Vehicle body structure performance prediction method and device

Abstract

The invention discloses a vehicle body structure performance prediction method and device, and relates to the technical field of vehicle structure design and CAE simulation. The invention aims to solve the problems of long time consumption and lack of quick verification tools in finite element simulation calculation in the existing vehicle body rigidity optimization process. The method comprises the following steps: acquiring an initial thickness and an initial performance index of a vehicle body structural member; setting a thickness change range and generating sample points by adopting a sampling strategy; performing finite element calculation in batches to obtain sample performance indexes; constructing an approximate mathematical model of thickness and performance indexes based on the sample data; and receiving a correction thickness input by a user, and directly outputting a prediction performance index by using the model. According to the method, key parts are identified through sensitivity sorting, and the rigidity / weight cost performance is calculated. According to the method, the rigidity prediction time can be shortened from the hour level to the real-time level, and the efficiency of vehicle body lightweight design is greatly improved.

Description

Technical Field

[0001] This invention relates to the fields of computer-aided engineering (CAE) and vehicle structure design technology, and in particular to a method and apparatus for predicting the performance of vehicle body structures based on an approximate model. Background Technology

[0002] In the optimization phase of automotive body design, body stiffness and modal characteristics are important indicators for measuring vehicle performance. Traditionally, to obtain these indicators, engineers need to build detailed finite element method (FEA) simulation models and perform complex solution calculations.

[0003] However, existing technical solutions have significant drawbacks. The closest existing technology typically employs a "trial and error" approach, where engineers manually adjust the material thickness parameters of components in a simulation model, repeatedly solving the problem to confirm its effectiveness in improving modal stiffness or in reducing weight while ensuring minimal modal changes. This method is time-consuming (usually taking more than half an hour per solution) and, due to its lack of directionality, often requires numerous blind attempts, resulting in extremely low efficiency.

[0004] While existing technology two incorporates contribution analysis to predict which parts are more sensitive and thus adjust the material thickness accordingly, its drawback is that even after identifying critical parts and adjusting the material thickness, there is still no tool for rapid verification. It is necessary to run the time-consuming finite element solution again to confirm the specific improved values.

[0005] In summary, current technologies lack a tool that can quickly, in real-time, and accurately predict the impact of material thickness variations on vehicle body stiffness without repeatedly running time-consuming simulation models. Communication barriers exist between designers and simulation engineers, and numerous sporadic calculations consume valuable R&D time. Summary of the Invention

[0006] To address the problems of long simulation calculation time, lack of rapid verification tools, and inability to efficiently guide lightweight design in the background technology, this invention provides a method, device, electronic device, and storage medium for predicting vehicle body stiffness. The aim is to establish a rapid mathematical mapping relationship between material thickness and stiffness / modality, achieve millisecond-level performance prediction, and significantly improve the efficiency of vehicle body structure optimization.

[0007] To address the aforementioned technical problems, in a first aspect, the present invention provides a method for predicting the performance of a vehicle body structure, comprising the following steps:

[0008] Obtain the initial thickness parameters and corresponding initial performance indicators of multiple structural components in the target vehicle body structure system;

[0009] Set a variation range based on the initial thickness parameter, and generate multiple sets of sample thickness data within the variation range using a preset sampling strategy;

[0010] Use a finite element analysis tool to calculate the multiple sets of sample thickness data to obtain corresponding sample performance indicators;

[0011] According to the sample thickness data and the sample performance indicators, construct an approximate mathematical model describing the mapping relationship between the thickness parameter and the performance indicator;

[0012] Receive the corrected thickness parameter input by the user for the target structural member, and directly output the corresponding predicted performance indicator using the approximate mathematical model.

[0013] Through the above solution, the present invention replaces the traditional finite element solution process with a high-precision linear approximate mathematical model. In the optimization design stage, designers only need to input the modified material thickness, and the system can instantly output the predicted values of stiffness and mode based on the pre-constructed mathematical relationship, without waiting for more than half an hour of simulation time, greatly improving the design iteration efficiency.

[0014] As a preferred solution of the present invention, the preset sampling strategy is Latin hypercube sampling;

[0015] Generating multiple sets of sample thickness data within the variation range specifically includes:

[0016] Define the thickness parameters of all structural members as variables, and set the variation range as the upper and lower limits of a preset percentage of the initial thickness parameter;

[0017] Select the number of sample points M, where n < M ≤ 3n, and n is the number of the structural members;

[0018] Take the initial thickness parameter as a reserved sample point, and combine it with the generated M - 1 sample points to form an overall sample set.

[0019] This technical solution realizes uniform sampling in the multi-dimensional variable space through Latin hypercube sampling (LHS), ensuring the representativeness of the samples; at the same time, controlling the number of samples between n + 1 and 3n, which not only ensures the information volume required for model construction but also avoids excessive simulation calculation costs.

[0020] As a preferred solution of the present invention, the approximate mathematical model is a first-order linear approximate model, and it adopts the following first-order linear regression formula:

[0021]

[0022] where y is the predicted performance indicator, t i is the thickness parameter of the i-th structural member, β0 is the intercept, βi Let be the main effect coefficient of the i-th structural component.

[0023] This technical solution is based on the physical fact that the thickness variation and stiffness of the body structural components are strongly linear within a small range (such as ±25%). It uses a first-order linear model for fitting, which simplifies the model complexity and ensures sufficient prediction accuracy.

[0024] As a preferred embodiment of the present invention, after constructing the approximate mathematical model, a model accuracy verification step is further included:

[0025] Calculate the fitting accuracy of the approximate mathematical model within the range of variation;

[0026] If the fitting accuracy is lower than the preset threshold, the number of sample points is increased, and the finite element calculation and model building steps are re-executed.

[0027] The increase in the number of sample points includes: increasing the number of sample points by a preset proportion based on the original number of sample points, or setting a larger upper limit for the number of sample points.

[0028] The preset threshold includes a determination coefficient or standard deviation controlled above 95%.

[0029] This scheme introduces an adaptive accuracy control mechanism to ensure that the generated prediction model has high reliability (>95%) and avoids prediction bias caused by nonlinear factors.

[0030] As a preferred embodiment of the present invention, the method further includes sensitivity analysis and ranking steps:

[0031] Obtain the stiffness matrix and static displacement of each structural component;

[0032] The sensitivity of each structural component's thickness to stiffness is calculated using the sensitivity calculation formula;

[0033] All structural components are sorted according to their sensitivity to identify those with high contribution and prioritize them for adjustment.

[0034] By ranking by sensitivity, the system can proactively recommend critical components that have a ripple effect on the entire system to designers, avoiding blind adjustments and improving the targeted nature of optimization.

[0035] As a preferred embodiment of the present invention, the method further includes a weight and cost-effectiveness analysis step:

[0036] Based on the corrected thickness parameters, calculate the total weight and weight change rate of the target vehicle body structure system; based on the predicted performance indicators, calculate the performance change rate; calculate the ratio of the performance change rate to the weight change rate as a cost-effectiveness indicator for evaluating stiffness performance optimization.

[0037] This solution not only focuses on stiffness itself, but also introduces a cost-effectiveness evaluation dimension of "stiffness / weight", providing a quantitative basis for decision-making in achieving the goal of lightweighting.

[0038] In a second aspect, the present invention provides a vehicle body structure performance prediction device, comprising:

[0039] The data acquisition module is used to acquire the initial thickness parameters and corresponding initial performance indicators of multiple structural components in the target vehicle body structure system.

[0040] The sample generation module is used to set a variation range based on the initial thickness parameter and generate multiple sets of sample thickness data within the variation range using a preset sampling strategy.

[0041] The simulation calculation module is used to call the finite element analysis tool to calculate the thickness data of the multiple sets of samples and obtain the corresponding sample performance indicators.

[0042] The model building module is used to construct an approximate mathematical model describing the mapping relationship between the thickness parameter and the performance index based on the sample thickness data and the sample performance index.

[0043] The prediction interaction module is used to receive the corrected thickness parameters input by the user for the target structural component, and directly output the corresponding prediction performance index using the approximate mathematical model.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] 1. Significantly improves efficiency: By establishing a mathematical substitution model, the simulation calculation that originally required more than 0.5 hours is compressed to a millisecond-level response, enabling real-time verification of the design scheme.

[0046] 2. Reduced communication costs: The packaged prediction equipment can be directly used by structural designers, reducing the need to repeatedly delegate calculations to simulation engineers and shortening the vehicle development cycle.

[0047] 3. Optimize decision support: By ranking sensitivity and analyzing cost-effectiveness (stiffness / weight ratio), key components can be quickly located, helping designers find the optimal balance between lightweighting and performance goals.

[0048] 4. High accuracy assurance: A sample generation strategy based on Latin hypercube sampling and a dynamic accuracy verification mechanism are introduced to ensure that the prediction accuracy reaches more than 95% within the range of material thickness variation (such as ±25%). Attached Figure Description

[0049] To more clearly illustrate the technical solutions of the embodiments disclosed in this invention, the accompanying drawings of the embodiments will be briefly described below. These drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention.

[0050] Figure 1 This is a flowchart illustrating the vehicle body stiffness prediction method provided in Embodiment 1 of the present invention.

[0051] Figure 2 This is a detailed logic flowchart of model construction and accuracy verification in Embodiment 1 of the present invention;

[0052] Figure 3 This is a structural block diagram of the vehicle body stiffness prediction device provided in Embodiment 2 of the present invention;

[0053] Figure 4 This is a schematic diagram of the prediction interaction interface provided in Embodiment 3 of the present invention. Detailed Implementation

[0054] The technical solutions (including preferred technical solutions) of the present invention will be further described in detail below with reference to the accompanying drawings and by way of listing some optional embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0055] Example 1

[0056] This embodiment provides a method for predicting vehicle body stiffness, aiming to solve the problem of time-consuming repeated simulation calculations in existing vehicle body designs. For example... Figure 1 and Figure 2 As shown, the method includes the following steps:

[0057] S1: Obtain initial data

[0058] The system first reads the finite element model information of the target body structure system (such as body-in-white). It identifies all structural components with adjustable material thickness (thickness), denoted as n parts. The initial thickness parameter T of these structural components is then obtained. initial= {t1, t2, ..., t n The initial performance indices (such as bending stiffness, torsional stiffness, first-order modal frequency, etc.) obtained through finite element analysis under this initial state are denoted as Y. initial .

[0059] It should be noted that although this embodiment uses a "first-order linear approximation model" as an example, in other embodiments of the present invention, the approximate mathematical model may also employ a response surface model, a Kriging model, or a neural network model, depending on the degree of nonlinearity between thickness variation and performance indicators. Furthermore, the performance indicators are not limited to stiffness (such as bending stiffness and torsional stiffness), but may also include the first-order modal frequency, second-order modal frequency, or other structural dynamic indicators of the vehicle body.

[0060] S2: Setting the Scope and Sampling

[0061] The thickness of all structural components is defined as a design variable. The range of thickness variation is set. In this embodiment, the range is set to ±25% of the initial thickness. Sample points are generated using a Latin Hypercube Sampling (LHS) strategy.

[0062] Specifically, the number of sample points M is selected. To ensure the solvability and accuracy of the linear regression equation, the number of samples M is at least n+1. In this preferred embodiment, M is set between n+1 and 3n. The initial value of the retained variable is placed as a sample point in the overall sample points, and the remaining M-1 sample points are uniformly generated in the n-dimensional design space using the LHS algorithm, ensuring that the sample points have good filling and representativeness in the design space.

[0063] S3: Batch Simulation Calculation

[0064] The generated M sets of sample points are batch-calculated using finite element analysis software. Each set of samples corresponds to a specific full-vehicle material thickness distribution, and the stiffness and modal results are extracted after calculation. This process can be automated by a script.

[0065] S4: Construct a linear approximation model

[0066] Based on the results of the above batch calculations, a mathematical relationship between material thickness and stiffness / modal characteristics is established. Since the stiffness of the vehicle body structure and the sheet metal thickness typically exhibit a strong linear relationship within a ±25% engineering adjustment range, this embodiment adopts a first-order linear approximation model:

[0067]

[0068] Where y is the response variable (e.g., torsional stiffness), t i β is the thickness of the i-th structural component, β0 is the intercept (the average response when all factors are at zero level or center point), β i It is the main effect coefficient (representing the linear effect of the thickness of structural component i on stiffness y, i.e., sensitivity).

[0069] S5: Model Accuracy Verification and Iteration

[0070] The fitting accuracy of the calculated mathematical model is usually expressed using the coefficient of determination R. 2 Or it can be measured by the standard deviation s:

[0071]

[0072] The model is required to maintain an accuracy of over 95% within a 25% variation range in material thickness.

[0073] If the fitting accuracy is found to be below 95% in step S4, it indicates that the current sample points are insufficient to describe the design space, or that nonlinear factors exist. In this case, the system automatically increases the number of sample points (e.g., if the current sample number is M, it increases to 1.5M or 2M, or directly to more than 3n), and returns to step S3 to recalculate the finite element method and update the mathematical model. If the accuracy still does not meet the requirements after increasing the sample number, the system may prompt the user to reduce the thickness variation range (e.g., to ±10%), or prompt the user to switch to a higher-order nonlinear approximation model.

[0074] S6: Prediction and Output

[0075] The verified mathematical relationships described above are encapsulated into a prediction tool. This tool receives the modified material thickness T from the user input. new Substitute this value into the above formula to directly calculate and output the predicted stiffness value. Simultaneously, the system has a boundary check function: if the user-input material thickness adjustment exceeds the preset ±25% range, the system will indicate that the basic model and alternative model are no longer applicable, and recommend that the user update the basic model again according to process S1-S5.

[0076] In addition to stiffness prediction, this model is also applicable to modal frequency prediction, based on the same principle: establishing a linear relationship of Modul=f(t1, t2, ...).

[0077] Example 2

[0078] This embodiment focuses on the sensitivity ranking function before prediction and the comprehensive evaluation function after prediction.

[0079] 1. Sensitivity Ranking

[0080] In the early stages of design, faced with hundreds or even thousands of body parts, designers often don't know which parts to adjust most effectively. In this embodiment, sensitivity calculation is performed after or simultaneously with step S1.

[0081] The formula for calculating sensitivity is:

[0082]

[0083] Where u is the static displacement vector, K is the overall stiffness matrix, and t is the thickness of the part. This formula reflects the displacement change caused by a unit thickness change (i.e., the rate of change of flexibility / stiffness).

[0084] The system calculates the sensitivity values ​​of all structural components and sorts them by absolute value. The top-ranked components are identified as "high-contribution components," serving as a key reference for subsequent design modifications. This addresses the deficiency in existing technologies that "cannot identify which components are important."

[0085] 2. Results Packaging and Evaluation

[0086] like Figure 4 As shown, the system interface is divided into an input area and an output area.

[0087] Input area: Lists all structural components (such as "Sheet Metal 1", "Sheet Metal 2", etc.) and provides a material thickness input box on the side.

[0088] Output area:

[0089] Modal / stiffness results: Displays base values ​​and predicted values.

[0090] Weight statistics: The system calculates the total weight W and the weight change rate ΔW in real time based on the input material thickness and part area / density.

[0091] Stiffness statistics: Calculate the rate of change of stiffness ΔK.

[0092] Cost-effectiveness index: Calculate the ratio of ΔK% / ΔW%. The larger this ratio, the greater the increase in stiffness per unit weight increase, or the greater the weight reduction per unit stiffness loss, and the higher the cost-effectiveness.

[0093] Example 3

[0094] This embodiment provides a vehicle body stiffness prediction device that can run on a high-performance computer or workstation. For example... Figure 3 As shown, the device includes:

[0095] Data acquisition module 100: used to read the finite element model of the vehicle body and its initial parameters.

[0096] Sample generation module 200: Built-in Latin hypercube sampling algorithm, used to generate n+1 to 3n sample design schemes.

[0097] Simulation Calculation Module 300: It is equipped with an interface with commercial finite element software (such as Nastran and Abaqus) to automatically submit calculation tasks and extract result files (such as .pch or .f06 files).

[0098] Model building module 400: Used to perform multiple linear regression analysis, calculate regression coefficient β, and perform accuracy verification.

[0099] Predictive Interaction Module 500: Provides, for example Figure 4 The human-computer interaction interface shown is included in this module.

[0100] Structural component thickness input port: Supports users to manually input or batch import new thickness values.

[0101] Performance results output port: Real-time refresh display of predicted modes and stiffness.

[0102] The statistical results display area shows the weight, rate of change, and cost-effectiveness ratio.

[0103] It will be readily understood by those skilled in the art that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, combinations, substitutions, improvements, etc., made under the spirit and principles of the present invention are included within the protection scope of the present invention.

Claims

1. A method for predicting the performance of a vehicle body structure, characterized in that, It includes the following steps: Obtain the initial thickness parameters of multiple structural components in the target vehicle body structure system and the corresponding initial performance indicators; Set a range of variation based on the initial thickness parameters, and use a preset sampling strategy to generate multiple sets of sample thickness data within the range of variation; Use a finite element analysis tool to calculate the multiple sets of sample thickness data to obtain the corresponding sample performance indicators; According to the sample thickness data and the sample performance indicators, construct an approximate mathematical model describing the mapping relationship between the thickness parameters and the performance indicators; Receive the corrected thickness parameters input by the user for the target structural component, and directly output the corresponding predicted performance indicators using the approximate mathematical model; 2. The method for predicting the performance of vehicle body structure according to claim 1, characterized in that, The preset sampling strategy is Latin hypercube sampling; The generating of multiple sets of sample thickness data within the range of variation specifically includes: Define the thickness parameters of all structural components as variables, and set the range of variation as the upper and lower limits of a preset percentage of the initial thickness parameters; Select the number of sample points M, where n < M ≤ 3n, and n is the number of the structural components; Use the initial thickness parameters as a reserved sample point, and combine with the generated M - 1 sample points to form an overall sample set; 3. The method for predicting the performance of vehicle body structure according to claim 1, characterized in that, The approximate mathematical model is a first-order linear approximation model, and it adopts the following first-order linear regression formula: Where y is the prediction performance metric, t i Let β be the thickness parameter of the i-th structural component, β0 be the intercept, and β... i Let be the main effect coefficient of the i-th structural component.

4. The method for predicting the performance of vehicle body structure according to claim 3, characterized in that, After constructing the approximate mathematical model, it further includes a model accuracy verification step: Calculate the fitting accuracy of the approximate mathematical model within the range of variation; If the fitting accuracy is lower than a preset threshold, increase the number of sample points, and re-execute the finite element calculation and model construction steps; The increasing of the number of sample points includes: increasing a preset proportion of sample points based on the original number of sample points, or resetting a larger upper limit of the number of sample points; The preset threshold includes that the coefficient of determination or the standard variance is controlled above 95%; 5. The method for predicting the performance of vehicle body structure according to claim 1, characterized in that, The method further includes a sensitivity analysis and ranking step: Obtain the stiffness matrix and static displacement of each structural component; Use a sensitivity calculation formula to calculate the sensitivity of the thickness of each structural component to the stiffness; Rank all structural components according to the magnitude of the sensitivity to identify high-contribution structural components as the priority adjustment objects; The sensitivity calculation formula is based on the following relationship: where t is the thickness of the structural component, u is the static displacement, and K is the stiffness matrix; 6. The method for predicting the performance of vehicle body structure according to claim 1, characterized in that, The method further includes a weight and cost performance analysis step: According to the corrected thickness parameters, calculate the total weight and weight change rate of the target vehicle body structure system; According to the predicted performance indicators, calculate the performance change rate; Calculate the ratio of the performance change rate to the weight change rate as a cost performance index for evaluating the stiffness performance optimization; 7. The method for predicting the performance of vehicle body structure according to claim 1, characterized in that, The method further includes a boundary monitoring step: Judge whether the corrected thickness parameters input by the user exceed the range of variation; If it exceeds the range of variation, issue a warning or automatically trigger the re-execution of steps: set a new range of variation, generate samples, perform finite element calculation and model construction; 8. A vehicle body structure performance prediction device, characterized in that, It includes: A data acquisition module for obtaining the initial thickness parameters of multiple structural components in the target vehicle body structure system and the corresponding initial performance indicators; The sample generation module is used to set a variation range based on the initial thickness parameter and generate multiple sets of sample thickness data within the variation range using a preset sampling strategy. The simulation calculation module is used to call the finite element analysis tool to calculate the thickness data of the multiple sets of samples and obtain the corresponding sample performance indicators. The model building module is used to construct an approximate mathematical model describing the mapping relationship between the thickness parameter and the performance index based on the sample thickness data and the sample performance index. The prediction interaction module is used to receive the corrected thickness parameters input by the user for the target structural component, and directly output the corresponding prediction performance index using the approximate mathematical model.

9. The vehicle body structure performance prediction device according to claim 8, characterized in that, The device further includes: The sensitivity analysis module is used to calculate the sensitivity of each structural component to the performance index and rank their contributions, and then highlight or recommend high-sensitivity structural components in the prediction interaction module. The comprehensive evaluation module is used to simultaneously calculate and output the total weight, weight change rate, stiffness change rate, and stiffness-to-weight ratio of the vehicle body structure system based on the input corrected thickness parameters.

10. The vehicle body structure performance prediction device according to claim 8, characterized in that, The model building module is specifically used for: Fitting the first-order linear equation using the least squares method To verify the model's accuracy, if the accuracy is below 95%, the sample generation module is triggered to increase the number of samples, and the simulation calculation module is triggered to recalculate.

11. The vehicle body structure performance prediction device according to claim 8, characterized in that, The prediction interaction module includes an encapsulated user interface, which is configured with: The structural component thickness input port allows users to edit the thickness of each structural component; The performance results output port is used to display the predicted modal and stiffness values; The statistical results display area shows the total weight, percentage change in weight, percentage change in stiffness, and cost-effectiveness ratio.

Images