Wind power generation systems encounter numerous disturbances and uncertainties, which make the stable control of these systems a focal point of research. Model predictive control (MPC), as an optimal control algorithm, predicts the future behavior of a system model and selects the most suitable control behavior based on the control objective. To achieve more stable control of doubly-fed induction generators (DFIGs) under the influence of various uncertain environmental factors, the rotor-side control strategy of DFIGs is designed. Firstly, this paper applied the vector control principle with model predictive control theory to the super local model of DFIGs. And then, the extended state observer is designed to estimate the disturbance item of this super local model. Based on this model, the d- and q- axis rotor current and output power values at future moments are predicted. According to the value function of the output power, the optimal three voltage vectors are obtained and applied to the rotor-side converter to control the generator to output the desired power value. Finally, the performance of the proposed control strategy was tested through simulation experiments. Compared with the traditional vector control strategy, although the execution time is longer, it can effectively reduce the steady-state error. In the fault experiments of three-phase or two-phase grounding short circuits, the strategy still provides the system with good dynamic performance, demonstrating strong robustness.