Subjective Evaluation On Sound Quality Of Motorcycles

With the development of industrial technology, there is an ongoing and increasing focus on the NVH(Noise,Vibration,Harshness) performance of Vehicle. A better sound quality has a better chance of winning on the market, and it is helpful to build up unique brand significance. The sound samples on the driver's position of motorcycles at different engine speed were selected as the evaluation objects. Based on the results and methods of sound quality by domestic and overseas subjective researches, this thesis aimed on the subjective evaluation and its analytical method of the sound quality of motorcycles.Firstly, a summary of research on subjective evaluation is given and then a brief introduction of theoretic basis of sound quality is presented. The preference subjective evaluation of 40 noise samples was carried out with the paired comparison method. In error rate analysis of subjective evaluation results, i-i error, ij-ji error and circular triads were investigated in detail. The criterion for eliminating misjudgment data called weighted factor of consistency were applied in this work, and some data were eliminated for assuring the weighted factor of consistency greater than 82%.The correlation analysis between subjective evaluation and four psychoacoustic parametric, as well as A-weighted sound pressure level, had been studied. The results indicate that the others parameters interference is serious in partial correlation analysis. By controlling the interference, the correlation coefficient between preference subjective scores and loudness was changed from -0.845 to 0.18. The correlation coefficient between preference subjective scores and roughness was changed from -0.605 to 0.186. It was finally obtained that preference subjective evaluation value had a strong correlation with A-weighted sound pressure level, roughness and fluctuation strength.A model for the subjective evaluation score in relationship with the psychoacoustic parameters was obtained gained through the multiple linear regression analysis. And it is verified that the model have a good predictability with subjective experiments. From the parameters of the model, it could be concluded that the preference subjective evaluation was effected by A-weighted sound pressure levels, roughness and fluctuation strength. The A-weighted sound pressure levels is the most influential factor, roughness ranks the second while fluctuation strength has the least contribution to the result. The thesis presents some applicable measures to improve the sound quality of motorcycle by analyzing the frequency component of well-accepted noise samples with the combination of psychoacoustics parameters. To obtain a better sound quality, it is necessary to decrease the power of high frequency.