Wheel bearing fault diagnosis based on minimum entropy deconvolution method
WANG Han1， HE Liu2
（1. Central Academy of CSR Corporation Limited，Beijing 100036，China；
2. State Key Laboratory of Traction Power，Southwest Jiaotong University，Chengdu 610031，China）
Abstract: A new approach to diagnose wheel bearing failure has been proposed with minimum entropy deconvolution（MED） to extract weak fault features of wheel bearings in strong background noise and ensure in actual signal detections that the detection signals are undistorted when passing from fault points to detection points. The core of this new approach was to design an optimal filter via MED， which was used to filter the vibration signals of wheel bearing axle boxes and make them close to the original impact signals， that is， to eliminate the interfering signals of propagation paths. The signals， after filtering， were analyzed with envelope spectrum to diagnose wheel bearing failure. Experiments have indicated that the MED method can accurately detect the fundamental frequency and harmonic components of wheel bearing faults.
Keywords: wheel bearings； MED； envelope spectrum； fault diagnosis