Complex Singular Spectrum Decomposition and its Application to Rotating Machinery Fault Diagnosis

In recent years, singular spectrum decomposition (SSD) has been recognized as a powerful signal decomposition technique and has become a useful tool for fault signature extraction of rotating machinery. A complex SSD (CSSD) approach is proposed in this paper, extending SSD to the complex domain. Moreover, the equivalent filter banks property of CSSD is studied by using the Gaussian noise. Based on the CSSD and Hilbert transform (HT), two fault feature extraction algorithms, named CSSD Hilbert time-frequency spectrum (CSSD-HTFS) and CSSD Hilbert envelope spectrum (CSSD-HES), are exploited. In order to fuse the information of multi-sensors and extract sufficient fault features, a novel fault diagnosis framework for rotating machinery is developed. Specifically, the vibration signals collected along the X-direction and Y-direction are firstly employed to construct a complex-valued signal. Next, the CSSD of the complex-valued signal is carried out to obtain a series of decomposed components associated with the positive and negative frequency components. Following, the CSSD-HTFS or the CSSD-HES algorithm is adopted to extract the fault signatures of the decomposed components. Finally, the fault pattern of the rotating machinery is recognized based on the extracted fault signatures. Several examples of fault diagnosis applications validate the developed methodology.