Survey of switch fault diagnosis for modular multilevel converter

This study presents a survey on the existing fault diagnosis methods (FDMs) of the switch devices for the rapidly developing modular multilevel converters (MMCs). Three categories, namely mechanism-based, signal processing-based and artificial intelligence-based FDMs, are evaluated and summarised depending on the operating principles. Mechanism-based FDMs detect the faults by comparing the inner characteristics of MMC or their derived parameters with the expected values. Signal processing-based FDMs detect the faults via comparing the processed output voltage or current with their expected values. Artificial intelligence-based FDMs detect the faults in the way of employing a trained intelligent classifier. Methods belonging to each category are introduced in detail via comparing a lot of criteria of the FDMs. Then, a figure-of-merit is defined to evaluate various FDMs. Finally, the summary is given and the developing tendency is recommended for future work.