A recognition method for gearbox wear state based on EEMD and INPP

Abstract

Wear is the main cause of failure of gear transmission system. In order to solve the problem that it is difficult to extract fault features from complex environmental noise for condition recognition, this paper proposes a method based on EEMD and INPP for gearbox wear condition recognition. Firstly, EEMD method is used to decompose the original vibration signal of gearbox, and then the decomposition results are sorted by kurtosis criterion, and the components with large kurtosis index are selected for time-frequency domain analysis to get the time-frequency domain high dimensional feature set; then the improved Neighborhood Preserving Project (INPP) algorithm is used to reduce the dimension of high-dimensional features, and then the reduced dimension features are obtained for state recognition. Finally, the algorithm is verified by the vibration response data of gearbox and compare with several algorithms, and the results show that the proposed algorithm has stable dimensionality reduction effect, good classification effect, and shows the effectiveness of the method.