Intelligent Bearing Fault Diagnosis of Switched Reluctance Motor Using Hilbert-Huang Transformation and Region-Based Convolutional Neural Network

With the widespread trend of transportation electrification, much research has been focused on switched reluctance motor (SRM) because of its robustness and wide speed range. Though the SRM has a promising electric fault tolerance, the rotor support bearings subjected to mechanical friction are vulnerable to brinelling and corrosion. Furthermore, traditional bearing fault diagnosis approaches adopted for conventional motors are not well suited for SRM bearings because of its high torque ripples and noisy operation. Aiming at offering an efficient solution, this work proposes a novel technique by fusing the Hilbert-Huang transformation of vibration signal with region based Convolutional Neural Network. The core idea of the method is to decompose the vibration signal and employ a new kind of offline region generation algorithm for the detection of fault characteristics. The effectiveness of the designed model is tested and evaluated by a series of experiments. The results show that the proposed method achieved a remarkable diagnosis accuracy of 99.6%. Conclusively, a systematic comparison of the proposed model with the other state of the art models is carried out.