Gearbox fault detection by Haar wavelet and envelope spectrum analysis of motor current

Abstract

Motor current signature analysis (MCSA) provides a non-invasive approach to diagnose the gear fault. The performance of MCSA at different speed and load torque conditions is studied using Fast Fourier Transform (FFT) to extract fault features. The results show that it would be better to perform fault diagnosis under low speed and heavy load conditions. And it fails when adopting FFT to diagnose the fault in some cases with high speed and light load. Then this paper proposes a new fault detection method which combines Haar wavelet transform and envelope spectrum analysis (ESA). Firstly, the current signal is decomposed to 1 level with Haar wavelet transform to highlight the impact component caused by the local gear fault, then envelope spectrum is carried out with the detail signal to trace the fault characteristic frequency. Experimental results show that the proposed method is more effective than conventional FFT to extract fault features. Using this method, the performance of MCSA is improved, and the gear fault can be detected in a wider range of speed and load conditions.