基于SHCA-T算法的车身骨架多工况耐撞性优化设计

Multi Workig Condition Crashworthiness Optimization Design of Body Frame Based on SHCA-T Algorithm

为解决多变量非线性动态结构优化效率低、难以收敛等问题,提出求解车身骨架厚度优化的子区域混合元胞自动机(SHCA-T)算法以及多工况SHCA-T算法,实现车身骨架多工况耐撞性高效优化设计。该方法包括内外两层循环:外层循环主要开展碰撞仿真分析、计算输出响应,更新目标质量,实现结构质量的最小化;内层循环主要根据当前元胞及其邻胞的内能密度,按照PID控制策略调整元胞厚度,使内层循环的当前质量收敛于目标质量;最终使元胞内能密度分布尽可能逼近阶跃式目标内能密度函数。为了验证SHCA-T和多工况SHCA-T算法的精度和效率,将其用于求解侧面碰撞和侧面柱碰工况下车身骨架的厚度优化问题,并与基于伪CEI准则的并行约束EGO(EGO-PCEI)算法的优化结果进行对比。结果表明:在收敛精度相当的条件下,SHCA-T和多工况SHCA-T算法具有更高的全局搜索效率。

To solve the problems of low efficiency and difficult convergence of multi-variable nonlinear dynamic structure optimization,this paper proposes a Subdomain Hybrid Cellular Automata for Thickness optimization algorithm(SHCA-T)and multi working condition SHCA-T algorithm to solve the thickness optimization of the body skeleton,so as to realize efficient optimization design of the crashworthiness of the body skeleton under multiple working conditions.The algorithm includes an outer loop and an inner loop:the outer loop is to realize minimization of target mass by conducting crash finite element analysis(FEA),calculating output responses and updating target mass;the inner loop is to make the current mass of the inner loop to converge to the target mass by adjusting cell thicknesses according to internal energy densities of current cell and neighboring cells and PID control strategy,and finally make the cell internal energy density distribution as close as possible to the step target internal energy density function.In order to validate the accuracy and efficiency,the SHCA-T and multi working condition SHCA-T algorithms are used to solve the thickness optimization problem of body frame under side collision and side column collision.The optimized results obtained from the SHCA-T and multi working condition SHCA-T algorithms are compared with the results from parallel Pseudo expected improvement criterion for parallel EGO algorithm.The results show that the SHCA-T and multi working condition SHCA-T algorithms have higher global search efficiency under the condition of equal convergence accuracy.