Reverse Optimization Of Material Parameters In Collision Simulation

With the development of computer technology and the continuous improvement of finite element simulation technology,the finite element collision simulation analysis technology provides an important help for the research of rail vehicle collision energy absorption.The energy-absorbing box is a typical energy absorption device at the front of rail vehicles and its crushing impact energy absorbing characteristics directly influence the passive safety performance of vehicles.In the design of energy absorption box,the accuracy of geometric parameters,boundary conditions and material parameters directly influence the simulation accuracy of energy absorption box crushing collision.In the crushing simulation modelling of energy absorbing box,the accuracy of geometric parameters and boundary conditions is relatively easy to be guaranteed,while the data curve used to describe the material stress-strain characteristics of energy absorbing box needs to be obtained by means of tensile test bar,calibration of typical parts,etc.When there is no condition to conduct material tensile test to obtain material stress-strain parameters aiming at the complex nonlinear problem between material parameter variables and simulation results in energy-absorbing box collision simulation in order to ensure the accuracy of the obtained base metal stress-strain parameters of energy absorbing box,taking typical energy absorbing box test elements as the research object,based on finite element collision simulation.A reverse optimization method based on Latin hypercube design,RBF neural network approximation model,and NSGA-II algorithm is proposed to optimize the stress-strain parameters of base metal of energy-absorbing box.The main content is divided into four aspects.(1)This paper introduced the basic theory of reverse optimization of material stressstrain parameters and gives the selection criteria of three elements of optimization design in reverse optimization of material parameters and illustrates the relevant characteristics of Latin hypercube design,RBF neural network model and NSGA-II algorithm to support optimization solution.(2)The evaluation index of energy absorption performance of energy-absorbing box was given.According to the physical dimensions and actual boundary conditions of the energy absorbing box,the initial collision simulation model of the energy absorbing box is established through the coordinated application of multiple software.Based on the compression test data of typical energy absorbing box,the data between the initial energy absorbing box collision simulation and physical test are compared.(3)Based on the compression reaction curve of typical energy absorbing box,the first method is directly based on NSGA-II algorithm to optimize the stress-strain parameters of base metal of energy absorbing box in collision simulation.Secondly,the Latin hypercube test design,RBF neural network approximation model and NSGA-II algorithm are used to optimize the stress-strain parameters of the base metal of the energy absorbing box.(4)The accuracy of the stress-strain parameters of the base metal of the energy absorbing box obtained by the two reverse optimization methods was verified by taking the energy absorbing performance evaluation index of the energy absorbing box as the evaluation standard.The results show that,compared with the initial crash simulation of the energy absorbing box,the material parameters obtained by the two reverse optimization methods are in good agreement with the actual compression test.However,compared with the direct NSGA-II algorithm based reverse optimization,the proposed algorithm is more efficient,the hybrid algorithm is more efficient than the NSGA-II algorithm.The results show that the hybrid method of Latin hypercube design,RBF neural network approximation model and NSGA-II algorithm is effective.It provides a reference method for obtaining material characteristic parameters with higher accuracy.