A Voltage-Based Open-Circuit Fault Detection and Isolation Approach for Modular Multilevel Converters With Model-Predictive Control

Fault detection and isolation (FDI) is currently considered a crucial way to increase the reliability of modular multilevel converters (MMCs), which consist of a large number of power electronics submodules (SMs). This paper proposes a fast FDI approach to identifying single open-circuit faults of insulated-gate bipolar transistors in SMs for MMCs with model-predictive control (MPC). The fault detection approach is simply implemented by checking the voltage errors between the measured arm voltages and the estimated ones in the former control cycle. The fault isolation is achieved by checking the switching state directly. The proposed FDI scheme is straightforward, and no additional transducer or measurement is required. Compared with the phase-shifted pulse-width-modulation-based scheme, the MPC has a known and unchanged switching state in a sampling period, which can be utilized for fast location of open-circuit faults. Experimental results show that an open-circuit fault in the MMC can be accurately detected and located in several sampling periods.