Model-Free Predictive Current Control of Doubly Fed Induction Generator Based on Ultra-Local Model

Abstract

Model predictive control (MPC) has been proposed as an effective method for power control of doubly fed induction generator (DFIG). However, conventional MPC relies heavily on the system model and accuracy of machine parameters. Furthermore, the steady state performance of conventional MPC is limited due to the application of only one voltage vector during one control period. To solve the problems of parameter robustness and steady state performance, this paper proposes a multiple-vector model-free predictive current control (MFPCC) based on ultral-local model. On one hand, an ultra-local model is used to replace the mathematical model of the machine. Since only the measured stator voltage and current values are required in the final control expression, the control system achieves good parameter robustness. On the other hand, by applying two voltage vectors during one control period, the steady-state performance is improved in terms of power ripples and current harmonics. The proposed method is compared to the prior multiple-vector MPC and the experimental results confirm its effectiveness.