Research Of Surrogate Models Technology And Its Application In Vehicle Crashworthiness Design

Passive safety of automotive has been one of the research hotspots incontemporary automotive industry, and Structural crashworthiness is a very importantresearch point in the automotive passive safety. The collision simulation technologyhas been widely adopted in the research of automobile crashworthiness. Thesimulation techniques can be used to analyze and optimize the crash performance ofautomobiles. With the development of the collision simulation technology, seekingmethods which are more suitable for stimulating optimization design of vehicle’scrash safety, has a very important engineering significance and academic value.In the function fitting process based on response surface method, radial basisfunction is usually used to replace the pure polynomial primary functions to constructshape functions due to the non-singularity of the moment matrix. The shape parameterof the radial basis function will significantly influence the shape of the shape function,the condition number of moment matrix and the accuracy of interpolation result. So,how to select the shape parameter of radial basis function is a key and difficultproblem in the study of the surrogate model. In this paper, we will study how to selectthe shape parameters of the common radial basis function through the comparativeanalysis of relative error in contrast design point. The radial basis functions we havestudied include: MQ and EXP radial basis function and MQ, EXP coupling linearpolynomial radial basis function.In this paper, the fitting accuracy of radial basis function model and polynomialbasis functions model have been compared and we obtain the optimal surrogate model,and the surrogate model has been used in function fitting for the100%frontal impactand moving barrier side impact, and we can get the crashworthiness optimizationproblem objectives of the vehicle body and the regression expression of constraintfunctions, and establish the mathematical model of optimization problems, and makeour multi-objective crashworthiness optimization design combined with themulti-island genetic algorithm and NSGA-П multi-objective genetic algorithm.The results of research show that we can get the optimal value range of the shapeparameters of radial basis functions and the optimal radial basis function by selectingdifferent shape parameters of the radial basis function and making comparativeanalysis of the relative error in additional design points, which provides some suggestions for the application of surrogate model in engineering. In this paper, theoptimization method based on the surrogate model of the MQ radial basis function isvery effective in the vehicle multivariate and multi-objective crashworthinessoptimization design, and the optimization method of surrogate model with appropriateshape parameter will help to solve the problems more accurately in crashworthinessoptimization design. The method and the results of this study provide a significantreference in crashworthiness optimization problem of the vehicle body in the future．...