Diagnosing Sliding Bearing Failures using the Frequency Domain Integration and PCA Methodology

Given the difficulty of the fault diagnosis of sliding bearings in induction motors, this paper designs a fault diagnosis method of the sliding bearing based on frequency domain integration and principal component analysis (PCA) algorithm. Taking the acceleration signal of the actual faulty motor bearing as an example, the displacement signal was obtained by frequency domain integration. The direct plotting of the axis trajectory led to a confused graph making it difficult to identify the fault. Then, the raw displacement signal was denoised by using the PCA-based signal denoising algorithm, and the axis trajectory was redrawn using the denoised displacement signal. The results showed that the purified axis trajectory is clear and the features are obvious, and the evident rotor misalignment fault of the motor can be identified. Compared with the denoising method of clustering ensemble empirical mode decomposition (EEMD) and empirical wavelet transform (EWT), the method presented in this paper can obtain a clearer axis trajectory, which helps to realize the diagnosis of motor sliding bearing faults.