As wind power industry develops rapidly, the control strategy for active power of wind farms tends to be regional. In order to regulate reasonably the active power of wind powercluster and increase the amount of wind power accommodation for reducing the number of wind turbines, a stratified active power control strategy for wind power cluster considering ordered wind turbines is proposed. Firstly, according to the location of wind farms and ultrashort term prediction of wind power, the wind power of regions are divided into three control layers, namely wind farm group layer, wind farm layer and wind turbine layer. The wind farm group layer and wind farm layer are optimized to increase the accommodation of wind power by rolling optimization at different time scales. The wind turbine level calculates and ranks the control capacity scores of each turbine by selecting the evaluation indexes that affect the control ability of the wind turbines, combining the entropy method and the membership function. In this way, the number of control times of the wind turbines is reduced. Finally, the data of actual wind farms in Shanxi power grid are used to analyze by GAMS and Matlab platform. The results show that the proposed control strategy reduces the number of wind turbine control times while improving the wind power accommodation.