Abstract: Aiming at the problems of low prediction accuracy of single wind speed prediction method and large error of tracking power dispatching instruction when power distribution of wind farm units is carried out according to wind speed ratio, a combined wind speed prediction method is proposed. Based on the unit classification of the predicted wind speed, current wind speed and output power, a method of active power distribution for wind farms is proposed. The fuzzy C-means classification method is used to classify the units. The power regulation priority is determined according to the classification results. The power instructions are allocated to different types of units, and then allocated to each unit according to the proportion of the output power of each unit of a certain class, so as to realize the active power distribution of the whole wind farm. Wind speed prediction and active power distribution simulation research are carried out based on actual wind speed data of a wind farm. The simulation results show that the combined wind speed prediction method and active power distribution method proposed in this paper have the advantages of high wind speed prediction accuracy, high output power tracking accuracy and small number of units involved in active power regulation.