The method of scale - energy fuzzy clustering based on series wavelet analysis in studying fault diagnosis of gearbox

Vibration signals is the information carrier of gearbox working, in order to monitor and diagnose the gearbox running states. This paper proposes a method of scale - energy fuzzy clustering based on series wavelet analysis processing gearbox vibration (non-stationary) signals that we draw the energy of the signal under various running states with different measurement as eigenvector. Because vibration signals of gear and roller bearings in the gearbox are not only related to frequency but also to time. Series wavelet analysis is a time-frequency (time-scale) analysis method. It can express both time and frequency in the same time. The eigenvector of scale-energy shows similarity with same state and dissimilarity with different states. So we use the fuzzy clustering methods to detect and diagnose different conditions of gearbox. The diagnosis result is satisfactory. It shows that series wavelet analysis can supply a convincing analysis means for gearbox fault diagnosis.