Complexity based on synchrosqueezing analysis in gear diagnosis

Abstract

The Synchrosqueezing is a special case of the reassignment method and concentrates the timefrequency representation (TFR) in a scale dimension. Compared to other TFR enhancement methods, the synchrosqueezing offers better adaptability, less deformation for IF profile and an exact reconstruction formula for constituent components. This paper deals with the investigation of descriptors based on the combination of the synchrosqueezing transform (SST) and Lempel-Ziv complexity methods. This last one transforms the analyzed signal into a data sequence. In the first part, the vibration signal components are extracted by using the synchrosqueezing transform and the reconstruction method. Afterward, the Lempel-Ziv complexity values are calculated. Since the complexity values are not dependent on the magnitude of the measured signal, the proposed method is less sensitive to the data sets measured under different data acquisition conditions. This approach is applied for monitoring and diagnosing the defects during a fatigue test on a first gear reducer and also when varying the load on a second gear reducer by using the recorded vibration signals. It can also provide a new way for feature extraction and recognition of gear system faults.