This paper presents high performance improved direct power control (DPC) based on model reference adaptive control (MRAC) and neuro-fuzzy control (NFC) for grid connected doubly fed induction generator (DFIG), to overcome the drawbacks of conventional DPC which was based only on PID controllers, namely the speed/efficiency trade-off and divergence from peak power under fast variation of wind speed. A mathematical model of DFIG implemented in the d-q reference frame is achieved. Then, a direct power control algorithm for controlling rotor currents of DFIG is incorporated using PID controllers, and space-vector modulation (SVM) is used to determine a fixed switching frequency. The condition of the stator side power factor is controlled at unity level via MPPT strategy. The MRAC which is based on DPC is investigated instead of PID regulators. Also, the performances of NFC based on DPC are tested and compared to those achieved using MRAC controller. The results obtained in the Matlab/Simulink environment using robustness tests show that the NFC is efficient, has superior dynamic performance and is more robust during parameter variations.