Diagnosis of bearing faults using an advanced feature extraction based on WPT and EEV methods

Detecting faults in rolling-element bearings can be achieved by monitoring and analyzing various signals such as vibration signal, acoustic noise, and stator current. Frequency spectral analysis using signal processing technique is one of many methods which has been used in fault diagnosis in this field. Other techniques as FFT, MUSIC, ESPRIT, STFT, and DWT are used to detect electrical or mechanical faults in induction motors (IMs). Motor current signature analysis using FFT tool (MCSA-FFT) is the most commonly utilized method. However, it has significant drawbacks, including resolution issues and the limitation of being unsuitable for analyzing non-stationary signals. To address these limitations, a method for analyzing stator current based on the Wavelet Packet Transform (WPT) technique is proposed for detecting this mechanical faults (rolling element bearing faults). The stator current signal is decomposed into nine levels using the Wavelet Packet Decomposition (WPD) method. Subsequently, EEV estimation is applied to the resulting coefficients to identify bearing faults. To confirm the effectiveness of the proposed technique, we examine the performance of a healthy and faulty motor under both no-load and full-load conditions. The findings indicate that the proposed method is more effective and efficient than traditional approaches.