Fault diagnosis of marine main engine shaft using support vector machines

Abstract

In this paper, a method is proposed to diagnose faults of marine main engine shaft. Since empirical studies show various faults of fuel oil system induces the variation of vibration signals, we propose to diagnose faults of marine main engine shaft using vibration signal from engine. The proposed method consists of three steps. First, a wavelet analysis method is used to characterize the power spectrum of the vibration signal. Next, principal component analysis (PCA) is used to extract the most distinctive feature for faults diagnosis. Finally, the extracted features are fed into a set of pre-trained support vector machines (SVM) for fault diagnosis. Importantly, we use a cascade framework to organize a set of SVMs, for identifying different types of faults. Experimental results are presented to show that our proposed method is able to detect and identify different types of faults accurately.