Vehicle body rigidity performance model optimization system and method, electronic equipment and storage medium
By using deep learning surrogate models and mixed variable processing frameworks, the problems of insufficient fitting ability and poor parameter selection realism in automotive body stiffness performance optimization models are solved, reducing R&D costs and achieving efficient multi-objective optimization and decision support.
Patent Information
- Application Number
- CN202511197698.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-26
- Publication Date
- 2025-12-19
AI Technical Summary
Existing automotive body stiffness performance optimization models suffer from problems such as insufficient model fitting ability, lack of realistic parameter selection, and increased costs due to the use of commercial optimization software.
We employ a deep learning-based surrogate model training module, combined with a neural network incorporating multilayer perceptron, one-dimensional convolution, temporal memory, and self-attention mechanisms. We handle mixed variable types through a hierarchical mapping strategy and utilize an elite-preserving genetic algorithm to construct a mixed variable processing framework for multi-objective optimization and global sensitivity analysis.
It achieves efficient fitting capability for the vehicle body stiffness performance model, improves the realism of parameter selection, reduces R&D costs, and provides multi-objective decision support.
Smart Images
Figure CN121168005A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of three-dimensional design, in particular to a vehicle body stiffness performance model optimization system, a vehicle body stiffness performance model optimization method, a vehicle body stiffness performance model optimization device, an electronic device, a storage medium and a design platform. BACKGROUND
[0002] In the development process of the AI optimization model of the vehicle body stiffness performance, a proxy model is modeled by using data, and optimization code suitable for the scene is developed. However, the existing optimization model development has the following deficiencies:
[0003] Weak model fitting capability: the existing Kriging or simple MLP neural network has the problem of poor fitting capability for complex parameters and insufficient generalization capability;
[0004] Lack of authenticity in parameter selection: in the real production process, the thickness and material selection of some parts may not only have a coupling relationship, but also have a non-continuous problem;
[0005] Use of commercial optimization software: the current mainstream optimization software is commercial paid software, which can only be legally used by purchasing authorization permission through an official channel, increasing the cost of enterprises;
[0006] Therefore, it is urgent to develop an automobile body stiffness performance optimization model using open source Python code, meeting the authenticity of parameter selection and using a deep learning proxy model with stronger fitting capability, so as to improve the development efficiency and reduce the research and development cost. SUMMARY
[0007] The purpose of the present application is to provide a vehicle body stiffness performance model optimization system, a vehicle body stiffness performance model optimization method, a vehicle body stiffness performance model optimization device, an electronic device, a storage medium and a design platform, which at least solve the core problems of poor model fitting capability, poor authenticity of parameter selection, use of commercial paid optimization software to increase cost and other problems in traditional vehicle body stiffness optimization design, and solve one of the technical problems in the problem of parameter sensitivity quantization in multi-objective trade-off.
[0008] The present application provides the following solutions:
[0009] According to a first aspect of the present application, a vehicle body stiffness performance model optimization system is provided, which comprises:
[0010] a data preprocessing module, a proxy model training module, a multi-objective optimization module and a parameter sensitivity analysis module;
[0011] The data preprocessing module is used to process mixed variable types in vehicle body design through a hierarchical mapping strategy;
[0012] The agent model training module is configured to construct a multi-element deep learning architecture integrating four neural networks, i.e., a multi-layer perceptron, a one-dimensional convolution, a time memory, and a self-attention mechanism.
[0013] The multi-objective optimization module is configured to construct a mixed variable processing framework based on an elite reservation genetic algorithm.
[0014] The sensitivity analysis module is configured to quantify design influences by using global sensitivity analysis.
[0015] Further, the mixed variable type in the vehicle body design is processed by a hierarchical mapping strategy, which includes:
[0016] The discrete engineering parameters are converted according to a material and thickness mapping relationship library to realize automatic conversion and mapping of the thickness variable and the material variable.
[0017] The stiffness and safety subsets are intelligently separated from the original data, and the geometric variable and the thickness / material variable are jointly normalized, and a user-defined normalization range is supported.
[0018] The geometric variable is a continuous variable, and the thickness / material variable is a discrete variable.
[0019] A data integration engine is developed to realize automatic alignment and joint normalization of the torque stiffness and the vehicle body mass data, to provide standardized multidisciplinary input data for the agent model, and to ensure the collaborative processing capability of different dimension parameters.
[0020] Further, the multi-element deep learning architecture integrating the four neural networks includes:
[0021] An end-to-end model optimization is realized by an automatic training framework.
[0022] A dynamic learning rate scheduler combined with an early stopping mechanism is used to optimize the training process.
[0023] A dual-target verification module is developed to simultaneously monitor the prediction accuracy of the torque stiffness and the mass.
[0024] A displacement distribution curve and a predicted-real value scatter matrix are generated to realize model performance visualization.
[0025] The model parameter saving and the training process backtracking are supported to ensure experimental reproducibility and provide a high-precision prediction model for further optimization.
[0026] Further, the mixed variable processing framework based on the elite reservation genetic algorithm includes:
[0027] For the geometric variable, a real number coding combined with a specific evolution operator is used.
[0028] An index mapping mechanism is designed for thickness / material variables to realize discrete value conversion through a predefined engineering combination library;
[0029] wherein,
[0030] In the optimization problem definition, constraint automatic checking is implemented to avoid invalid engineering combinations;
[0031] wherein,
[0032] The optimization process outputs a Pareto frontier solution set, automatically generates a structured report containing design variable values, torque stiffness values and vehicle body mass data, and visually displays the optimal trade-off scheme of lightweighting and stiffness through visualization to provide multi-objective decision support.
[0033] Further, the global sensitivity analysis quantifies the design impact, including:
[0034] A multi-objective sensitivity evaluation system is constructed;
[0035] The multi-objective sensitivity evaluation system includes identifying key parameters through core indicators and positioning multi-objective conflict points using conflict indexes;
[0036] It also includes developing a triple visualization tool;
[0037] The triple visualization tool includes a distributed scatter plot to display parameter impact by target, a horizontal bar chart to label trade-off relationships, and a heat map to associate engineering variables and actual impact;
[0038] It also includes outputting a structured report of the analysis results;
[0039] The structured report of the analysis results includes original sensitivity values, relative impact degrees and conflict markers, providing quantitative decision-making basis for design iteration and clarifying parameter adjustment priorities.
[0040] According to a second aspect of the present application, a vehicle body stiffness performance model optimization method is provided, which includes:
[0041] The steps of preparing data, pre-processing and model training, multi-objective optimization and parameter sensitivity analysis;
[0042] The step of preparing data includes preparation of original design data and auxiliary configuration files;
[0043] The step of pre-processing and model training includes intelligent data preprocessing and model training by a surrogate model, and outputting model results;
[0044] The step of multi-objective optimization includes optimization execution by NSGA-II after mixed variable optimization configuration, and outputting optimization results;
[0045] The parameter sensitivity analysis steps include global sensitivity sampling, sensitivity quantification analysis, output of analysis results, and triple visualization of the output results.
[0046] Further, the preparation of the original design data includes:
[0047] The preparation includes a CSV file mixed with design variables, including continuous variables of geometric parameters, thickness / material parameters;
[0048] Further, a mapping file of thickness / material parameters is prepared;
[0049] The mapping file of thickness / material parameters is an Excel format file used to define the engineering constraint relationship between material ID and actual thickness;
[0050] Further, a variable definition file is prepared;
[0051] The variable definition file is an Excel file format used to clearly define the variables required for stiffness / safety analysis and their value ranges;
[0052] Further, a torque / quality label file is prepared;
[0053] The torque / quality label file includes Excel data of torque stiffness and actual values of vehicle body mass.
[0054] According to a third aspect of the present application, an electronic device is provided, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus;
[0055] The memory stores a computer program, which, when executed by the processor, causes the processor to execute the steps of the vehicle body stiffness performance model optimization method.
[0056] According to a fourth aspect of the present application, a computer readable storage medium is provided, which stores a computer program executable by an electronic device, and when the computer program runs on the electronic device, the electronic device executes the steps of the vehicle body stiffness performance model optimization method.
[0057] According to a fifth aspect of the present application, a design platform is provided, comprising:
[0058] An electronic device is used to implement the steps of the vehicle body stiffness performance model optimization method;
[0059] A processor runs a program, and when the program runs, the data output from the electronic device executes the steps of the vehicle body stiffness performance model optimization method;
[0060] A storage medium for storing a program, which, when executed, performs the steps of the vehicle body stiffness performance model optimization method on data output from the electronic device.
[0061] Through the above scheme, the following beneficial technical effects are obtained:
[0062] The present application solves the core problems in traditional vehicle body stiffness optimization design, such as poor model fitting ability, poor parameter selection authenticity, and increased cost of using commercial paid optimization software, by developing automatic agent model training code, automatic single-discipline optimization, and automatic parameter sensitivity analysis code, and constructing a full-process automatic vehicle body stiffness optimization design system.
[0063] The present application realizes automatic mapping and normalization of thickness / material variables through an intelligent data preprocessing module.
[0064] The present application realizes automatic optimization of vehicle body multivariate space by constructing a mixed variable multi-objective optimization framework, combining NSGA-II algorithm and mixed variable processing strategy, and avoids the inefficiency and local optimal risk of traditional manual trial and error.
[0065] The present application solves the parameter sensitivity quantification problem in multi-objective trade-off by developing a global sensitivity analysis module based on the Morris method, automatically identifying key design parameters and conflict parameters, and providing design guidance for engineers. BRIEF DESCRIPTION OF DRAWINGS
[0066] Figure 1 is a structural diagram of a vehicle body stiffness performance model optimization system provided by one or more embodiments of the present application.
[0067] Figure 2 is a flowchart of a vehicle body stiffness performance model optimization method provided by one or more embodiments of the present application.
[0068] Figure 3 is a structural diagram of a vehicle body stiffness performance model optimization device provided by one or more embodiments of the present application.
[0069] Figure 4 is a schematic diagram of the system architecture provided by one specific embodiment of the present application.
[0070] Figure 5 is a schematic diagram of label preprocessing provided by one specific embodiment of the present application.
[0071] Figure 6 is a schematic diagram of input preprocessing provided by one specific embodiment of the present application.
[0072] Figure 7 is a schematic diagram of the agent model training module provided by one specific embodiment of the present application.
[0073] Figure 8 is a schematic diagram of a multi-objective optimization module provided by one embodiment of the present application.
[0074] Figure 9 is a schematic diagram of a parameter sensitivity analysis module provided by one embodiment of the present application.
[0075] Figure 10 is a structural block diagram of an electronic device of a vehicle body stiffness performance model optimization method provided by one or more embodiments of the present application. DETAILED DESCRIPTION
[0076] The technical solutions of the present application will be described clearly and completely below with reference to the drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.
[0077] Figure 1 is a structural diagram of a vehicle body stiffness performance model optimization system provided by one or more embodiments of the present application.
[0078] As shown in Figure 1 , the vehicle body stiffness performance model optimization system comprises:
[0079] a data preprocessing module, a proxy model training module, a multi-objective optimization module and a sensitivity analysis module;
[0080] The data preprocessing module is used to process mixed variable types in vehicle body design through a hierarchical mapping strategy.
[0081] The proxy model training module is used to construct a multi-element deep learning architecture integrating four kinds of neural networks, i.e., a multi-layer perceptron, a one-dimensional convolution, a time memory and a self-attention mechanism.
[0082] The multi-objective optimization module is used to construct a mixed variable processing framework based on an elite reservation genetic algorithm.
[0083] The sensitivity analysis module is used to quantify design influences by using global sensitivity analysis.
[0084] Specifically, the present application is an automatic vehicle body stiffness optimization platform based on deep learning and multi-objective optimization, which adopts a modular architecture design, as shown in Figure 4 , comprising four core modules of data intelligent preprocessing, proxy model training, multi-objective optimization and parameter sensitivity analysis, and realizing full-process automatic control through Python.
[0085] Between the present application, through the development of automated agent model training code, automated single discipline optimization, automatic parameter sensitivity analysis code, build the whole process of automatic body stiffness optimization design system, solve the traditional body stiffness optimization design model fitting ability is poor, parameter selection authenticity is poor, use commercial paid optimization software to increase the cost and other core problems.
[0086] Through the intelligent data preprocessing module, the automatic mapping and normalization of thickness / material variables are realized. By constructing a mixed variable multi-objective optimization framework, combining the NSGA-II algorithm and the mixed variable processing strategy, the automatic optimization of the body multi-variable space is realized, avoiding the inefficiency and local optimal risk of traditional manual trial and error. Further development of a global sensitivity analysis module based on the Morris method automatically identifies key design parameters and conflict parameters, providing design guidance for engineers and solving the parameter sensitivity quantification problem in multi-objective trade-off.
[0087] In the present embodiment, the mixed variable types in the body design are processed by a hierarchical mapping strategy, including:
[0088] According to the material and thickness mapping relationship library, the discrete engineering parameters are converted to realize the automatic conversion and mapping of thickness variables and material variables;
[0089] The stiffness and safety subsets are intelligently separated from the original data, and the geometric variables and thickness / material variables are jointly normalized to support custom normalization range;
[0090] Among them, the geometric variable is a continuous variable, and the thickness / material variable is a discrete variable;
[0091] A data integration engine is developed to realize automatic alignment and joint normalization of torque stiffness and body mass data, providing standardized multidisciplinary input data for the proxy model and ensuring the collaborative processing capability of different dimension parameters.
[0092] Specifically, in one specific embodiment, as shown in Figure 5 , 6 The data preprocessing module processes the mixed variable types in the body design through a hierarchical mapping strategy. The automatic conversion and mapping of thickness variables and material variables are realized, and the discrete engineering parameters are converted into continuous design variables according to the safety material and thickness mapping relationship library. The stiffness and safety subsets are intelligently separated from the original data, and the geometric variables (continuous type) and thickness / material variables (discrete type) are jointly normalized to support custom normalization range. A data integration engine is developed to realize automatic alignment and joint normalization of torque stiffness and body mass data, providing standardized multidisciplinary input data for the proxy model and ensuring the collaborative processing capability of different dimension parameters.
[0093] In this embodiment, a multi-element deep learning architecture is constructed, integrating four kinds of neural networks including multilayer perceptron, one-dimensional convolution, time series memory and self-attention mechanism.
[0094] An end-to-end model optimization is realized through an automated training framework.
[0095] Among them, a dynamic learning rate scheduler combined with an early stopping mechanism is used to optimize the training process.
[0096] A double-target verification module is developed to monitor the prediction accuracy of torque stiffness and mass simultaneously.
[0097] A displacement distribution curve and a predicted-real value scatter matrix are generated to realize model performance visualization.
[0098] Among them, model parameter saving and training process backtracking are supported to ensure experimental reproducibility and provide high-precision prediction models for further optimization.
[0099] Specifically, in one specific embodiment, as shown in Figure 7 The agent model training module constructs a multi-element deep learning architecture, integrating four kinds of neural networks including multilayer perceptron, one-dimensional convolution, time series memory and self-attention mechanism. An end-to-end model optimization is realized through an automated training framework: a dynamic learning rate scheduler combined with an early stopping mechanism is used to optimize the training process; a double-target verification module is developed to monitor the prediction accuracy of torque stiffness and mass simultaneously; a displacement distribution curve and a predicted-real value scatter matrix are generated to realize model performance visualization. Model parameter saving and training process backtracking are supported to ensure experimental reproducibility and provide high-precision prediction models for optimization.
[0100] In this embodiment, a hybrid variable processing framework is constructed based on the elite reservation genetic algorithm, including:
[0101] For geometric variables, real number coding is used combined with specific evolution operators.
[0102] For thickness / material variables, an index mapping mechanism is designed to realize discrete value conversion through a pre-defined engineering combination library.
[0103] Among them,
[0104] In the optimization problem definition, automatic constraint checking is realized to avoid invalid engineering combinations.
[0105] Among them,
[0106] The optimization process outputs a Pareto frontier solution set, automatically generates a structured report containing design variable values, torque stiffness values and vehicle body mass data, and visually displays the optimal trade-off scheme of lightweight and stiffness through visualization, providing multi-objective decision support.
[0107] Specifically, in one specific embodiment, as shown inFigure 8 As shown, the multi-objective optimization module constructs a mixed variable processing framework based on an elitist reserved genetic algorithm. Real number coding is adopted for geometric variables combined with specific evolution operators; an index mapping mechanism is designed for thickness / material variables to realize discrete value conversion through a pre-defined engineering combination library. Automatic constraint checking is realized in the optimization problem definition to avoid invalid engineering combinations. The optimization process outputs a Pareto frontier solution set, automatically generates a structured report containing design variable values, torque stiffness values and vehicle body mass data, and visually displays the optimal trade-off scheme of lightweighting and stiffness through visualization to provide multi-objective decision support.
[0108] In the embodiment, global sensitivity analysis is used to quantify design influence, including:
[0109] A multi-objective sensitivity evaluation system is constructed.
[0110] The multi-objective sensitivity evaluation system includes identifying key parameters through core indicators and positioning multi-objective conflict points using conflict indexes.
[0111] It also includes developing a triple visualization tool.
[0112] The triple visualization tool includes a distributed scatter plot to display parameter influence by target, a horizontal bar chart to label trade-off relationships, and a heat map to associate engineering variables with actual influence.
[0113] It also includes outputting a structured report of analysis results.
[0114] The structured report of analysis results includes original sensitivity values, relative influence degrees, and conflict markers, providing quantitative decision-making basis for design iteration and clarifying parameter adjustment priorities.
[0115] Specifically, in one specific embodiment, as shown in Figure 9 The parameter sensitivity analysis module uses global sensitivity analysis to quantify design influence. A multi-objective sensitivity evaluation system is constructed by identifying key parameters through core indicators and positioning multi-objective conflict points using conflict indexes. A triple visualization tool is developed, including a distributed scatter plot to display parameter influence by target, a horizontal bar chart to label trade-off relationships, and a heat map to associate engineering variables with actual influence. Analysis results are output in a structured report, including original sensitivity values, relative influence degrees, and conflict markers, providing quantitative decision-making basis for design iteration and clarifying parameter adjustment priorities.
[0116] Figure 2 is a flowchart of a vehicle body stiffness performance model optimization method provided by one or more embodiments of the present application.
[0117] The vehicle body stiffness performance model optimization method shown in Figure 2 includes:
[0118] Step S1 of preparing data, step S2 of preprocessing and model training, step S3 of multi-objective optimization and step S4 of parameter sensitivity analysis;
[0119] Step S1 of preparing data includes preparation of original design data and auxiliary configuration files;
[0120] Step S2 of preprocessing and model training includes training by an agent model after intelligent data preprocessing, and output of model results;
[0121] Step S3 of multi-objective optimization includes optimization by NSGA-II after optimization configuration of mixed variables, and output of optimization results;
[0122] Step S4 of parameter sensitivity analysis includes sensitivity quantification analysis after global sensitivity sampling, analysis result output and triple visualization of output results.
[0123] In the embodiment, the preparation of original design data includes:
[0124] The CSV file including mixed design variables includes continuous variables of geometric parameters, thickness / material parameters;
[0125] Further, a mapping file of thickness / material parameters is prepared;
[0126] The mapping file of thickness / material parameters is an Excel format file, which is used to define the engineering constraint relationship between material ID and actual thickness;
[0127] Further, a variable definition file is prepared;
[0128] The variable definition file is an Excel file format, which is used to clearly define the variables required for stiffness / safety analysis and their value ranges;
[0129] Further, a torque / quality label file is prepared;
[0130] The torque / quality label file includes Excel data of true values of torque stiffness and vehicle body mass.
[0131] Specifically, in one specific embodiment, the use process of the full-process automatic vehicle body stiffness optimization design system is disclosed.
[0132] 1. Data preparation.
[0133] 1.1 Preparation of original design data: preparation of a CSV file (such as lhs_samples_all.csv) containing mixed design variables, including geometric parameters (continuous variables) and thickness / material parameters (discrete variables);
[0134] Material-thickness mapping file: Excel file defining the engineering constraint relationship between material ID and actual thickness;
[0135] Variable definition file: Excel file specifying the variables required for stiffness / safety analysis and their value ranges;
[0136] Torque-mass label file: Excel data including the true values of torque stiffness and vehicle body mass.
[0137] 1.2 Auxiliary configuration file;
[0138] Geometric variable boundary definition: records the engineering feasible range of each geometric parameter;
[0139] Material thickness combination library: stores the discrete value space of material thickness combinations;
[0140] Normalization parameter configuration: set the normalization range of geometric variables, thickness variables, mass / torque.
[0141] 2 Data preprocessing and model training.
[0142] Run data processing and training program:
[0143] 2.1 Intelligent data preprocessing: automatically identify and separate the data subsets of stiffness analysis and safety analysis;
[0144] Convert thickness / material discrete variables, follow engineering constraints;
[0145] Jointly normalize geometric parameters and thickness / material parameters;
[0146] Integrate torque and mass data for collaborative normalization processing;
[0147] 2.2 Proxy model training: automatically divide training set / validation set / test set (70% / 20% / 10% ratio);
[0148] Parallel training of four types of neural network models (MLP / CNN / LSTM / Self-Attention);
[0149] Dynamically adjust learning rate and implement early stopping strategy to optimize training process;
[0150] Real-time generation of prediction accuracy visualization charts (R² index, predicted-real value scatter plot);
[0151] 2.3 Output model results: generate optimal model parameter file (.pth format);
[0152] Output training process log and performance evaluation report;
[0153] Save the test set prediction results and visualize the chart.
[0154] 3 Multi-objective optimization.
[0155] Run the optimization solver:
[0156] 3.1 Mixed variable optimization configuration: Set optimization objectives: maximize torque stiffness, minimize body mass;
[0157] Define design variable space: geometric parameters (continuous space), thickness / material (discrete combination);
[0158] Load the pre-trained surrogate model as the objective function evaluator.
[0159] 3.2 NSGA-II optimization execution: Real number coding for geometric variables, integer index for thickness / material combination;
[0160] Implement constraint automatic checking to avoid invalid engineering combinations; parallel evaluation of torque stiffness and body mass of design scheme.
[0161] 3.3 Optimization result output: Generate Pareto front solution set CSV file (including design variable values, torque values, mass values); automatically draw three-dimensional Pareto front visualization chart; output optimization process log and performance analysis report.
[0162] 4 Parameter sensitivity analysis.
[0163] Run the sensitivity analysis program:
[0164] 4.1 Global sensitivity sampling: Use the Morris method to generate design variable perturbation samples; calculate the response values of torque / mass through the surrogate model;
[0165] 4.2 Sensitivity quantification analysis: Calculate the μ* sensitivity index of each variable to torque stiffness; calculate the μ* sensitivity index of each variable to body mass; identify key influencing parameters (above a certain threshold); detect conflicting parameters (significantly affect both objectives simultaneously);
[0166] 4.3 Analysis result output: Generate sensitivity analysis CSV report (including original values, relative sensitivity, conflict flag);
[0167] 4.4 Output triple visualization results: σ-μ* scatter plot shows parameter sensitivity distribution; conflict parameter horizontal bar chart labels trade-off relationship; heat map correlates engineering variables with actual impact; provide key parameter adjustment suggestion report.
[0168] The core technology route of the above embodiments is: 1. Prepare data → 2. Preprocessing and model training → 3. Multi-objective optimization → 4. Parameter sensitivity analysis.
[0169] End-to-end automation: seamless connection from raw data to optimized decision-making throughout the whole process;
[0170] Engineering constraint guarantee: strictly follow the material-thickness combination and other engineering practical constraints;
[0171] Standardized output: structured data report + visualization chart to support design decision-making;
[0172] Breakthrough performance: compress traditional optimization period of several weeks to minutes.
[0173] Figure 3 Figure 1 is a structural diagram of a vehicle body stiffness performance model optimization device provided by one or more embodiments of the present application.
[0174] As shown in Figure 3 The vehicle body stiffness performance model optimization device includes:
[0175] a data preparation module, a preprocessing and model training module, a target optimization module, and a parameter sensitivity analysis module;
[0176] The data preparation module is used for preparation of original design data and auxiliary configuration files.
[0177] The preprocessing and model training module is used for training by an agent model after intelligent data preprocessing, and outputs model results.
[0178] The target optimization module is used for optimization execution by NSGA-II after mixed variable optimization configuration, and outputs optimization results.
[0179] The parameter sensitivity analysis module is used for sensitivity quantification analysis after global sensitivity sampling, and outputs analysis results and performs triple visualization on the output results.
[0180] It is worth noting that although the present system / device only discloses the data preparation module, the preprocessing and model training module, the target optimization module, and the parameter sensitivity analysis module, it does not mean that the device is limited to the above basic functional modules. On the contrary, the present application intends to mean that on the basis of the above basic functional modules, those skilled in the art can add one or more functional modules to form infinite embodiments or technical solutions in combination with existing technologies. That is to say, the present system / device is open rather than closed, and it cannot be considered that the protection scope of the present application is limited to the above disclosed basic functional modules just because the present embodiment only discloses individual basic functional modules.
[0181] Technical route of the present application:
[0182] 1. A mixed variable multi-objective collaborative optimization framework: achieve the collaborative optimization of continuous geometric parameters and discrete material-thickness combinations, and innovatively construct a dual-path mixed variable processing mechanism. Real number coding is adopted for geometric variables in combination with evolutionary algorithms, and an engineering combination library index mapping technology is introduced for material thickness to automatically check engineering feasibility and avoid invalid designs during optimization evaluation, and a torsional stiffness-mass Pareto frontier surface is generated to provide visual decision basis for lightweighting and stiffness balance.
[0183] 2. A global sensitivity-driven conflict parameter quantification technology: break through the parameter sensitivity quantification bottleneck in multi-objective trade-off, and construct a torque stiffness and vehicle body mass parallel evaluation channel. A conflict index positioning algorithm is innovatively designed to automatically identify key conflict points, output a three-dimensional diagnostic system including a distribution scatter plot, a trade-off bar chart and an influence heat map, realize visual and accurate positioning of parameter sensitivity, and provide quantitative priority guidance for design iteration.
[0184] 3. A mixed optimization architecture based on an open source ecosystem: integrate the PyTorch deep learning framework and the pymoo multi-objective optimization library to construct an open source collaborative computing engine. PyTorch is used to realize efficient training and deployment of a neural network proxy model, NSGA-II algorithm of pymoo is used to realize multi-objective optimization of mixed variables, and a seamless open source technology chain from parameterized modeling to Pareto frontier search is established to break through the dependence barrier of commercial software.
[0185] Figure 10 is a structural block diagram of an electronic device provided by one or more embodiments of the body stiffness performance model optimization method.
[0186] As shown in Figure 10 , the present application provides an electronic device, comprising: a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus;
[0187] The memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the body stiffness performance model optimization method.
[0188] The present application also provides a computer readable storage medium storing a computer program executable by an electronic device, which makes the electronic device execute the steps of the body stiffness performance model optimization method when the computer program runs on the electronic device.
[0189] The present application also provides a design platform, comprising:
[0190] An electronic device for implementing the steps of the body stiffness performance model optimization method;
[0191] The processor runs a program, and the program performs the steps of the vehicle body stiffness performance model optimization method when the program runs on the data output from the electronic device.
[0192] The storage medium is used to store the program, and the program performs the steps of the vehicle body stiffness performance model optimization method when the program runs on the data output from the electronic device.
[0193] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.
[0194] The electronic device includes a hardware layer, an operating system layer running on the hardware layer, and an application layer running on the operating system. The hardware layer includes hardware such as a central processing unit (CPU), a memory management unit (MMU), and a memory. The operating system can be any one or more computer operating systems that realize the control of the electronic device through a process, such as a Linux operating system, a Unix operating system, an Android operating system, an iOS operating system, or a windows operating system, etc. In the embodiments of the present application, the electronic device can be a handheld device such as a smart phone or a tablet computer, or an electronic device such as a desktop computer or a portable computer, which is not particularly limited in the embodiments of the present application.
[0195] The execution subject of the electronic device control in the embodiments of the present application can be an electronic device, or a functional module in the electronic device that can call and execute a program. The electronic device can obtain a firmware corresponding to the storage medium, and the firmware corresponding to the storage medium is provided by a supplier. The firmware corresponding to different storage media can be the same or different, which is not limited herein. After the electronic device obtains the firmware corresponding to the storage medium, the electronic device can write the firmware corresponding to the storage medium into the storage medium, specifically, burn the firmware corresponding to the storage medium into the storage medium. The process of burning the firmware into the storage medium can be realized by using the prior art, which is not described in detail in the embodiments of the present application.
[0196] The electronic device can also obtain a reset command corresponding to the storage medium, and the reset command corresponding to the storage medium is provided by a supplier. The reset command corresponding to different storage media can be the same or different, which is not limited herein.
[0197] At this time, the storage medium of the electronic device is the storage medium in which the corresponding firmware is written, and the electronic device can respond to the reset command corresponding to the storage medium in the storage medium in which the corresponding firmware is written, so that the electronic device resets the storage medium in which the corresponding firmware is written according to the reset command corresponding to the storage medium. The process of resetting the storage medium according to the reset command can be implemented in the prior art, and will not be described in detail in the embodiments of the present application.
[0198] For the convenience of description, the above device is described as various units and modules in function. Of course, the functions of the units and modules can be implemented in one or more software and / or hardware in the implementation of the present application.
[0199] Those skilled in the art can understand that, unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art to which the present application belongs. It should also be understood that terms such as those defined in general dictionaries should be understood as having meanings consistent with those in the context of the prior art, and unless specifically defined, should not be interpreted in an idealized or overly formal sense.
[0200] For the convenience of description, the above device is described as various units and modules in function. Of course, the functions of the units and modules can be implemented in one or more software and / or hardware in the implementation of the present application.
[0201] From the above description of the embodiments, those skilled in the art can clearly understand that the present application can be implemented by means of software and the necessary general hardware platform. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which can be stored in a storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for causing a computer device (which can be a personal computer, a server or a network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments of the present application.
[0202] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit the present application; although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions recorded in the above embodiments can be modified, or some or all of the technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.
Claims
1. A vehicle body stiffness performance model optimization system, characterized in that, The vehicle body stiffness performance model optimization system includes: The module includes a data preprocessing module, a surrogate model training module, a multi-objective optimization module, and a parameter sensitivity analysis module. The data preprocessing module is used to process mixed variable types in vehicle body design through a hierarchical mapping strategy; The surrogate model training module is used to construct a multi-dimensional deep learning architecture that integrates four types of neural networks: multilayer perceptron, one-dimensional convolution, temporal memory, and self-attention mechanism. The multi-objective optimization module is used to construct a mixed variable processing framework based on the elite-preserving genetic algorithm; The sensitivity analysis module is used to quantify the impact of design using global sensitivity analysis.
2. The vehicle body stiffness performance model optimization system according to claim 1, characterized in that, The method of handling mixed variable types in vehicle body design through hierarchical mapping strategy includes: Based on the material and thickness mapping relationship library, discrete engineering parameters are automatically converted and mapped to thickness variables and material variables; The system intelligently separates stiffness and safety from the raw data and performs joint normalization on geometric variables and thickness / material variables, supporting custom normalization ranges. Among them, geometric variables are continuous variables, while thickness / material variables are discrete variables; Develop a data integration engine to achieve automatic alignment and joint normalization of torque stiffness and vehicle body mass data, providing standardized multidisciplinary input data for the proxy model and ensuring the collaborative processing capability of parameters with different dimensions.
3. The vehicle body stiffness performance model optimization system according to claim 1, characterized in that, The constructed multi-dimensional deep learning architecture integrates four types of neural networks: multilayer perceptron, one-dimensional convolution, temporal memory, and self-attention mechanism. End-to-end model optimization is achieved through an automated training framework; Among them, a dynamic learning rate scheduler combined with an early stopping mechanism is used to optimize the training process; Develop a dual-objective verification module to simultaneously monitor the prediction accuracy of torque stiffness and mass; Generate displacement distribution curves and a scatter matrix of predicted and actual values to visualize model performance; Among its features, it supports saving model parameters and backtracking the training process, ensuring the reproducibility of experiments and providing a high-precision prediction model for further optimization.
4. The vehicle body stiffness performance model optimization system according to claim 1, characterized in that, The framework for handling mixed variables based on the elite-preserving genetic algorithm includes: For geometric variables, real number encoding combined with specific evolutionary operators is used; For thickness / material variables, an index mapping mechanism is designed to achieve discrete value conversion through a predefined engineering combination library; in, In the optimization problem definition, automatic constraint verification is implemented to avoid invalid engineering combinations; in, The optimization process outputs the Pareto front solution set, automatically generating a structured report containing design variable values, torque stiffness values, and vehicle mass data. It also provides multi-objective decision support by visually and intuitively displaying the optimal trade-off between lightweighting and stiffness.
5. The vehicle body stiffness performance model optimization system according to claim 1, characterized in that, The use of global sensitivity analysis to quantify the design impact includes: Construct a multi-objective sensitivity assessment system; Constructing a multi-objective sensitivity assessment system includes identifying key parameters through core indicators and using conflict indices to locate points of conflict among multiple objectives. This also includes developing triple visualization tools; The development of triple visualization tools includes: scatter plots to display the impact of parameters on each target, horizontal bar charts to annotate trade-offs, and heat maps to link engineering variables with actual impacts; This also includes outputting structured reports of the analysis results; The analysis results output a structured report including the original sensitivity value, relative impact, and conflict markers, providing a quantitative basis for design iteration and clarifying the priority of parameter adjustments.
6. A method for optimizing a vehicle body stiffness performance model, characterized in that, The optimization method for the vehicle body stiffness performance model includes: The steps for data preparation, preprocessing and model training, multi-objective optimization, and parameter sensitivity analysis; The steps for preparing data include preparing the original design data and auxiliary configuration files; The preprocessing and model training steps include: intelligent data preprocessing followed by training by a proxy model, and outputting model results; The steps of multi-objective optimization include configuring mixed variables, performing optimization by NSGA-II, and outputting the optimization results; The steps of parameter sensitivity analysis include: performing sensitivity quantification analysis after global sensitivity sampling, outputting the analysis results, and performing triple visualization of the output results.
7. The method for optimizing the vehicle body stiffness performance model according to claim 6, characterized in that, The preparation of the original design data includes: Prepare a CSV file containing mixed design variables, including continuous variables of geometric parameters and thickness / material parameters; This also includes preparing a mapping file for thickness / material parameters; The thickness / material parameter mapping file is an Excel format file, used to define the engineering constraint relationship between material ID and actual thickness; This also includes preparing the variable definition file; The variable definition file is in Excel format and is used to specify the variables required for stiffness / safety analysis and their value ranges. This also includes preparing torque / mass label files; The torque / mass label file includes Excel data for the actual values of torque stiffness and vehicle mass.
8. An electronic device, characterized in that, include: The processor, communication interface, memory, and communication bus are connected, with the processor, communication interface, and memory communicating with each other via the communication bus. The memory stores a computer program that, when executed by a processor, causes the processor to perform the steps of the vehicle body stiffness performance model optimization method as described in any one of claims 6 or 7.
9. A computer-readable storage medium, characterized in that, The device stores a computer program executable by an electronic device, which, when run on the electronic device, causes the electronic device to perform the steps of the vehicle body stiffness performance model optimization method as described in any one of claims 6 or 7.
10. A design platform, characterized in that, include: An electronic device for implementing the steps of the vehicle body stiffness performance model optimization method as described in any one of claims 6 or 7; The processor runs a program that, when the program is running, executes the steps of the vehicle body stiffness performance model optimization method as described in any one of claims 6 or 7 from data output by the electronic device. A storage medium for storing a program that, when run, performs the steps of the vehicle body stiffness performance model optimization method as described in any one of claims 6 or 7 on data output from an electronic device.