基于组合代理模型的汽车碰撞安全性多目标优化研究

发布时间：2018-03-28 00:15

本文选题：汽车碰撞　切入点：代理模型　出处：《重庆大学》2016年硕士论文

【摘要】：随着汽车保有量的越来越大,对汽车安全的要求也越来越高。汽车安全主要分为主动安全及被动安全。主动安全可以在事故发生前主动规避,而一旦事故发生,则需要较好的整车被动安全性能,才能保证乘员的安全性。因此作为汽车安全的最后一道防线,汽车碰撞安全性能的优化至关重要。为了减少成本,对于汽车碰撞过程通常采用有限元仿真的方法。但仿真过程是非线性的动力学过程,耗时较长,无法满足迭代优化计算要求。因此需利用代理模型代替有限元模型,用以减少仿真时长,提高计算效率。本文主要的研究内容就是利用代理模型进行汽车碰撞安全性的优化设计。本文以汽车前部主要吸能板件的厚度作为设计变量,以B柱峰值加速度最小、脚踏板侵入量最小及整车质量最小作为优化目标,对汽车碰撞安全性进行了多目标优化计算。首先利用拉丁超立方试验设计方法选取样本点。其次建立高斯径向基模型、Multiquadric径向基模型及Kriging模型,并对三种模型的误差进行了分析,将精度较高的近似模型作为代理模型。其中B柱峰值加速度及脚踏板侵入量采用Kriging模型,整车质量采用c=4.5的Multiquadric径向基模型,用以降低计算代价,提高计算效率。最后,采用遗传算法及粒子群算法,求解了多目标优化问题,获得了Pareto最优解集,并在Pareto最优解集中根据安全性及汽车轻量化的要求,对方案进行了制定,分别根据两种优化算法制定了两套方案,其中,利用多目标遗传算法使B柱峰值加速度下降10.07%,脚踏板侵入量下降19.32%,整车质量下降0.9kg;利用多目标粒子群算法,使B柱峰值加速度下降4.9%,脚踏板侵入量下降6.31%,整车质量下降4.9kg,优化效果明显,均获得了较好的优化结果。对汽车碰撞安全性的优化设计具有理论和工程指导意义。
[Abstract]:With the increasing number of vehicles in possession, the requirements for vehicle safety are becoming higher and higher. Vehicle safety is mainly divided into active safety and passive safety. Active safety can be actively avoided before an accident occurs, but once an accident occurs, Therefore, as the last line of defense of vehicle safety, the optimization of vehicle collision safety performance is very important. The finite element method is usually used to simulate the vehicle collision process, but the simulation process is a nonlinear dynamic process, which takes a long time and can not meet the requirements of iterative optimization calculation. So it is necessary to use the agent model instead of the finite element model. In order to reduce the time of simulation and improve the efficiency of calculation, the main research content of this paper is to use the agent model to optimize the safety of vehicle collision. In this paper, the thickness of the main energy absorption plate in front of the vehicle is taken as the design variable. With the minimum peak acceleration of B column, the minimum of pedal invasion and the minimum of vehicle mass, as the optimization objective, The multi-objective optimization calculation of vehicle collision safety is carried out. Firstly, the sample points are selected by using Latin hypercube design method. Secondly, Gao Si radial basis model multiquadric radial basis function model and Kriging model are established, and the errors of the three models are analyzed. The approximate model with high precision is used as the agent model. The Kriging model is used for the peak acceleration and pedal intrusion of the B column, and the Multiquadric radial basis function model for the whole vehicle quality is adopted to reduce the calculation cost and improve the calculation efficiency. Genetic algorithm and particle swarm optimization algorithm are used to solve the multi-objective optimization problem, and the Pareto optimal solution set is obtained. According to the requirements of safety and vehicle lightweight, the scheme is formulated in the Pareto optimal solution set. According to the two optimization algorithms, two sets of schemes are developed, in which the peak acceleration of B column is reduced by 10.07, the pedal intrusion is reduced by 19.32kg and the vehicle mass is decreased by using multi-objective genetic algorithm, and the multi-objective particle swarm optimization algorithm is used. The peak acceleration of the B column is decreased by 4.9, the pedal invasion is reduced by 6.31, the mass of the whole vehicle is decreased by 4.9 kg, the optimization effect is obvious, and the better optimization results are obtained. It is of theoretical and engineering significance for the optimal design of the vehicle crash safety.
【学位授予单位】：重庆大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：U467.14;TP18

【相似文献】