Contrary to the smoothness of the time-varying subjective sound quality evaluation (SQE) sequence of the vehicle interior non-stationary noise,the prediction sequence output by the traditional objective evaluation model fluctuates rapidly in the time domain.In this paper,based on the traditional time-varying annoyance (TVA) evaluation model of auto-motive interior non-stationary noise,the Savitzky-Golay filter (SG filter) was used to smooth the prediction sequence in time domain.The results show that compared with the prediction sequence,the smoothed evaluation sequence has a better predic-tion performance.It improves the prediction accuracy by deceasing the root mean square of prediction errors by 11.40％,im-proves the stability by decreasing the variance by 31.50％,and improves the consistency by increasing the Pearson correla-tion coefficient by 9.95％.It is concluded that on the basis of the traditional model,the prediction sequence smoothed in the time domain is an better objective evaluation method for the TVA of automotive interior non-stationary noise with higher pre-diction accuracy.
维普
万方数据