基于Kriging模型及NSGA-Ⅱ算法的前围声学包优化

Dash sound package optimization based on Kriging model and NSGA-Ⅱ algorithm

利用统计能量分析方法,首先建立了计算汽车前围板传递损失的统计能量分析模型,采用最优拉丁超立方方法生成了36个声学包方案的试验点,其中以覆盖率、堵件厚度、PU泡沫厚度及EVA面密度为设计变量和以前围子系统隔声量和声学包重量为优化目标,并计算了不同声学包方案下前围子系统的传递损失。然后依此建立Kriging近似模型并验证模型的可信度,采用NSGA-Ⅱ算法进行以声学包隔声量及重量为目标的多目标优化,获得了Pareto最优解集。赋予声学包隔声性能及总重量一样的权重,得到了覆盖率、堵件厚度、PU泡沫厚度及EVA面密度的最优值,并任取三组试验点计算其传递损失和整车中的基于能量的隔声量和驾驶员头部的声压级,从而验证该结果的正确性。结果表明,该最优值能够使得声学包在隔声性能与重量之间取得最佳平衡。

Taking advantage of the statistical energy analysis （SEA）method,a SEA model for vehicle dash to calculate the sound transmission loss （TL）was established.By using the optimal Latin hypercube method,36 test points of the dash sound package were generated,based on four design variables in terms of coverage,blocking thickness of holes,PU foam thickness and EVA surface density as well as two optimization targets in terms of insulation performance and sound package weight.The TLs of different sound package schemes were calculated.Then the Kriging approximation model was built and also verified.The multi-objective optimizations of the insulation performance and the sound package weight were performed by appling NSGA-Ⅱ algorithm,and the Pareto optimal solution set was obtained.The optimal values of coverage,blocking thickness,PU foam thickness and EVA surface density were obtained,giving the same weight to the insulation performance and the sound package weight,and the results were verified by calculating at three randomly selected test points the TL,the power based noise reduction （PBNR）in the whole vehicle and the sound pressure level （SPL）on the driver＇s head.The results show that the best balance between the insulation performance and the weight of dash sound package can be acquired in accordance with the optimal values.