Short-term Wind Power Forecasting Algorithm Based On Similar Time Period Clustering

Wind energy has strong randomness and intermittency.When large-scale wind power is connected to the power grid,the fluctuation of output power will reduce the reliability of the power system and may even undermine the safe and stable operation of the system.Accurate wind power prediction can make scheduling department develop a more reasonable power generation plan,thereby improving the performance of the power grid,which is of great significance.Wind power is generated by rotating the fan blades to generate electricity.However,due to the inertia of the blades,the output of the fan is not only related to the wind power at this moment but also to the wind force of the previous period.This paper first analyzes historical data,gives the definition of similar periods,and determines the optimal length of similar periods by experiments.Secondly,in order to improve the search speed of similar period,K-means algorithm is used to cluster similar periods.During the search,the cluster is divided by calculating the distance between the baselines of similar periods and cluster centers.Finally,the optimal similar period sets are screened out.On this basis,this paper proposes a short-term wind power forecasting method based on clustering of similar time period.Taking the optimal similar period as the training sample,the GA-Elman neural network model is established based on the power curve and the meteorological information,and the wind power of the future period is iteratively predicted.The example shows that this method can mine the valuable information contained in the data more effectively,and can improve the short-term wind power prediction performance.In addition,in order to further improve the accuracy of power prediction,this paper also attempts to use different strategies to establish a combined model,and analyze the application of each combination model through the example data.