Fault detection using ANN for Three-level NPC Inverter fed induction motor drive

This paper is to present the investigation of ANN for fault detection Three-level Neutral Point Clamped (NPC) Inverter fed induction motor drive. Such detection requires only motor currents and DC-link current to be measured for detecting and identifying the power switch in which the open circuit fault or short circuit fault has occurred. Furthermore, a diagnosis controller using ANN for real-time fault detection of power switches is suggested. A computer simulation program is developed using Matlab/Simulink together with the Neural Network Toolbox for training the ANN-controller. This method has been validated experimentally using kit ACE 1104 (DSPACE).