An Approach to Control Multilevel Flying-Capacitor Converters Using Optimal Dynamic Programming Benchmark

The problem of balancing capacitor voltages is of utmost importance in multilevel converter topologies involving flying capacitors. In this study, a new minimum angular distance (MAD) algorithm is proposed to control the turning on and off of the switches, ensuring fast convergence of the capacitor voltages balancing problem in multilevel flying-capacitor converters. This algorithm was developed based on a preliminary analytical analysis of the capacitor voltage trajectories using the power-oriented model of the converter. Compared to other approaches, the proposed algorithm involves only simple and well-defined calculations, requires no training, and does not require any prediction of future values that the output current assumes. The proposed algorithm, implemented in the MATLAB/Simulink environment, is proven to give very good performance, verified against an optimal benchmark given by dynamic programming, in terms of capacitor voltages convergence time, efficiency, power loss, and total harmonic distortion.