Detection and diagnosis of sub-module faults for modular multilevel converters

This paper presents a new fault detection technique for the diagnosis and localization of sub-module faults that are developed due to failures in switching devices inside Modular Multilevel Converters (MMC). Unlike other fault detection strategies that have been proposed in literature, the proposed fault detection technique does not need extra sensors or special power circuits as it depends on an ADAptive LINEar Neuron (ADALINE) capacitor voltage estimation algorithm. The proposed fault detection technique is validated by Hardware In the Loop (HIL) real time simulations through which different case studies are simulated to test the algorithm under different operating conditions.