Pitting fault detection of a wind turbine gearbox using empirical mode decomposition

Abstract

The conventional method of detecting a gear fault is to demodulate the vibration signal collected from the gearbox based on the Hilbert transform; however, this requires human intervention and lacks sophistication. Empirical mode decomposition (EMD) is a significant timefrequency tool for adaptively decomposing vibration signals into a collection of intrinsic mode functions (IMFs); a fault feature can be extracted from one of IMFs to reveal the fault location and fault level of a gear or bearing in the mechanical drive system. In this paper, a multi-harmonic vibration model of a gearbox with fault modulation is presented, a conventional demodulation analysis using Hilbert transform is introduced, and the principle of EMD is illustrated. The Hilbert demodulation analysis and EMD are applied to processing field vibration signals collected from a wind turbine gearbox to detect a gear-pitting fault. The results show that EMD can extract the fault modulation information more adaptively and intelligently than Hilbert demodulation analysis can.