Expert System Based on Autoencoders for Detection of Broken Rotor Bars in Induction Motors Employing Start-Up and Steady-State Regimes

Induction motors are indispensable, robust, and reliable machines for industry; however, as with any machine, they susceptible to diverse faults. Among the faults that a motor can suffer, broken rotor bars (BRBs) have become one of most studied ones because under this fault condition continue operating apparent normality, yet severity quickly increase and, consequently, generate whole collapse motor, raising repair costs risk people or other around it. This work proposes an expert system detect BRB early, i.e., half-BRB, 1-BRB, 2-BRB, from current signal analysis by considering following two regimes: start-up transient steady-state. The method diagnose using either regime both regimes, where objective is somehow reliability result. Regarding proposed system, it consists application autoencoders, per regime, condition. To automatically separate regimes obtain envelope signal, Hilbert transform applied. Then, particle swarm optimization implemented compute separation point in signal. Once separated, autoencoders simple set if-else rules employed determine proved be effective tool, 100% accuracy diagnosing all conditions.