Correntrogram: A Robust Method for Optimal Frequency Band Selection to Bearing Fault Detection

Correntropy includes not only the second-order statistics of the signal, but also the higher-order statistics of the signal. Therefore, correntropy is an effective tool to deal with nonlinear and non-Gaussian signals. In order to solve the problem that it is difficult to select the optimal frequency band of bearing fault vibration signal under the interference of Gaussian and non-Gaussian Noise, a new optimal frequency band selection method is proposed, which is named as Correntrogram. Firstly, the correntropy of the signal is calculated. Then correntropy is decomposed into multiple frequency bands using the 1/3-binary tree structure and the optimal frequency band is selected according to the L-2/L-1 norm. Finally, the squared envelope spectrum of the optimal frequency band is calculated and bearing fault characteristics frequency can be accurately identified. The results of simulation and experiment show that Correntrogram can correctly select the optimal frequency band of bearing fault vibration signal under the interference of Gaussian and non-Gaussian noise, which has good robustness, and its performance is better than that of traditional Kurtogram.