Feature Extraction and Intelligent Fault Diagnosis of Marine Machinery

PurposeThe present work proposes a new method to realize the intelligent condition monitoring and fault diagnosis of marine machinery.MethodTo realize feature extraction, the time averaging decomposition method (TAD) is proposed to extract features from vibration signal. And a feature selection method, confusion score feature selection (CSFS), is proposed in this paper.ResultsThe simulation data and the experimental data were analyzed in this work. Several signal decomposing method is compared in this paper and TAD is performed better than other methods. And CSFS method has better performance than other feature selection methods compared in this paper. Besides, the CSFS method will not only improve the prediction accuracy but also reduce the classifier computing time.ConclusionThe proposed method is experimentally validated with a marine blower fault experiment, which proves the effectiveness of this proposed method.