Gearbox fault diagnosis method based on wavelet packet analysis and support vector machine

This paper presents an intelligent method for gear fault diagnosis based on wavelet packet analysis and support vector machine (SVM). For this purpose, two experiments were selected to verify the proposed method. One is a spur gear of the motorcycle gearbox system. Slight-worn, medium-worn, and broken-tooth were selected as the faults. In fault simulating, two very similar models of worn gear have been considered with partial difference for evaluating the preciseness of the proposed method. The other one is a helical gear of a gearbox system. Broken-tooth and crack in root of gear were selected as the faults. Raw vibration signals were segmented into the signals recorded during one complete revolution of the input shaft using tachometer information and then synchronized using cubic spline interpolation to construct the sample signals with the same length. Next, standard deviations of wavelet packet coefficients of the vibration signals which have been normalized and dimension deducted using principal component analysis (PCA) were considered as the feature vector for training purposes of the SVM. The parameters of SVM are optimized using particle swarm optimization (PSO). Its effectiveness is verified by experimental results.