摘要
以某轿车为例,建立了其FE-BEM的低频声学响应模型,并通过实车试验验证了模型的正确性。基于等矩形带宽(ERB)尺度,并以车速50km/h时采集的驾驶员右耳处20-200Hz频带的声信号为基础,运用正交试验设计生成16个声样本。编程计算了各样本的客观参量,并完成了主观评价试验。综合遗传算法与支持向量机,构建了声样本的声品质预测模型。以低频段各REB频带声压级为变量,声样本主观烦躁度最小为目标,建立了声品质优化模型,优化得到了比原来显著改善的声样本,通过主观评价结果验证了该方法的有效性。
A FE-BEM model for the low-frequency acoustic response of a car is established and validated by real vehicle test. Based on the metric of equal rectangle bandwidth( ERB) and the sound signal at the right ear of driver with a frequency range of 20-200 Hz collected at a vehicle speed of 50km/h,16 sound samples are generated by using orthogonal design of experiment. A program is developed to calculate the objective metrics of each sound sample and a subjective evaluation test is conducted. Then a sound quality prediction model for sound samples is built with the combination of genetic algorithm and support vector machine. Finally a sound quality optimization model is set up with minimizing the subjective annoyance of sound sample as objective and the subtotal of sound pressure level of each ERB band at low-frequency range as variables. An optimization is performed with an optimized sound sample significantly better than original one is obtained,and the effectiveness of the method proposed is verified by another subjective evaluation test.
出处
《汽车工程》
EI
CSCD
北大核心
2017年第9期1074-1080,共7页
Automotive Engineering
基金
重庆工程职业技术学院院级科研项目(KJB201712)
重庆市基础与前沿研究计划项目(CSTC2015jcyjBX0075)
中央高校基本科研业务费(106112016CDJZR335522)资助