Fault detection of rotor systems based on Orthogonal Matching Pursuit (OMP) algorithm

Rotor systems can be affected by environmental and operational factors, which often cause faults in their operations. Therefore, fault detection of rotor systems, which involves fault diagnosis and classification, is of great importance in engineering practice. Traditional fault detection methods often rely on experienced experts and are highly subjective. To address these issues, a novel automatic fault detection method is presented, using the time-frequency images of rotating machinery and the Orthogonal Matching Pursuit (OMP) algorithm. Firstly, the fault vibration signals generated by the rotor systems are transformed into time-frequency images using the wavelet transform. Secondly, a sparse image classification method using the OMP is introduced. Finally, a sparse image classification method, combined with a time-frequency image, is used to detect faults in the rotor system. The effectiveness of the proposed method is verified by detecting the misalignment and unbalance faults of rotor systems. The results show that the method is with high efficiency, and is more accurate than the traditional KNN method for rotor fault detections.