Research On Prediction Of Sound Quality Of Vehicle Door-slamming Sound Based On Improved Complex Analytical Wavelet Method

With the development of domestic automobile industry,the performance of NVH is more and more concerned by vehicle manufacturers and consumers.In this field,sound quality is becoming the focus of research.The current research focus has gradually shifted to unsteady sound quality.For the study of the unsteady sound quality,the traditional psychological parameters which are suitable for steady sound quality can not express the characteristics of the unsteady signals,which will have a great influence on the prediction accuracy of the sound quality.To solve this problem,a series of time-frequency analysis methods are used to extract the characteristics of acoustic sample signals as evaluation parameters at home and abroad.The selection of characteristic parameters of sound signals is the key.The main contents of this paper are as follows:Firstly,because the traditional objective psychological parameters predict sound quality unsteady inaccurately,the paper proposed the method that use the sound signal characteristics as objective parameters to predict sound quality of the Vehicle Door-slamming Sound,and the sound signal characteristics are obtained by time frequency signal method.And the proposed method was compared with the traditional objective psychoacoustic parameter to verify it's effects.Secondly,using a combination of EMD and analytic wavelet phase way to extract the sound signal feature parameters.Used EMD to decompose the signal into intrinsic mode function(intrinsic mode,function,IMF),and got rid of the IMF component.Based on frequency spectrum of effective IMF,we divided the critical frequency band of human ear hearing into several frequency bands,and used the bands as the frequency component of the complex analytic wavelets.Then sound signal is decomposed into a plurality of signal components with the complex analytic wavelet,and calculate the weight of each signal component after the energy ratio feature vector as the sound parameter.finally,using EMD and Complex analytic wavelet decomposition to extract the signal to energy ratio after weighting as objective parameters,and BP neural network as a mathematical model to predict the sound quality,and compare which method is suitable for predicting the sound quality of impact sound samples.This article takes the automobile door closing sound as the data sample,do some research on the objective evaluation parameter and prediction method of the car closed sound quality by the time-frequency processing method.Using EMD decomposition to extract frequency components and the decomposition process of complex analytic wavelets is optimized.The objective parameters extracted by the two time-frequency methods are compared and analyzed to predict the sound quality of impact process.And the objective psychoacoustic parameters of psychological acoustics are compared with them.The results show that the signal features extracted based on the time-frequency method are more suitable for evaluating the non-stationary noise quality than the objective psychoacoustic parameters.Compared with the traditional objective psychoacoustic parameters and the energy feature parameters extracted based on EMD decomposition,complex analytic wavelet transform has the best performance to predict the quality of door-slamming sound.