Fault diagnosis of rotating shaft systems based on wavelet entropy and GA-SVM

A new fault diagnosis method for rotating shaft system of marine diesel engine is proposed in this paper. The proposed method is an integrated application of wavelet packet, Shannon entropy, SVM (Support Vector Machine) and GA (Genetic Algorithm) theory. Based on the simulation platform for marine diesel engine shafting, wavelet packet decomposition and strong fault-tolerant Shannon entropy are used to compute the feature vectors of vibration signals, which are then served as the input vectors of SVM; GA is used to optimize the parameters of SVM when it is trained to achieve higher veracity. The study result shows that WPS-GS can get higher reliability and veracity than the conventional SVM and BP neural network, which means the provided method is more suitable for the condition monitoring and fault diagnosis of rotating shaft system. © 2013 Asian Network for Scientific Information.