Abstract:In this paper, fuel cell vehicle (fuel cell vehicle, FCV) is focused on as the reserch object and the sound signals collected in different locations at idle condition as the experimental samples. This paper uses paired comparison method to have subjective evaluation test for the 24 signal samples stated above, meanwhile, it calculates the six parameters of objective evaluation describing the sound characteristics and adopts BP neural network to establish FCV sound quality prediction model, which could be uesd to calculate the impact weight that means how objective evaluation of sound quality parameters impact on the results of subjective evaluation. It is the first time to propose to use the method of BP neural network to determine impact weight measuring objective evaluation parameters contributing to the results of subjective evaluation in the process of sound quality evaluation, then it is obtained main objective parameters determining sound quality good or not. This analysis, in both fuel-cell vehicle and other areas for sound quality evaluation and analysising, plays a significant guiding role.