Recent Advances of Neural Network based Methods in Induction Motor Fault Diagnosis

Preventing induction motors (IMs) from failure and shut down is important to maintain functionality of many critical loads in industry and commerce. This paper provides a comprehensive review of neural network (NN) based fault detection and diagnosis (FD/D) methods targeting all the major types of faults in IMs. While many pioneer works have been published up to 2014, the focus of the review is laid on the state-of-the-art of NN-based FD/D techniques after 2014, i.e. in the 5-years' time frame 2015-2019. The different NN-FD/D methods are introduced and classified into categories depending on the processing domain of the signal that will be used next for feature extraction and pattern recognition. Comparisons of the reviewed papers are given in tables for fast referring. Finally, a dedicated discussion is provided, which presents recent developments, trends and remaining difficulties regarding to FD/D of IMs, to inspire novel research ideas.