Mass minimization and sensitivity analysis of high power modular multilevel converter

Abstract

Optimal design and selection of passive components has been known as a challenging issue in the field of MMC converter especially in high power applications. Global optimization is an integrated optimization loop consisted of several sub-models such as circuit model, electromagnetic and thermal model, constraints, goal function and nonlinear solver. Global optimization algorithm is utilized to minimize the total converter volume regarding to the technical and manufacturing constraints. The most important factor to achieve the reliable optimization results is to employ the precise models. In this paper, several optimization algorithm with different level of complexity are proposed and developed to minimize the total MMC volume. The first model is consisted of a time-domain steady-state circuit model that determines the circuit value of passive components. Unlike the conventional model, the proposed circuit model does not neglect the switching frequency and saturation effect of arm inductance. The second model employs the circuit model in combination with the dimension and thermal model of arm inductance. Finally, a hybrid optimization algorithm is proposed which is consisted of an internal correction loop using finite element method to enhance the model accuracy. Also, a comprehensive sensitivity analysis has done to evaluate the total converter mass sensitivity against the different converter parameters and constraints.