车内异响自动识别方法研究

Research on Automatic Recognition Method of Vehicle Abnormal Noise

车内异响的识别是改进优化异响声源的前提。根据异响声音信号非平稳突变的特点,提出一种基于Mel频率倒谱系数、小波包能量和支持向量机的车内异响识别方法。分别提取各种异响声音的Mel频率倒谱系数和小波包能量参数,以此作为支持向量机的输入向量,实现对车内常见的4种异响声音的识别。实验结果表明,该方法可有效识别4种常见的车内异响声音,且与实验人员主观评价的方法相比,其分类的准确率可提高到90%以上。

Recognition of abnormal sound in vehicle is the premise of improving and optimizing abnormal sound source. Based on the characteristics of non-stationary mutation of abnormal sound signal, this paper proposes a method of vehicle interior abnormal sound recognition based on Mel frequency Cepstrum coefficient, wavelet packet energy and support vector machine.Mel frequency Cepstrum coefficients and wavelet packet energy parameters of various abnormal sounds are extracted respectively, which are used as input vectors of Support Vector Machine(SVM)to recognize 4 common abnormal sounds in the car. The experimental results show that this method can effectively recognize 4 kinds of common interior noise, and the classification accuracy can be improved to more than 90% compared with the subjective evaluation method of the experimenters.