Doubly-fed wind power generator (DFIG) type is the mainstream model of the wind turbine at home and abroad at present. To study the impact on reliability and stability of power system with large-scale wind power grid, it must have accurate model and parameters of wind turbine. Therefore, this paper studies the model parameter identification method of doubly-fed wind power generator. First, in the Matlab/Simulink environment, it sets up a simulation model for the wind turbine grid-connected and gets the measured data. Afterwards, by using the generator ?? coordinate system (mathematical) model, it analyzes the identifiability to illustrate the convergence for the proposed model, and then through the analysis of the mathematical model, gets an expression of stator self-induction and the mutual inductance. Finally, the rotor initial phase angle , stator self inductance, mutual inductance, and mutual inductance between rotor and stator can be identified by using genetic algorithm with identification strategy step by step. It provides reliable theory for the study of large-scale wind power grid.