This paper presents recently developed Variable Gain Super Twisting Algorithm (VGSTA) for Direct Power Control (DPC) of a grid connected Doubly-Fed Induction Generator (DFIG) system to regulate stator active and reactive power directly in stationary reference frame. Grid-connected DFIG system exhibit highly non-linear behaviour due to random wind fluctuations, uncertainties in machine parameters and variation in grid voltage and frequency due to power system dynamics. These uncertainties produces large ripple in drive train torque which may lead to damage of mechanical components and also causes flicker in weak grid. Second-order Sliding Mode (SOSM) control strategy has an essence of robustness for highly nonlinear system having uncertainties and external perturbation. The proposed control scheme is designed using Lyapunov function. The efficacy of the proposed algorithm is demonstrated by comprehensive simulations on 5 kW DFIG system for optimum wind power extraction under the step variations of wind speed and reactive power. The results endow that the proposed algorithm effectively regulates both the stator active and reactive powers satisfactorily with chattering attenuation and also ensure faster dynamic response due to its adaptive nature of the variable gains. The robustness of the proposed control scheme has also been proved during disturbances and model parameter variations. The results are compared with First-order Sliding Mode (FOSM)-DPC scheme to corroborate the superior performance of the proposed strategy.