The operating state of the state estimation modeling network is identified by the fusion of multi-source real-time data such as SCADA, PMU and security and stability control system. When the error of branch power flow or node voltage at the boundary is large, the active power of wind farm is determined by comprehensively considering the actual measurement collected by the wind farm centralized control system and the prediction information of the wind power prediction system. The wind power generators are grouped according to the topology in the wind farm, then each group is subdivided according to static characteristic and operating state in turn. The dynamic models and static parameters of equivalent wind power generators and equivalent transformers are calculated. The proposed method is proved to be fast and effective through the analysis of a practical power grid example.