Statistical Approach for Fault Diagnosis In Three Level NPC Inverters

Fault detection and identification are critical for enhancing the reliability of power conversion systems. Consequently, fault diagnosis is a dynamic research area in the power conversion field. In the same context, this paper proposes multiple statistics-based methods for diagnosing open-circuit faults in neutral point clamped inverters(NPC) by analyzing load currents' signal distortions. In contrast with conventional current-distortions-based methods, these proposed methods are robust against load current and modulation index variation, and low-frequency disturbances. Moreover, these methods allow open-circuit faults early detection and isolation of open-switch faults with minimal computational burden. Various simulation results demonstrated the effectiveness of the proposed approach.