Condition monitoring and SVM diagnosis of sliding bearings

Abstract

It is important to predict the condition of pump shaft at sliding bearing because of the wear in long time usage. However, it is difficult and not clearly known the relationship between the vibration signal and severity of wear in pump shaft. In this study, in order to monitor the condition of the slide bearings, an experimental device for the rotor systems with vertical shaft and horizontal shaft supported with water was developed. Long-term rotation experiments were conducted for both shafts. In these experiments, the torque, bearing temperature, and displacement in the x and y directions were measured and recorded. Torque and bearing temperature signals were used to label the condition as “normal” or “abnormal” state. Then, various features were calculated from displacement signals. Support vector machine (SVM) model was trained by some specific labeled feature signals. The trained SVM model could identify the state and transition of state in other time zones.