Dynamic gains robust differentiator based fault detection approach for cascaded H-bridge multilevel inverters

The deployment of a high-reliability Multilevel Inverter (MLI) able to operate effectively and unremittingly in case of partial failure is indispensable in high performance and high power grid-connected PV systems. Among the well-known MLI topologies, the Cascaded H-Bridge (CHB) inverter is characterized by its high modularity and fault-tolerance features due to the high flexibility in generating the output voltage levels. However, the effective operation of such topology needs a tight supervision to detect the failure and provide uninterrupted power supply. This paper proposes a novel open-circuit failure detection strategy based on a 1st-Order Dynamic Gains Robust Differentiator (DGRD) for 3-phase 9-cell CHB inverter. The proposed scheme estimates the current 1 ^st derivative in order to easily detect the amplified faults. This technique is considered as an improved version of the sliding mode based Levant derivative approach, where a constant gain is used instead. The presented simulation results confirm the accuracy of the proposed technique in real-time fault detection (distinction between noises and faults to avoid false alarms and misdetection).