奇异值分解结合频率切片小波的齿轮故障特征提取

摘要

Frequency slice wavelet transform is a powerful time-frequency analysis method. But its ability of fault characteristic identification is weak under the condition of strong noise background. Thus, a method of fault characteristic extraction combining the singular value decomposition with the frequency slice wavelet transform is proposed. First of all, the Hankel matrix is constructed using the original signal, the reconstruction order is determined based on the criterion of the unilateral maximum in the singular value difference spectrum, and the de-noising process is carried out. Secondly, the whole frequency domain analysis is performed for the de-noised signal using the frequency slice wavelet transform, and the distribution interval of the signal component is confirmed. Finally, the slice refinement analysis is performed for the signal with concentrated energy, and the fault characteristic of the gears can be extracted from the time-frequency spectrum of the reconstructed signal. Results of numerical simulation and signal measurement show that the proposed method can achieve accurate identification of the operation condition of the gears, and has some engineering significance.%频率切片小波变换是一种有力的时频分析方法，但在强背景噪声条件下其故障特征识别能力不足，故提出奇异值分解结合频率切片小波的故障特征提取方法。首先利用原始信号构造Hankel矩阵，根据奇异值差分谱单边极大值原则确定阶次并进行降噪处理，继而利用频率切片小波对降噪信号进行全频分析，确定信号分量分布区间之后，对能量集中的信号进行频率切片细化分析，用时频图及重构信号提取齿轮故障特征。通过仿真及实测齿轮的点蚀信号分析，表明该方法能够实现齿轮运行状态的准确判别，有一定的工程实际意义。