Bearing fault diagnosis with morphological gradient wavelet

In order to extract impulsive features and perform noise reduction in the vibration signals of defective bearings, morphological gradient wavelet (MGW) scheme is proposed. This approach is based on a combination of gradient operator and morphological wavelet theory. In order to validate the effectiveness of the proposed scheme, computer experiments are performed for two different scenarios. In these scenarios, vibration signals for two defective bearings are investigated; one with an inner race fault and the other with an outer race fault. The main advantages of the proposed MGW algorithm are its speed and simplicity of implementation. Therefore, it is suitable for real-time signal processing aimed at online condition monitoring. The proposed method is compared with two other similar algorithms namely, the morphological Haar wavelet (MHW) and the morphological undecimated wavelet decomposition (MUWD). Simulation results confirm the superiority of the proposed MGW in terms of effectiveness and speed.