Automatic Diagnosis Method for Rolling Bearing Using Measured Signal from Distant Points

摘要

Though acceleration sensors must be set as near the diagnosed bearings as possible, it is difficult to set acceleration sensors near the diagnosed bearings in some field equipment, and in some cases bearing faults are diagnosed at distant points from the diagnosed bearings. So, in this research, to solve the problem described above, we investigated the basic method by which bearing faults can be automatically diagnosed simply and precisely using the signals measured by the acceleration sensors at distant points from the diagnosed bearings. In this paper, first we proposed the method by which the optimal cut-off frequency of high-pass-filter is automatically searched and decided by genetic algorithm and tabu search to extract vibration signals of the abnormal bearings, and the method by which bearing faults can be automatically diagnosed by decision tree. Then, these methods were applied to bearing diagnosis using the vibration signals measured at distant points from the diagnosed bearings rolling at middle speed, and the efficiency of these methods have been verified by the results of the automatic bearing faults diagnosis.