Sound quality of vehicle exhaust noise prediction by using support vector regression

The Support Vector Regression (SVR) was proposed to establish the relationship between the objective psychoacoustic parameters and the sound quality of vehicle exhaust noise. Sensory pleasantness evaluation of vehicle exhaust noise samples were obtained under the same training and test sample sets and the results were compared with that obtained through Multiple Linear Regression (MLR) prediction models. The results showed that the prediction values of SVR were close to the measured values, the mean absolute percentage error is smaller than MLR and in the 8% error range. The SVR model represented the nonlinear of sensory pleasantness and objective parameters exactly. It is suggested that SVR is an effective and powerful tool for predicting sound quality of vehicle exhaust noise.