Three-Level NPC Inverter Fault Diagnosis using Wavelet Packet Transform and Statistical Analysis of the Emitted Near-Field

This research paper deals with the open-circuit fault diagnosis in three-phase three-level Neutral Point Clamped (NPC) inverter. The proposed non-intrusive method identifies the faulty MOSFET based on the emitted near-field above the power conversion system. To detect the faulty switch, the wavelet packet transform is applied to extract a distinguishable electromagnetic signature. For the fault isolation, we focus on the two statistical moments, which are the mean value and the kurtosis of the radiated emissions. The proposed approach is verified through experimental results.