A Novel Framework of Extended Active Power and Its Application Into Low Complexity MPC for Enhanced Steady-State Performance

Abstract

This article proposes a novel low complexity model predictive control (LC-MPC) for the performance improvement of the ac/dc converter under unbalanced grid voltages. Although the finite control set model predictive control (FCS-MPC) is a very effective control method, whose cost function is evaluated for all discrete switching states, presenting a high calculation burden for the efficient control. To obtain a reduced harmonic of the grid current and a low calculation burden under unbalanced grid conditions, a new framework of the extended active power is proposed and a negative conjugate of new complex power is chosen as a control variable. In the proposed LC-MPC method, the best voltage vector is obtained with only one prediction whether it is the proposed single-vector-based LC-MPC (SVLC-MPC) or double-vector-based LC-MPC (DVLC-MPC) method, and the optimal duration time using the framework of the extended active power is derived in detail based on the principle of minimizing the error of the new complex power. The proposed LC-MPC has a similar dynamic performance with the prior model predictive control (MPC) methods and obtains much better steady-state performance, i.e., the reduced harmonics of grid currents, under severely unbalanced grid voltages. The comparative experiments confirm the effectiveness of the proposed LC-MPC method.