Bearing fault diagnosis based on bispectrum and bispectrum entropy feature

Fault feature extraction and application is the key technology of fault diagnosis. In this paper, a fault diagnosis method using bispectrum and bispectrum entropy as the fault feature parameters is put forward. Bispectrum entropy as the information entropy in bispectrum domain can reflect the complexity of information energy. When the structure is failed, the distribution of bispectrum will be changed. bispectrum entropy can reflect this change and achieve good separation of the different types of fault. Vibration signal in different bearing states of a secondary drive gearbox is compared and analyzed, bispectrum energy spetrum and bispectrum entropy are extracted. Feature vector is set up via bispectrum entropy for the fault pattern recognition and diagnosis by BP neural network. The analysis result proves that bispectrum entropy is more sensitive to fault characteristic and can separate the fault of bearing.