A Bi-directional Temporal Sequence Approach for Condition Monitoring of Broken Rotor Bar in Three-Phase Induction Motors

Abstract

Condition monitoring and diagnosis of rotor broken bar (BRB) induction motors is critical for the proper functioning of imperative industrial applications. Traditional fault diagnosis methods typically employ deep learning techniques that are trained on the chronological sequence of the accumulated signals. Extracted features in the direction of inverse time-domain signals is usually neglected in these particular circumstances. This article suggests a Bidirectional LSTM (Bi-LSTM) to establish a fault diagnosis model of timely detection and monitoring of BRB faults, which is based on a conventional Long Short Term Memory (LSTM) network. The Bi-directional LSTM network is used to extract features from both directions of output signals from the induction motor. The results show that the method is able to undertake concise diagnostic operations.