Abstract: In this paper,in view of the fact that the traditional aggregation method cannot solve the problem of parameters variation after long-term operation of wind farms,a detailed equivalent model of Doubly Fed Induction Generators（DFIGs）wind farm and initialization method are developed. The trajectory sensitivity of parameters is analyzed. Parameters identification strategy is proposed that the non-time-varying parameters are fixed as aggregated values,while the Genetic Learning Particle Swarm Optimization（GLPSO）hybrid algorithm is used to identify time-varying parameters based on Phasor Measurement Unit（PMU） data at the common interconnection point of wind farm. The robustness and adaptability of the equivalent model of DFIG wind farm under different wind speeds,wake effects,unknown wind speed,different short-circuit fault locations and voltage sags depth and some DFIGs off-line are analyzed. The simulation results using the Western Electricity Coordinating Council benchmark test system show that the global searching capability to find the optimal solution of the proposed method is much higher than that of canonical particle swarm optimization（PSO）and genetic algorithm（GA）. Further,the maximum deriation between the identification results using the proposed method and the true values is less than 10% with high sensitivity parameters,which is much better than previous state-of-art work.