Offset-Free Model Predictive Control for the Power Control of Three-Phase AC/DC Converters

This paper describes an offset-free model predictive control (MPC) algorithm using a disturbance observer (DOB) to control the active/reactive powers of a three-phase AC/DC converter. The strategy of this paper is twofold. One is the use of DOB to remove the offset error, and the other is the proper choice of the weighting matrices of a cost index to provide fast error decay with small overshoot. The DOB is designed to estimate the unknown disturbances of the ac/dc converter following the standard Luenberger observer design procedure. The proposed MPC minimizes a one-step-ahead cost index penalizing the predicted tracking error by performing a simple membership test without any use of numerical methods. A systematic way for choosing the weights of the cost index, which guarantees the global stability of the closed-loop system, is proposed. Use of the DOB eliminates the offset tracking errors in the real implementation. Using a 25-kW ac/dc converter, it is experimentally shown that the proposed MPC enhances the power tracking performance while considerably reducing the mutual interference of the active/reactive powers as well as the output voltage.