Short-term wind power prediction based on multi-scale tuple matching

Wind power short-term prediction method generally depends on the meteorological data at present. This paper proposed time series power prediction method which is based on multi-scale tuple matching and can predict wind power well by making full use of historical data without affecting the computational efficiency to predict wind power on the occasion where power series can be obtained but the meteorological data not. This method used the multi-scale tuple matching technology to search target sequence quickly and accurately. Then calculate some characteristic parameters through the comparison analysis of the search results and use historical data and the characteristic parameters to predict future output. Simulation studies are carried to test the performance of this method using the data obtained from a wind farm in Northwest China. Results show that this method can predict power effectively. It is characterized by independence of meteorological information, cost saving, high prediction precision and strong real-time compared with traditional method.