Optimization of acoustic performance of a vehicle dash sound package

In this paper, four BEVs are chosen as study objects. For every BEV, the noise signals are measured in five different working conditions. The interior noise signals of BEVs in the frequency range from 400Hz~8000Hz are extracted as sound samples. Subjective evaluation tests are performed to acquire value of the irritability on sound samples, and psychoacoustic objective evaluation parameters of sound samples are calculated. Then, the BP neural network prediction model of sound quality of BEVs is established. SEA model of one of the four BEVs is established to analysis noise transfer path for finding out the plates with greater noise contribution. The method of positive development of sound package is used to design sound package with best acoustic performance, then the sound package is applied on the plates to perform simulation of SEA model to calculate the interior sound level of the BEV, through which the optimal psychoacoustic objective evaluation parameters are calculated. Based on the BP neural network prediction model of sound quality, the optimal psychoacoustic objective evaluation parameters are used to calculate the optimal value of irritability. By comparing with original sound quality of the BEV, the irritability is reduced and sound quality is improved after optimizing.