Computationally Efficient Long Horizon Model Predictive Direct Current ‎Control of DFIG Wind Turbines ‎

Abstract

Model predictive control (MPC) based methods are gaining more and more attention in power converters and electrical drives. Nevertheless, high computational burden of MPC is an obstacle for its application, especially when the prediction horizon increases extends. At the same time, increasing the prediction horizon leads to a superior response. In this paper, a long horizon MPC is proposed to control the power converter employed in the rotor side of DFIG. The main contribution of this paper is to propose a new comparative algorithm to speed up the optimization of the objective function. The proposed algorithm prevents examining all inputs in each prediction step to saving the computational time. Additionally, the proposed method along with the use of an incremental algorithm applies a sequence of weighting factors in the cost function over the prediction horizon to maximize the impact of primary samples on the optimal vector selection. Therefore, the proposed MPC strategy can predict a longer horizon with relatively low computational burden. Finally, results show that the proposed controller has the fastest dynamic response with lower overshoots compared to direct torque control and vector control method. In addition, the proposed strategy with more accurate response reduces the calculation time by up to 48% compared to classical MPC, for the prediction horizon of three.

Keywords

• model predictive control
• computational effort
• doubly fed induction generator, wind energy conversion system.‎