An Integral Sliding Mode Observer Based Fault Diagnosis Approach for Modular Multilevel Converter

Modular Multilevel Converters (MMCs) are distinguished by their modular nature that makes them suitable for a wide range of high-power and voltage applications. A practical MMC usually have a large number of power semiconductor devices hence the chance of failure of these components increases. Thus, for an MMC fault-detection and isolation (FDI) is very important. In order to improve the reliability of MMC, faulty sub-modules (SMs) should be rapidly isolated. In this paper, a fault-diagnosis approach based on integral sliding mode observer is presented that contains two phases: fault-detection and fault-localization. At first, the fault is detected when observed and measured states deviate from each other. While the later one pinpoints faulty SMs by comparing the error between average-voltage of all SMs and capacitor-voltage of one SM with a threshold value. The effectiveness and feasibility of the proposed method is validated by the simulation results.