Abstract: To improve the maximum power tracking efficiency of permanent magnet synchronous wind generator under variable working conditions, a model self-correcting composite active disturbance rejection control（MSCADRC） based on the optimal blade tip speed ratio method was designed to track the motor speed in real time. An extended state observer（ESO）, which could switch freely between linear and nonlinear, was used for real-time observation and estimation of the total disturbance caused by parameter perturbation or other uncertainties. A real-time predictive inertia observer of torque and moment was used to compensate the influence of parameter disturbance in the controller on ADRC. In the current loop ADRC control, q-axis current compensation factor was added to eliminate the interference of d-axis current, and the feedback compensation control method based on electromagnetic torque observer was adopted to realize the real-time compensation of q-axis current for accelerating the system response speed. The mathematical model of wind speed, wind machine and permanent magnet synchronous generator was used to design the wind power model self-correcting maximum power tracking system, and the Matlab/Simulink was used for simulation analysis. The simulation results show that the model MSCADRC has superiority. The wind power system under self-correcting model can perform better on power tracking control when wind speed changes sharply, and the system response speed and anti-external interference ability can be further improved.