Short Term Forecasting for Wind Power Based on Cluster Analysis

Short Term Forecasting for Wind Power Based  on Cluster Analysis

Yang Gao, Jing Xing, Aoran Xu, Liu Zhang, Gang Wang, Quanping Zou


Electric power college, Shenyang Institute of Engineering, Pu Chang Str.18 of Shenbei New Area. Shenyang City, Liaoning Province, China

In order to make full use of historical wind speed information behind the data, according to daily similarity of the wind speed and wind power, short term power forecasting method based on cluster analysis is presented in this paper. Through the original sample data is preprocessed, election history daily data that is similar with characteristic parameters of NWP of forecast day,  so as to establish training samples of model. NWP information of forecast day provided by Meteorological Department will be as the characteristic parameters of forecast day, and calculating Euclidean distance between characteristic parameters will be regarded as a basis of similarity measure. Finally forecasting model is founded by adopting similar samples based on cluster. Using NWP data as input parameters, the actual wind power as a target value, many kinds of short-term wind power forecasting model is gained by training. Through the actual wind farm test, forecasting accuracy is improved obviously.