A Novel Sensor-Reduction Condition Monitoring Approach for MMC Submodule IGBTs Based on Statistics of Inferred On-state Voltage

Condition monitoring of modular multilevel converters (MMCs) submodule insulated gate bipolar transistors (IGBTs) is of significance to ensure reliable operation. The on-state voltage of IGBT is a promising indicator to perform condition monitoring while its application in MMC is tricky due to high blocking voltage and numerous submodules. In this paper, a novel sensor-reduction approach is presented to infer the on-state voltages of MMC submodule IGBTs. Such an estimation method of the on-state voltages is based on Kirchhoff’s voltage law applied to an MMC arm. This method focuses on a statistical analysis of capacitor voltages and arm voltages. The switching-state combinations are specifically selected for the purpose. A case study is conducted on a single-phase seven-level MMC test bench to verify the proposed on-state voltage estimation method, which only shows an error of less than 3%. Furthermore, continuous on-state voltage estimations are made with this MMC test bench under steady-state operations and carry out statistical analysis of the inferred on-state voltage at various deterioration levels. Finally, a confusion matrix is proposed, whose maximum accuracy reaches 92.4% for the early-deterioration condition.