Wavelet analysis-application of PCA-SVM in state identification method of marine centrifugal pump

This paper proposes a state identification method of centrifugal pump based on algorithms of wavelet transform, principal component analysis (PCA), and support vector machine (SVM), and provides concrete steps of the fault diagnosis model establishment of marine centrifugal pump using the method. The analysis object is the pressure pulsation signal at the exit of the centrifugal pump, and the characteristic set is constructed by the wavelet packet algorithm. The test data of the centrifugal pump in the normal state is regarded as the sample data. After PCA, the dimension of the feature set is reduced, and the feature extraction model is established. Finally, the state identification and fault diagnosis of the marine centrifugal pump are realized by applying SVM. In the practical application, the method improves the limitation of the traditional state intelligent identification method and the difficulty of setting the threshold, and its validity has been verified in the experiment.