LAUSR

People’s primary impression of the sound quality of a car usually comes from door closing sound, which brings the direct subjective feelings to customers. This article obtains a sample library of the sound of closing doors, conducts subjective evaluation and extracts objective evaluation parameters. The correlation between subjective and objective evaluation parameters is analyzed in this study. Then the Back Propagation (BP) neural network and support vector machine (SVM) prediction model is established with loudness and sharpness as inputs and subjective evaluation results as outputs. After comprehensive comparison, support vector machine prediction model with better training effect is used to predict the sound quality of door closing. Subsequently, the finite element model (FEM) and boundary element model (BEM) of automobile door are established for door closing sound simulation. The acceleration of the door obtained from the finite element simulation analysis is taken as the boundary condition, which is imported into the BEM of the door. The frequency domain diagram of the sound pressure of the car door is obtained using the boundary element method. The loudness and sharpness are extracted from the sound pressure diagram. They are imported into SVM prediction model to obtain the evaluation value of door closing sound quality. Eventually, the experimental research on the vibration and noise of car door closing is carried out. The acoustic signals obtained from the experiment and simulation are compared to verify the accuracy of the simulation results. After completing the above verification, from the perspective of sound quality, the sound quality of the door closing is improved by optimizing the structure of the door inner panel. Since then, a set of flow from door closing sound quality simulation to optimization to evaluation and prediction has been established. It provides an important reference for related research.