New fault diagnosis method for rolling bearing based on PCA

A fault diagnosis approach is proposed for rolling bearing based on the principal component analysis (PCA). Multiple features selected from the time-frequency domain parameters of vibration signals are analyzed. Firstly, by wavelet packet transformation, the wavelet packet energy spectrums of vibration signals are extracted from the different frequency bands. Meanwhile, the time domain statistical features, such as mean value and kurtosis, are also calculated. Then the PCA is used to obtain the best description features from the combination of energy spectrums and statistical features. Finally, a neural network model is established to implement the diagnosis of rolling bearing faults. Practical rolling bearing experiment data is used to verify the effectiveness of the proposed method.