基于降维的驾驶员侧约束系统高维目标优化

为解决将高维目标变为单目标优化时各子目标不能同时较优,而多目标算法直接用于高维目标优化时又存在难以找到一个有代表性的Pareto非劣解集问题,在某轿车驾驶员侧约束系统的优化过程中提出了乘员损伤准则与多目标算法协同优化的方法。在已有相关损伤准则基础上根据最新版的FMVSS 208和ECE R94法规提出了适合研究问题的损伤准则;以提出的损伤准则为媒介,将一个高维目标优化问题降为一个低维目标优化问题,通过灵敏度分析、实验设计、多项式近似模型筛选出优化设计变量并得到近似模型,用多目标算法NSGA-Ⅱ对近似模型进行计算得到Pareto非劣解集,将得到的Pareto非劣解集中的每个解代入损伤准则损伤值计算公式,升序排列得到各子目标同时较优而损伤值最小的优化解。最终的优化结果表明：该方法很好地解决了乘员约束系统的高维目标优化问题,优化效果明显。

High-dimensional Objective Optimization of Driver Side Restraint System Based on Reduction of Dimension

To solve the problems that each sub-goal cannot reach optimum at the same time when multiple objectives were integrated into a single objective optimization, and using multi-objective genetic algorithm for high-dimensional objective optimization it was hard to find a representative non-inferiority Pareto solution, an optimization method coordinated occupant injury criteria with multi-objective evolutionary algorithm was proposed for optimizing the driver side restraint system of a passenger car. Based on the existed injury criterion and the latest version of FMVSS 208 and ECE R94, a new injury criterion (IC) was proposed for the problems mentioned above. The IC was used to reduce a high-dimensional objective optimization problem to a low-dimensional objective optimization problem. Sensitivity analysis and experimental design were conducted to get an approximate model, and NSGA-Ⅱ evolutionary algorithm was used to calculate non-inferiority Pareto solution. Then every non-inferiority Pareto solution was put into the injufy formula, which ordered in ascending order, to obtain an optimization solution that each sub goal can reach optimum at the same time, meanwhile, the injury value was minimum. The optimized results show that high-dimensional optimization problems of occupant restraint system are solved, and the effect of optimization is significant.