Method for performing short-term prediction on active power of wind power plant in combination mode

The invention discloses a method for performing short-term prediction on active power of a wind power plant in a combination mode, and relates to the technical field of power prediction. The method specifically comprises the following steps: preprocessing original power and wind speed data used by the method; performing fan clustering by using a density peak value clustering algorithm according to the sorted power sequence; and performing short-term prediction on each cluster sorted by the clustering algorithm by using the optimized gated recurrent neural network to obtain total short-term prediction power. The method is mainly used for short-term power prediction of the wind power plant with complex wind speed change, complex terrain and a large amount of historical data. 本发明公开了一种组合方式下对风电场有功功率进行短期预测的方法，涉及功率预测技术领域。本发明具体方法流程包括：将方法所用到的原始功率、风速数据进行预处理；根据整理好的功率序列运用密度峰值聚类算法进行风机分簇；采用优化后的门控循环神经网络对聚类算法整理出来的每簇分别进行短期预测，得到总的短期预测功率。本发明主要可用于风速变化复杂、地形复杂、具有大量历史数据的风电场进行短期功率预测。