Study of the predictive capability of modular multilevel converter simulation models under parametric and model form uncertainty

Abstract

The deviation of real system behavior from the predictions made from modeling and simulation is inevitable due to variations in different model input parameters as well as inaccurate modeling. The potential sources of uncertainty in modular multilevel converters (MMCs) is significant in medium- and high-voltage applications where each arm consists of several power electronics building blocks (PEBBs) connected in series. Therefore, assessing the predictive proficiency of the model, in the presence of various uncertainties, is critical in gaining confidence in modeling and simulation results. This paper investigates the predictive capability of MMC simulation models when adding more PEBBs. The relationship between the total uncertainty in modeling and simulation, and the number of PEBBs in each arm is presented for different model outputs. The results reveal an interesting feature of MMC — despite the fact that the number of potential sources of uncertainty increases by adding more PEBBs in each arm, the total uncertainty in the prediction of a system response quantity remains the same or decreases, depending on the selected model output response.