Bearing Fault Detection Via Stator Current Noise Cancellation and Statistical Control

Abstract

This paper proposes a new approach to detect in situ bearing faults via stator current monitoring. For in situ bearing faults, the characteristic bearing fault frequencies may not exist, particularly at an early stage. In addition, the bearing fault signatures are usually subtle compared to the dominant components in the sampled stator current. Therefore, in this paper, a noise cancellation technique is used to suppress those dominant components that are not related to a potential bearing fault. The remaining components, i.e., the noise-cancelled stator current, are then closely related to the health condition of the bearing. Furthermore, it is observed that under the presence of a bearing fault, the noise-cancelled stator current displays a significant amount of degrees of uncontrolled variation in its magnitude. The uncontrolled variation is detected by observing the samples falling outside the three-sigma limits on Shewhart's control charts. Therefore, it is possible to detect in situ bearing faults by detecting the variation in magnitude of the noise-cancelled stator current, as verified by online experiments performed in this paper.