中文摘要：

建立了轿车FE-SEA混合模型,把仿真值与试验结果进行对比,验证了模型的可靠性。以防火墙和地板各层声学包材料的种类和厚度为设计变量,驾驶员右耳旁声压级、声学包总重量和总价格为优化目标,采用正交试验和灰色关联分析相结合的方法确定了各层材料种类和厚度的最优水平。以各层材料的厚度为设计变量进行进一步研究,采用最优拉丁超立方设计抽取75组样本点并通过仿真计算其响应值,任选67组样本点建立Kring近似模型,其余8组样本点验证近似模型的可靠性。最后以该近似模型为基础执行多目标优化,与原始声学包相比,驾驶员头部声腔声压级降低了0.4 d B（A）,重量减轻了33.21%,价格降低了14%。

英文摘要：

FE-SEA（ Finite element-statistical energy analysis） hybrid model of a passenger car was established,and verified the reliability of the model by comparing simulation results with experiment results. Taking each layer sound packages＇ material types and thickness of the firewall and floor as design variables,and selecting the sound pressure level（ SPL） at driver＇ s right ear,the total weight and total price of sound packages as optimization targets,the grey relational analysis（ GRA） was coupled with orthogonal experiment methods to determine the optimal level of material types and thickness in each layer. Using thickness of each layer as design variables for the further study,the method of optimal Latin hypercube design was adopted to generate 75 sample points,and their response values were acquired through simulations,choose 67 sample points discretionarily to establish a Kring approximate model,and the remaining 8 sample points was used to verify the reliability of the approximate model.Finally,multi-objective optimization was performed based on this approximation model,compared with original sound packages,the SPL at driver＇s right ear decreases by 0.4 d B（ A）,the total weight and total price of sound packages decreases by 33.21% and 14% respectively.