基于Gabor变换和盲源分离的旋转机械故障诊断方法

摘要

In the traditional blind source separation( BSS ),the condition of actual mechanical equipment is very difficult to satisfy that the source signals must be mutually statistically independent. A new method of rotating machinery fault diagnosis based on Gabor transform and BSS is proposed. Firstly,the common frequency components of source signals can be obtained by the ratios of the coefficients of the mixed signals in Gabor transform coefficient. Then,the new observed signals are obtained by filtering,and the jointly approximate diagonalization of eigen-matrix( JADE)is applied to the new observed signals. Even if the source signals are correlative,or there is more than one Gaussian signal in the sources,the new method can get better separation performance. Simulation and experiment results verify the effectiveness and feasibility of the proposed method.%传统的盲源分离方法要求源信号相互统计独立，但是实际机械设备很难满足这个条件。为此，提出了一种基于Gabor变换和盲源分离相结合的旋转机械故障诊断方法。首先通过不同混合信号的Gabor变换系数之间的相互关系，得到源信号间的公共频率成分，然后对观测信号进行滤波处理，得到新的观测信号，最后利用矩阵联合对角化方法进行分离，实现相关源信号盲分离。该方法突破了传统盲源分离方法中要求源信号相互统计独立且最多只能有一个高斯信号的限制，仿真和实验结果验证了该方法的有效性和可行性。