Approach to extracting gear fault feature based on mathematical morphological filtering

Abstract

To extract fault feature of gear, a novel approach is proposed according to the signal characteristics based on morphological filtering. As a nonlinear filtering algorithm for digital signal processing, morphological filtering is able to identify the feature of fringe and shape of the signal. Lorenz signal is processed by mathematical morphological filtering via various structuring elements, and the effect of noise reduction and nonlinear feature reservation of morphological filtering is validated. The vibration signal of gear teeth broken is processed by morphological closing operation via the flat structuring elements, and the length of the structuring elements is 0.6 to 0.8 times to the length of gear impact period. Then, the filtered signal is analyzed by Fourier frequency spectrum. The results show that the impact feature, which can not be identified from noisy data directly, is successfully extracted by morphological filtering.