Research On The Optimization Design Of Vehicle Bottom Lightning Protection Components Considering Mixed Variable

Mixed variables refer to that in the structural optimization design,the design variables include both continuous variables and discrete variables.The selection of various structural materials in the optimization design of bottom protective components belongs to typical discrete variables.Based on this,this paper first studies various improvements of Gaussian process model in dealing with discrete variables,three improved methods based on Gaussian process model are compared and analyzed: multiplication covariance(MC＿GP),addition covariance(ADD＿GP)and latent variables Gaussian process model(LVGP).It is found that LVGP method has more advantages in dealing with discrete variables.The test and finite element simulation of the response of small equivalent explosive target are carried out.MC＿GP、ADD＿GP and LVGP are used to establish the surrogate model of small equivalent explosion target deformation,which verifies the advantages of LVGP method and the feasibility of LVGP method in the process of explosion impact.Secondly,the design scheme analysis,experimental analysis and finite element simulation model verification of the vehicle bottom protective components are carried out,and the improvement scheme of the bottom protection assembly is proposed.Aiming at the problem that the optimization of the bottom protection components involves multiple response values,a kernel function selection multi-response value latent variables Gaussian process(KS-MR LVGP)is improved on the basis of LVGP method,KS-MR LVGP method is used to establish the surrogate model of bottom protection components,and its effectiveness is verified.Finally,based on the surrogate model established by KS-MR LVGP method,the particle swarm optimization algorithm is used to solve the optimization of the bottom protection components.At the same time,in order to compare the influence of the accuracy of the surrogate model on the optimization results,the neural network method is used to create the surrogate model of the bottom protection components,and the particle swarm optimization algorithm is used to optimize the solution,the optimization results obtained by the two methods are verified by finite element simulation.It is found that the optimization results based on KS-MR LVGP method are better and the accuracy of the optimization results is higher.