Crank Bearing Wear Fault Diagnosis of Emulsion Pump Based on Fuzzy Support Vector Machine

The experiment research shows that as the crank bearing gradually wears the relative energy contents of low-frequency bands slightly reduce and these of high-frequency bands greatly increase. The relative energy distribution of frequency bands of vibration signal provides the quantitative evidence to the identification of the crank bearing wear fault. So the feature extraction of wavelet packet energy was used to extract the relative energy signature of frequency bands from different signals as the input feature vector of fuzzy support vector machine (FSVM). The membership grade of sample was introduced and the pattern recognition algorithm of FSVM that made full use of the fuzzy information of the crank bearing wear condition was put forward. Further, the diagnosis model of the crank bearing wear fault integrated the feature extraction of wavelet packet energy and FSVM was established. The diagnosis results show that the diagnosis model can realize the early identification of the crank bearing wear fault.