Review of Deep Learning Algorithms in Motor Bearing Fault Detection

Abstract

Bearing fault in any rotary machines can cause equipment to break down thus causing critical safety, environmental or economic effect. Many mechanical equipment operate under tough working environment, which makes them vulnerable to various types and degrees of faults. As a result, bearing fault detection (BFD) and consistent monitoring of the health status of bearings has become important so as to ensure efficiency, avert complete breakdown or any catastrophic event and prevent/reduce financial loss. This has attracted researchers to work on BFD during the past few years because of its great influence on the operational continuation of many industrial processes. This paper provides a survey on some deep learning (DL) methods for motor BFD. Some common existing DL methods are brie y reviewed, highlighting their contributions, drawbacks and their significance in motor BFD. Finally, we point out a set of promising future works and draw our own conclusions by recommending long short term memory (LSTM) autoencoder (AE) as the best method to use for BFD based on certain advantage that we presented in this paper.