An on-line neurofuzzy approach for detecting faults in induction motors

Abstract

A broken rotor bar is one of the most common type of faults that may occur in an induction motor system. This paper is devoted to investigating the possibility of performing online monitoring of the condition of asynchronous machines. The fault detection scheme uses a neurofuzzy model of the static characteristics of the motor to generate residuals. Although the influence of a cracked rotor bar and an increase in the motor loading are similar, simulation results show that the neurofuzzy model-based fault detector is able to detect the presence of a partially broken bar regardless of the loading conditions. © 2001 IEEE.