基于时频偏相干分析的关门异响声源识别

Identification of Door Closing Abnormal Noise Source Based on Time-frequency Partial Coherence Analysis

针对基本的偏相干算法识别非稳态噪声源的误差缺陷,提出一种适用于瞬态问题的时频偏相干算法。该算法在声振信号的时间均匀切片基础上,以一定的采样间隔求取测点信号与响应点噪声的偏相干关系,并根据信号的重叠率对结果进行平均。利用该算法对某车型关门振动噪声的振源进行识别,分别从时域和频域的角度分析车门系统各部件对关门振动噪声的贡献。结果表明:在时域角度,玻璃、前导轨以及后导轨的振动贡献时间较长;在频域角度,比较关门噪声的主要频率的峰值带宽以及偏相干函数的RMS值,发现玻璃和内板的问题较为严重。依据此分析结果对车门进行改进,关门振动噪声的声品质改善明显,验证了该算法对瞬态噪声源识别的有效性。

In order to identify the unsteady noise sources more accurately,a time-frequency partial-coherence analysis was proposed.Based on the time uniformly slicing of the signal,the partial coherent relation between the vibration signal at the measurement point and the noise signal at the response point was calculated every sampling interval.Then,the result was averaged according to the overlap rate.This algorithm was used to identify the vibration source of the door closing noise of a vehicle and analyze the contribution of different parts to the door closing noise in the time domain and frequency domain.The results showed that in the time domain,the vibration contribution durations of glass,front guide and rear guide were longer.In the frequency domain,comparing the peak bandwidth of the main frequency of closing noise with the RMS value of partial coherence function,it was found that the noise problem of glass and inner plate was more serious.Based on the analysis results,the door structure was optimized,and the sound quality of the door closing noise was improved obviously,which verified the effectiveness of the method to identify the transient noise source.