Abstract
Winding insulation failures are among the most severe electric faults in industrial induction motors. It is believed that these failures start as minor inter-turn faults that eventually lead to serious insulation breakdowns. Therefore, using an online detection technique in early stages could be quite advantageous for avoiding serious failures. In this study, we have focused on the asymmetries in a three-phase circuit, and investigated the negative sequence components of the line current and voltage under various complicated conditions, such as unbalanced supply voltages, different values of the number of shorted turns, and different loads. Based on the experimental results and steady-state equivalent circuit models of an induction motor, we have developed a highly accurate analysis algorithm for the inter-turn stator winding fault, to provide an early warning of impending failures. The proposed algorithm utilizes an off-diagonal term of the sequence component admittance matrix calculated from negative sequence current, voltage and admittance. To improve fault detection performance, the algorithm is exposed to the normal operating range of the motor in the learning stage.
