CAE optimization method based on vehicle multi-working-condition collision analysis

Abstract

The invention discloses a CAE (Computer Aided Engineering) optimization method based on vehicle multi-working-condition collision analysis, which comprises the following steps of: S1, establishing a vehicle finite element model for optimization analysis based on a CAE technology and grid division software Ansa, setting the unit size to be 10mm, setting 2002989 units and 1806803 nodes for a whole vehicle, and using 48 materials including glass, structural steel, aluminum alloy, rubber and damping rubber for the whole vehicle, the finite element model is composed of a body-in-white, a power assembly, a steering system, front and rear suspension systems, front and rear bumpers and a battery pack. According to the optimization method, a comprehensive and scientific analysis means is provided for automobile comprehensive performance optimization, and the rationality and safety of the automobile structure can be comprehensively evaluated in the design stage by accurately simulating the performance of the automobile under multi-working-condition collision, so that design improvement is effectively guided, the design period is remarkably shortened, the design cost is reduced, and the design efficiency is improved. And the overall competitiveness of automobile products is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile safety performance optimization, and in particular to a CAE optimization method based on vehicle multi-operating condition collision analysis. Background Art

[0002] In a car traffic accident, collisions between cars and road facilities can cause injuries to people. Countries around the world have formulated car collision regulations, such as my country's GB11551-2003 "Occupant Protection in Frontal Collisions of Passenger Cars", GB20071-2006 "Occupant Protection in Side Collisions of Passenger Cars", and GB / T20913-2007 "Occupant Protection in Frontal Offset Collisions of Passenger Cars", which put forward clear numerical requirements for car safety performance and dummy injury values.

[0003] Computer-aided engineering (CAE) and optimization design methods currently play an important role in the research and development and manufacturing of automobiles and other fields, but the commonly used automotive CAE optimization methods have limitations. Most of the existing methods are single linear condition shape optimization or automotive component sensitivity analysis and material thickness optimization. There are few optimization analysis methods for nonlinear conditions such as collisions, and there are almost no optimization analysis rules that comprehensively consider side collisions, side column collisions, and roof compression. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a CAE optimization method based on vehicle multi-condition collision analysis.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] The present invention provides a CAE optimization method based on vehicle multi-operating condition collision analysis, and the specific steps are as follows:

[0007] Vehicle finite element model establishment

[0008] CAE technology and meshing software Ansa are used to build a vehicle finite element model for optimization analysis. The model unit size is set to 10mm, and the whole vehicle contains 2,002,989 units and 1,806,803 nodes. In terms of material usage, the whole vehicle uses 48 different materials, including glass, structural steel, aluminum alloy, rubber, shock-absorbing rubber, etc., to accurately simulate the real structure and physical properties of the vehicle. The finite element model fully includes the key components of the vehicle, such as the body in white, powertrain, steering system, front and rear suspension systems, front and rear bumpers, battery packs, etc., to ensure the integrity and accuracy of the model.

[0009] Basic performance analysis

[0010] The basic performance analysis of the vehicle finite element model is carried out with the help of LS-dyna software, mainly including side collision, side column collision and roof compression simulation analysis. In the side collision simulation, the speed is set to 60km / h; in the side column collision simulation analysis, the speed is set to 32km / h to simulate the response of the vehicle under different collision scenarios. During the simulation process, a detailed collision analysis calculation file is output, which contains important physical quantity data, such as the acceleration at the bottom of the B-pillar, the Y-direction intrusion of the upper, middle and lower parts of the B-pillar, and the energy change curve during the calculation process. To ensure the reliability of the simulation results, the energy change curve must meet the requirements of natural changes and smooth transitions, and the hourglass energy must be less than 5% of the total energy. Afterwards, the calculation file is set as the corresponding analysis file (Tpl file) in Hyperstudy and imported into Hyperstudy to prepare for subsequent calculations and analysis.

[0011] Grouping Design Variables

[0012] Based on function, thickness and shape, the longitudinal beam frame, passenger compartment frame and floor frame of the vehicle are reasonably divided into 18 groups, and the cross-sectional variation size of the B-pillar is carefully divided into 5 groups. To ensure the accuracy and consistency of the subsequent joint optimization, the grouping of the side collision, side column collision and roof compression simulation models remains completely consistent, thus determining a total of 23 optimization variables, which will serve as key adjustment factors in the optimization process.

[0013] Defining the optimization problem

[0014] The optimization problem is clarified, and the acceleration of the B-pillar, the intrusion of the upper, middle and lower B-pillar, the deformation of the frame and the vehicle mass obtained in the previous simulation process are set as the optimization response function. The Hammersley method is used to perform DOE (experimental design) analysis on the plate material thickness of the longitudinal beam frame, the passenger compartment frame, the floor frame and the cross-sectional variation size of the B-pillar, providing data support for subsequent optimization decisions.

[0015] Screening Design Variables and Experimental Design

[0016] According to the analysis results of DOE, the sensitivity of each design variable to the response function is deeply examined. By eliminating the design variables with low sensitivity, the optimization model is effectively streamlined, thereby improving the optimization efficiency. Subsequently, the Latin hypercube method is used to carefully design the experiments for each retained design variable, and finally the Latin hypercube experimental design sample distribution is obtained, laying the foundation for building an accurate approximate model.

[0017] Approximate model building

[0018] A radial basis function neural network (RBF) model is constructed based on the experimental design sample distribution data. After the construction is completed, the accuracy of the model is comprehensively evaluated to ensure that it meets the accuracy requirements. Only an approximate model with accurate accuracy can accurately reflect the complex relationship between the design variables and the response function, providing a reliable basis for subsequent optimization.

[0019] Optimization solution

[0020] Based on the established approximate model, the sequential quadratic programming algorithm (SQP) and the multi-objective genetic algorithm in the multidisciplinary optimization algorithm are used to optimize the main cross-sectional dimensions of the B-pillar and the thicknesses of the longitudinal beam frame, the passenger compartment frame, and the floor frame to obtain the optimal design variables and achieve optimized improvement of vehicle performance.

[0021] Verification

[0022] Substitute the optimized solution obtained through optimization calculation into the three basic collision simulation models and recalculate. By comparing the calculation results with the target value requirements, strictly verify the effectiveness of the optimization plan to ensure that the performance of the optimized vehicle under multiple working conditions collision is effectively improved.

[0023] The present invention has the following beneficial effects:

[0024] Improve overall vehicle performance and R&D efficiency

[0025] The optimization method of the present invention provides a comprehensive and scientific analysis method for optimizing the comprehensive performance of automobiles. By accurately simulating the performance of vehicles under multiple collision conditions, the rationality and safety of vehicle structures can be comprehensively evaluated during the design phase, thereby effectively guiding design improvements, significantly shortening the design cycle, reducing design costs, and improving the overall competitiveness of automobile products.

[0026] Enhance vehicle safety and economy

[0027] In terms of automobile safety performance, this method comprehensively considers multiple key collision conditions such as side collision, side column collision, and roof compression, and can optimize the vehicle structure in a targeted manner, enhance the impact resistance of key parts (such as B-pillars, longitudinal beam frames, passenger compartment frames, floor frames, etc.), effectively reduce the risk of injury to occupants in the vehicle due to collisions, and improve the passive safety performance of the vehicle. At the same time, by optimizing design variables (such as plate material thickness, B-pillar cross-sectional dimensions, etc.), while ensuring safety performance, it helps to achieve lightweight design of the vehicle, reduce the weight of the vehicle, and thus improve the fuel economy of the vehicle, reduce energy consumption and exhaust emissions, which has important economic and environmental significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is the multi-operating condition collision optimization solution of the present invention;

[0029] Figure 2 The automobile multi-operating condition collision optimization flow chart of the present invention;

[0030] Figure 3 This is the automobile simulation grid computing model of the present invention. DETAILED DESCRIPTION

[0031] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0032] A CAE optimization method based on vehicle multi-operating condition collision analysis

[0033] Vehicle finite element model establishment

[0034] The meshing software Ansa was started to mesh the vehicle according to the actual structure and geometry of the vehicle with a unit size of 10 mm. During the meshing process, the body-in-white, powertrain, steering system, front and rear suspension systems, front and rear bumpers, battery pack and other components were accurately distinguished to ensure that the mesh quality and connection relationship of each component were accurate.

[0035] Based on the vehicle's material list, each unit in the model is assigned corresponding material properties, including 48 materials such as glass, structural steel, aluminum alloy, rubber, and shock-absorbing rubber, so that the model can truly reflect the physical characteristics of the vehicle.

[0036] After completing the meshing and material assignment, the model was checked for integrity to ensure that the vehicle had a total of 2,002,989 units and 1,806,803 nodes, and that all parts of the model were tightly connected without loopholes, providing a reliable basic model for subsequent analysis.

[0037] Basic performance analysis

[0038] Open the LS-dyna software and import the established vehicle finite element model.

[0039] For the side collision simulation analysis, the collision speed is set to 60km / h in the software. According to the regulatory requirements and the actual collision scenario, the parameters of the mobile deformable barrier are accurately set, that is, the longitudinal median vertical plane trajectory of the mobile deformable barrier is perpendicular to the longitudinal median vertical plane of the collided vehicle, and the distance between the longitudinal median vertical plane of the mobile deformable barrier and the transverse vertical plane passing through the "R" point of the front seat on the collision side of the test vehicle is within ±25mm. Start the simulation calculation to obtain the acceleration of the bottom of the B-pillar, the Y-direction intrusion of the upper, middle and lower parts of the B-pillar, and the energy change curve of the calculation process to ensure that the energy change curve meets the requirements of natural change, smooth transition and hourglass energy less than 5% of the total energy.

[0040] For the simulation analysis of the side pole collision, the collision speed is set to 32km / h. When the vehicle is driven laterally to the rigid pole, it is ensured that when contact occurs, a collision angle of 75° is formed between the vertical plane parallel to the vehicle collision speed vector and the longitudinal centerline of the vehicle. Simulation calculations are also performed to obtain relevant physical quantity data and energy change curves.

[0041] In the compression simulation analysis of the vehicle roof, the corresponding loading conditions are set according to the actual test standards and requirements, and the relevant performance data are calculated and obtained.

[0042] Output the calculation files obtained from each collision analysis, set them correctly in Hyperstudy, define them as corresponding analysis files (tpl files), and prepare for subsequent joint analysis.

[0043] Grouping Design Variables

[0044] Based on the functional characteristics of the vehicle structure, the thickness differences and shape characteristics of the components, the longitudinal beam frame, the passenger compartment frame and the floor frame are systematically grouped. For example, parts with similar load-bearing functions, similar thickness ranges or similar shapes are grouped together, and finally 18 groups are determined.

[0045] The cross-sectional dimensions of the B-pillar are divided into five groups according to their changes in different positions and functional requirements. During the grouping process, the grouping of the side collision, side column collision, and roof compression simulation models is ensured to be completely consistent, so that the variable relationship under each working condition can be accurately reflected in the subsequent joint optimization.

[0046] Defining the optimization problem

[0047] In the Hyperstudy software, the acceleration of the B-pillar in the previous simulation of side collision, side pillar collision, roof compression simulation, the intrusion amount of the upper, middle and lower B-pillar, the deformation of the frame and the mass of the whole vehicle are clearly set as optimization response functions. These functions will serve as key indicators to measure the optimization effect.

[0048] The Hammersley method was selected to perform DOE analysis on the plate material thickness of the longitudinal beam frame, passenger compartment frame, floor frame, and the cross-sectional variation size of the B-pillar in Hyperstudy. During the analysis process, the DOE parameters were reasonably set to ensure that the analysis results can accurately reflect the influence of each design variable on the response function.

[0049] Screening Design Variables and Experimental Design

[0050] View the DOE analysis results in Hyperstudy, and determine the sensitivity of each design variable to the response function through data comparison and analysis. According to the sensitivity, filter out the design variables with low sensitivity and eliminate them to simplify the optimization model and improve the optimization efficiency.

[0051] The Latin hypercube method is used to conduct experimental design for the retained design variables in Hyperstudy. According to the number and range of variables, the parameters of the Latin hypercube experiment are reasonably set to generate the sample distribution of the Latin hypercube experiment design, providing sufficient and reasonable data samples for the establishment of subsequent approximate models.

[0052] Approximate model building

[0053] Import the sample distribution data of Latin hypercube experimental design into appropriate modeling software (such as Matlab, etc.), and use the radial basis function (RBF) algorithm to build an approximate model. In the modeling process, according to the data characteristics and optimization requirements, the parameters of the RBF function, such as the center and width of the basis function, are reasonably selected to ensure that the model can accurately fit the data.

[0054] The constructed RBF approximate model is evaluated for accuracy using methods such as cross-validation. By comparing the difference between the model prediction value and the actual simulation value, it is determined whether the accuracy of the model meets the requirements. If the accuracy is not satisfied, the model parameters are adjusted or sample data is added, and the model is rebuilt until an approximate model that meets the accuracy requirements is obtained.

[0055] Optimization solution

[0056] In Hyperstudy, based on the established approximate model, the sequential quadratic programming algorithm (SQP) and the multi-objective genetic algorithm in the multidisciplinary optimization algorithm are called respectively.

[0057] For the main cross-sectional dimensions of the B-pillar and the design variables such as the longitudinal beam frame, the passenger compartment frame, and the floor frame thickness, the SQP and multi-objective genetic algorithms are used for optimization calculations based on the optimization objectives (such as minimizing the intrusion of the B-pillar, the deformation of the frame, and controlling the mass of the vehicle) and constraints (such as material strength restrictions and manufacturing process requirements). During the calculation process, the algorithm parameters such as the number of iterations and convergence accuracy are reasonably set to obtain the optimal combination of design variables.

[0058] Verification

[0059] The optimal design variables obtained by optimization are substituted into the original three basic collision simulation models (side collision, side pillar collision, and roof compression), and the simulation calculation is re-performed in the LS-dyna software.

[0060] Compare the recalculated results (such as B-pillar acceleration, intrusion, frame deformation, and vehicle mass) with the pre-set target values. If all indicators meet the target value requirements, it means that the optimization scheme is effective; otherwise, it is necessary to analyze the reasons, adjust the optimization method or parameters, and re-optimize the process until a satisfactory optimization effect is achieved.

[0061] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A CAE optimization method based on vehicle multi-operating condition collision analysis, characterized in that: The following steps are involved: S1. A vehicle finite element model for optimization analysis is established based on CAE technology and meshing software Ansa. The unit size is set to 10 mm. A total of 2,002,989 units and 1,806,803 nodes are set for the whole vehicle. The whole vehicle uses 48 materials including glass, structural steel, aluminum alloy, rubber, and shock-absorbing rubber. The finite element model consists of a body-in-white, a powertrain, a steering system, front and rear suspension systems, front and rear bumpers, and a battery pack. S2. Use LS-dyna software to perform basic performance analysis on the vehicle finite element model, including side collision, side column collision, and vehicle roof compression simulation analysis. The side collision simulation speed is 60 km / h, and the side column collision simulation analysis speed is 32 km / h. Output the calculation file of the collision analysis, which contains the acceleration at the bottom of the B-pillar, the Y-direction intrusion of the upper, middle and lower parts of the B-pillar, and the energy change curve of the calculation process. The energy change curve must meet the requirements of natural change, smooth transition, and hourglass energy less than 5% of the total energy. Set the calculation file in Hyperstudy as the corresponding analysis file, i.e., Tpl file, and import it into Hyperstudy to calculate each analysis condition defined before. S3. Based on the grouping principle of function, thickness and shape characteristics, the longitudinal beam frame, passenger compartment frame and floor frame of the vehicle are divided into 18 groups, the cross-sectional variation size of the B-pillar is divided into 5 groups, and the grouping of the side collision, side column collision and vehicle top compression simulation models is kept consistent as the variables for joint optimization, with a total of 23 optimization variables; S4. Define the optimization problem, take the acceleration of the B-pillar, the intrusion of the upper, middle and lower B-pillar, the deformation of the frame and the vehicle mass obtained in the previous simulation of side collision, side pillar collision and vehicle top compression as the optimization response function, and use the Hammersley method to perform DOE analysis on the plate material thickness of the longitudinal beam frame, passenger compartment frame, floor frame and the cross-sectional variation size of the B-pillar; S5. Based on the analysis results of DOE, check the sensitivity of each design variable to the response function, eliminate the design variables with low sensitivity, use the Latin hypercube method to conduct experimental design for each design variable, and obtain the Latin hypercube experimental design sample distribution; S6. Establish a radial basis neural network model based on the experimental design sample distribution data, evaluate its accuracy, and construct an approximate model that meets the accuracy requirements; S7. Based on the established approximate model, the sequential quadratic programming algorithm in the multidisciplinary optimization algorithm and the multi-objective genetic algorithm are used to optimize the main cross-sectional dimensions of the B-pillar and the thickness of the longitudinal beam frame, the passenger compartment frame, and the floor frame to obtain the optimal design variables; S8. Substitute the obtained optimized solution into the three basic collision simulation models, recalculate, and verify whether the target value requirements are met.

2. The CAE optimization method based on vehicle multi-operating condition collision analysis according to claim 1, characterized in that: During the side impact analysis, the longitudinal median vertical plane trajectory of the mobile deformable barrier is perpendicular to the longitudinal median vertical plane of the impacted vehicle, and the distance between the longitudinal median vertical plane of the mobile deformable barrier and the transverse vertical plane passing through the "R" point of the front seat on the impact side of the test vehicle is within ±25mm.

3. The CAE optimization method based on vehicle multi-operating condition collision analysis according to claim 1 is characterized in that: The side pole collision analysis is performed by driving the vehicle laterally to the rigid pole. When contact occurs, a collision angle of 75° is formed between a vertical plane parallel to the vehicle collision velocity vector and the longitudinal centerline of the vehicle.

Images