Real-time hardware-in-loop based open circuit fault diagnosis and fault tolerant control approach for cascaded multilevel inverter using artificial neural network

Abstract

Multilevel inverters (MLI) are finding widespread in various engineering and commercial applications owing to their immense performance. The cascaded H-bridge (CHB) inverter is the most potential MLI topology for renewable energy applications. The successful operation of the CHB-MLI depends on the integrity of the semiconductor devices and capacitors. Irrespective of its benefits the huge number of switches decreases the reliability of the inverter. Concerning reliability, this article proposes a fault-tolerant (FT) CHB MLI for solar photovoltaic applications. The proposed CHB MLI can withstand both the single and multiple open circuit faults in all the H-bridges of the CHB topology. The diagonally opposite switch pairs of CHB topology have similar fault features which lead to difficulty in finding the fault switches using the analytical fault diagnosis methods. Hence an artificial intelligence (AI) based fault diagnosis (FD) and FT operation of CHB MLI are interpreted. The proposed model offers complete FD and FT operation within one fundamental cycle which is advantageous relative to the existing methods. Compared to the existing methods, the proposed AI-based fault diagnosis strategy achieves a shorter diagnosis time and provides 96% classification accuracy between various fault conditions. Further, the simulation and HIL results demonstrated that the voltage magnitude and THD have been maintained at 8.24% before and after the fault state. In addition, the suggested FT structure ensures the constant output power over the post-fault operation for both single and multiple switch failure instances while improving the MLI resilience. The feasibility and performance of the proposed method have been investigated through related case studies using simulation and hardware-in-the-loop (HIL) tests on a single-phase fifteen-level CHB MLI.