A Statistical Submodule Open-Circuit Failure Diagnosis Method for Modular Multilevel Converters (MMCs) With Variance Measurement
Heya Yang,Weihao Zhou,Jing Sheng,Haoze Luo,Chushan Li,Wuhua Li,Xiangning He
DOI: https://doi.org/10.1109/OJPEL.2020.2992754
2020-01-01
IEEE Open Journal of Power Electronics
Abstract:Submodule (SM) open-circuit failures severely affect the reliable operation of modular multilevel converter (MMCs). A comprehensive SM open-circuit failure diagnosis process includes three steps with failure detection, localization, and classification. However, the existing diagnosis methods perform these steps separately, which makes the diagnosis processcomplicated. Furthermore, the efficiency is relatively low with separate diagnosis steps. To address this issue, a diagnosis method with variance measurement is proposed in this paper to combine failure detection, localization, and classification. The proposed method makes use of two fault characteristics of SM open-circuit failures: 1) SM capacitor voltages have distinct distributions upon different faulty IGBTs; 2) Numerous SMs in the MMC make the failure diagnosis equivalent to statistical outlier analysis. On account of the fault characteristics, a variance-based index called VAR is proposed to evaluate the distributions of SM capacitor voltages. The VAR values are then analyzed with quartile analysis, which is a statistical outlier analysis method, to complete the diagnosis process. The proposed failure diagnosis method features high reliability, high compatibility, and high cost-effectiveness. Finally, the feasibility of the proposed method is verified through both simulations and experiments under various failure scenarios.