Fault Detection and Classification in Five-Level Reduced Device Count Multilevel Inverter Using Fuzzy Logic System

The reliable performance of multilevel inverters (MLIs) is critical to the advancement of power electronic systems used in renewable energy conversion, microgrid operation, and electric vehicle technologies. However, these systems are often susceptible to open-circuit faults in power switches, which can adversely affect output waveform quality and overall system stability if not detected early. This study presents an intelligent fault detection and localization strategy based on a fuzzy inference system (FIS) for a reduced device count cross-connected-source MLI. The proposed diagnostic framework employs output voltage and current signals as diagnostic parameters, enabling precise identification of fault conditions. A comprehensive set of fuzzy rules is developed to differentiate between single and double switch faults under diverse fault inception angles. Real-time validation is carried out using the dSPACE DS1202 controller in a hardware-in-the-loop environment. The findings confirm that the proposed FIS-based approach achieves high fault classification accuracy, rapid response time, and strong adaptability across varying operating conditions, highlighting its suitability for real-time condition monitoring in modern power conversion systems.