Short-term Wind Speed Forecasting using Model Structure Selection and Manifold Algorithm

Abstract

Nonlinear model structure selection (MSS) is an important step in the modeling theory of the nonlinear system. The proper model structure and model parameters estimation can be used to reduce unnecessary computations and improve the model forecast accuracy. This paper mainly investigates the short-term wind speed forecasting (STWSF) by utilizing the comprehensive MSS technique based on real data of the wind speed plant in East China. The MSS and manifold algorithm are respectively used to design proper model structure and reduce the computational complexity to improve the forecasting accuracy and promote the computational efficiency. The experimental evaluation by using support vector regression based on the real data is given to demonstrate the performance of the proposed method.