This is an outdated version published on 2022-03-23. Read the most recent version.
Preprint has been submitted for publication in journal
Preprint / Version 1

Open-Circuit Submodule Fault Diagnosis in MMCs Using Support Vector Machines

##article.authors##

  • Ardavan Mohammadhassani Virginia Tech
  • Ali Mehrizi-Sani Virginia Tech

DOI:

https://doi.org/10.31224/2231

Abstract

Series connection of semiconductor submodules (SM) in a modular multilevel converter (MMC) makes the MMC prone to open-circuit (OC) IGBT failures inside SMs. If left undetected, these faults degrade the operation of the MMC and lead to its instability. This paper proposes a method to detect, localize, and classify single OC SM faults in an MMC using support vector machines (SVM) trained with data obtained from the capacitor voltage balancing block of the MMC control system. The proposed method relies on data extracted from the sorted capacitor voltage arrays of the upper and lower phase arms. Therefore, it does not require extra measurements and hardware. Additionally, it offers a fixed time for detecting and localizing OC SM faults. This method is easy to implement as SVM has a simple decision function. Time-domain simulation case studies are performed on a three-phase nine-level MMC to evaluate the performance of the proposed method.

Downloads

Download data is not yet available.

Downloads

Posted

2022-03-23

Versions