基于MLSM-GRSM模型的客车侧碰安全与轻量化优化OA

To address the problems of large computational cost and inefficiency in achieving multi-objective optimization during iterative finite element method(FEM)optimization,the static,modal,and side-impact finite element(FE)models of an electric bus body,together with 55 sets of Latin hypercube sampling(LHS)sample data,were established.Taking the vehicle mass and side-impact intrusion as optimization objectives,and constraining the static strength,stiffness,and modal frequency responses,surrogate models were constructed using the least squares regression(LSR)and moving least squares regression(MLSR)methods.The frame was then optimized and compared using the multi-objective genetic algorithm(MOGA)and the global response search method(GRSM).The results showed that,compared with the traditional LSR method,the MLSR surrogate model achieved higher fitting accuracy for vehicle mass,stress,and modal frequencies.The convergence efficiency of the GRSM algorithm was improved by 69.6%compared with MOGA.Consequently,the vehicle body weight was reduced by 11.11%,the peak intrusion values of the upper and lower side-impact battery protection beams on the frame were reduced by 15.11%and 18.57%,respectively,and the peak static stress was 224.3 MPa(safety factor of 1.53).The accuracy and efficiency of surrogate models in complex engineering optimization were thereby validated,providing a new approach for the lightweight design of electric buses.