Abstract: With high penetration of wind power, the automatic generation control (AGC) of distributed generation system has to deal with the problems caused by natural environment uncertainty. An improved AGC method is proposed, to reduce the negative impacts of uncertainty. Firstly, an AGC model for the regional power system with wind power participating in the frequency regulation is built. The virtual inertial control of wind turbine is introduced to respond to short-term output fluctuation. A predictive optimal control strategy with second-order proportional integral derivative controller is proposed. By establishing AGC model with PIDD² controller, the prediction algorithm is used to obtain the optimal prediction sequence of the AGC system which is used to adjust the reference signal for better AGC performance. The simulation result shows that the proposed method can effectively solve the problem that the traditional PID controller with fixed parameters is inadaptable to the dynamic change of the system with high penetration wind power.