used to analyze and process the parameter combination and performance prediction error of the global optimal solution, and to compare the convergence termination judgment criteria to obtain the structural parameter configuration.

[0121] Through the collaborative operation of its various components, the system achieves closed-loop optimization of the entire capacitor structure design process, including key stages such as structural modeling, physical simulation, performance extraction, parameter optimization, and result evaluation, effectively improving the automation level and performance accuracy of the design. The structural modeling module 201 provides a parametric configuration model as the system's input foundation; the simulation analysis module 202 solves multi-physics problems based on the modeling results, obtaining multi-dimensional performance data such as electrical, thermal, and mechanical properties; the performance extraction module 203 performs quantitative analysis of the simulation results, extracting key indicators for subsequent optimization; the optimization solution module 204 executes iterative algorithms based on the objective function and constraints to find the optimal structural scheme; and the result evaluation module 205 verifies the optimization results for errors and determines convergence, ensuring the effectiveness and stability of the structural configuration. Through the integrated operation of these modules, the system can accurately and efficiently complete the capacitor structure design optimization task, suitable for collaborative design of structural performance under complex constraints.

[0122] above Figure 4 The capacitor structure design optimization method system in this embodiment of the invention is described in detail from the perspective of modular functional entities. The capacitor structure design optimization device in this embodiment of the invention is described in detail from the perspective of hardware processing.

[0123] Reference Figure 5 This invention also provides a capacitor structure design optimization device, which can be a server, and its internal structure can be as follows: Figure 5 As shown. The capacitor structure design optimization device includes a processor, memory, display screen, input device, network interface, and database connected via a system bus. The processor, designed by the computer, provides computing and control capabilities. The memory of the capacitor structure design optimization device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the capacitor structure design optimization device stores the data corresponding to this embodiment. The network interface of the capacitor structure design optimization device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the above-described method.

[0124] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the capacitor structure design optimization device to which the present invention is applied.

[0125] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the capacitor structure design optimization method.

[0126] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0127] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a capacitor structure design optimization device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0128] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing capacitor structure design, characterized in that, The capacitor structure design optimization method includes: S1. Model the capacitor structure parameters and discretize and combine the capacitor structure parameters in the model space to obtain a structural configuration model dataset, wherein the capacitor structure parameters include geometric parameters and material properties. S2. Set input conditions based on the geometric parameters and material properties, and obtain the calculation dataset through a preset multiphysics simulation model; S3. Based on the extraction of the calculation dataset, performance index data is obtained, including capacitance value, electric field strength, heat distribution and mechanical stress value; S4. Set the objective function and constraints according to the performance index data, and then execute the optimization algorithm iterative process to obtain the global optimal solution parameter combination. The optimization algorithm iterative process is as follows: set the algorithm type and initialize the parameters, introduce the surrogate model and Bayesian auxiliary algorithm iterative process until the iteration termination condition is met to obtain the global optimal solution parameter combination. S5. Analyze and process the parameter combination and performance prediction error of the global optimal solution, and compare the convergence termination judgment criteria to obtain the structural parameter configuration. S4 includes: selecting and weighting the target parameters corresponding to the performance indicators to establish an objective function expression; constructing a set of constraint functions based on the range of structural parameters, performance requirements, and engineering constraints; inputting the objective function expression and the set of constraint functions into the optimization control module to set an initial population or search point; driving the optimization algorithm to iteratively search in the solution space according to a preset iteration upper limit and convergence criterion until the termination condition is met; and filtering the solution set that meets the convergence condition in the iteration results to obtain the global optimal solution parameter combination. The process of driving the optimization algorithm to iteratively search the solution space according to a preset iteration upper limit and convergence criterion until the termination condition is met includes: setting the optimization algorithm type and parameter configuration; initializing the initial search space of the algorithm; obtaining the optimization starting state dataset; judging the preset convergence control conditions to obtain the optimization process termination judgment criterion set, wherein the preset convergence control conditions include an error threshold and the iteration upper limit; performing search processing based on the optimization algorithm, and evaluating the structure parameter combination using an objective function to obtain a fitness score dataset, while introducing a surrogate model for auxiliary prediction to obtain candidate evaluation results; analyzing the current convergence region and performing local perturbation sampling on the solution space, and then using a Bayesian update mechanism to generate a set of candidate solutions for fine optimization; recording the iteration path and intermediate solutions of each round, and analyzing the optimization convergence trajectory and learning scheduling history to obtain an optimization execution path dataset.

2. The capacitor structure design optimization method according to claim 1, characterized in that, S1 includes: The basic configuration of the capacitor structure is analyzed, and a set of adjustable structural parameters is obtained by extracting and processing geometric parameter features. The parameter set is ranged and classified, and the results of the constraint boundary setting are standardized to obtain the structural variable space. The constraint boundary setting results include the range of structural parameters and performance index limits. The structure variable space is discretized, and multiple sets of parameter combination samples are generated based on the combination of each variable dimension. Modeling is performed based on the parameter combination samples, and the three-dimensional structure generation process is executed in batches to obtain an initial structure model set; The initial structural model set is subjected to a unified format conversion process, and the structural configuration model dataset is obtained by organizing the readable model dataset.

3. The capacitor structure design optimization method according to claim 1, characterized in that, S2 includes: The geometric parameters and material properties are extracted, and the parameters are organized according to the preset input format requirements; The input conditions are set based on the extracted input parameters and boundary conditions, and the simulation region mesh is initialized. The electric field distribution equation, heat conduction equation, and mechanical equilibrium equation are modeled separately and coupled to obtain the multiphysics simulation model. The multiphysics simulation model is numerically solved, and the response data of each physics field are extracted to obtain the simulation results. The simulation results data are formatted and the results of various performance indicators are labeled and summarized to obtain the computational dataset.

4. The capacitor structure design optimization method according to claim 3, characterized in that, The process of modeling the electric field distribution equation, heat conduction equation, and mechanical equilibrium equation, and coupling them together to obtain the multiphysics simulation model, includes: Establish the electric field distribution equation and set the electrode boundary conditions; Establish the heat conduction equation and define the material thermal conductivity and heat source term; Establish the aforementioned mechanical equilibrium equations and define stress boundaries and displacement constraints; The variables of the three types of control equations (electric, thermal, and mechanical) are correlated, and the interaction terms of the coupled boundary are uniformly processed. The coupled control equations are encapsulated, and a multiphysics simulation solution interface is constructed to obtain the multiphysics simulation model.

5. The capacitor structure design optimization method according to claim 1, characterized in that, S3 includes: The multiphysics simulation results are classified, and various physical response data are extracted. Analyze the electric field distribution data and obtain the capacitance value based on the potential difference and charge density data between the electrodes; Gradient calculations are performed on the electric field intensity data, and then the extreme regions are marked to obtain the electric field intensity distribution map; Statistical analysis was performed on the temperature field simulation results, and heat distribution parameters were obtained based on the temperature rise trend data. The displacement and stress field data are solved, and the mechanical stress value is extracted after identifying the region of maximum stress.

6. The capacitor structure design optimization method according to claim 1, characterized in that, S5 includes: Extract the structural parameters and performance indices from the global optimal solution parameter combination to form an optimal solution candidate set; The relative error value is calculated based on the prediction performance and simulation verification results of the optimal solution candidate set; The relative error value is compared with the set accuracy threshold to determine whether the prediction accuracy meets the convergence requirement. For the solution set that does not meet the accuracy requirement, it is returned to the optimization module to re-execute the iteration process. The optimal parameter configuration that meets the error range and performance requirements is used as the structural parameter configuration.

7. A capacitor structure design optimization system, used to implement the capacitor structure design optimization method as described in any one of claims 1 to 6, characterized in that, The capacitor structure design optimization system includes: The structural modeling module is used to model the structural parameters of the capacitor, and to discretize and combine the structural parameters of the capacitor in the model space to obtain a structural configuration model dataset. The structural parameters of the capacitor include geometric parameters and material properties. The simulation analysis module is used to set input conditions based on the geometric parameters and material properties, and obtain a calculation dataset through a preset multiphysics simulation model; The performance extraction module is used to extract performance index data based on the computational dataset. The performance index data includes capacitance value, electric field strength, thermal distribution and mechanical stress value. The optimization solution module is used to set the objective function and constraints according to the performance index data, and then execute the optimization algorithm iterative process to obtain the global optimal solution parameter combination. The optimization algorithm iterative process is as follows: set the algorithm type and initialize the parameters, introduce the surrogate model and Bayesian auxiliary algorithm iterative process until the iteration termination condition is met, and obtain the global optimal solution parameter combination. The result evaluation module is used to analyze and process the parameter combination and performance prediction error of the global optimal solution, and to compare the convergence termination judgment criteria to obtain the structural parameter configuration. The optimization solution module is further used to select and weight the target parameters corresponding to the performance indicators, and establish the objective function expression; construct a set of constraint functions based on the range of structural parameters, performance requirements, and engineering constraints; input the objective function expression and the set of constraint functions into the optimization control module, and set the initial population or search point; drive the optimization algorithm to iteratively search in the solution space according to the preset iteration upper limit and convergence criterion until the termination condition is met; filter the solution set that meets the convergence condition in the iteration results, and obtain the global optimal solution parameter combination; The process of driving the optimization algorithm to iteratively search the solution space according to a preset iteration upper limit and convergence criterion until the termination condition is met includes: setting the optimization algorithm type and parameter configuration; initializing the initial search space of the algorithm; obtaining the optimization starting state dataset; judging the preset convergence control conditions to obtain the optimization process termination judgment criterion set, wherein the preset convergence control conditions include an error threshold and the iteration upper limit; performing search processing based on the optimization algorithm, and evaluating the structure parameter combination using an objective function to obtain a fitness score dataset, while introducing a surrogate model for auxiliary prediction to obtain candidate evaluation results; analyzing the current convergence region and performing local perturbation sampling on the solution space, and then using a Bayesian update mechanism to generate a set of candidate solutions for fine optimization; recording the iteration path and intermediate solutions of each round, and analyzing the optimization convergence trajectory and learning scheduling history to obtain an optimization execution path dataset.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is run by the processor, it causes the processor to execute the capacitor structure design optimization method as described in any one of claims 1 to 6.

Images