基于时频分析的客车加速通过噪声声源识别

应用小波包分析理论发展了基于时频分析的客车加速通过噪声声源识别技术。根据选择运行及部分覆盖方法，分别去除各声源影响，同时测得客车通过测试区域的车外加速噪声。对于噪声信号进行多层小波包分解并重构，计算每个小波包的能量及该层小波包的总能量，通过不同信号的小波包能量比较达到声源识别的目的。研究表明，该方法不仅可以清晰地分辨各噪声源能量在时-频域的分布情况，而且通过能量计算可以识别主要噪声源，为采取相适应的降噪措施提供了理论基础。

Coach Noise Source Identification Based on Time-frequency Analysis

A coach noise source identification technique based on time-frequency analysis is developed by applying wavelet packet theory. By removing, covering or shutting down the components or assemblies or closing some sound transmitting channels one by one, the vehicle passing-by noises are measured respectively. Then the decomposition and reconstruction of multi-layer wavelet packets are conducted on these noise signals, and the energy of each wavelet packet and the total energy of that layer of wavelet packets are calculated. Through the comparison of wavelet packet energy of different noise signals, the noise sources are identified. It is shown that the technique can not only clearly discern the energy distribution of different noise source in time-frequency domains, but also identify the main noise sources through energy calculation, laying a theoretical foundation for adopting appropriate measures to reduce the acceleration noise of the vehicle.