Abstract
Integration of wind energy into electric power systems requires accurate and reliable forecasting of wind generation in the short-term horizon. This paper proposes a short-term wind power forecasting approach based on a LNF model and the HS optimization algorithm. The LNF model is empowered by LSSVMs, which are capable of dealing with complex and nonlinear time series (e.g. wing generation series), and is trained by a HSPS. In the HSPS, a harmony search optimization algorithm is used to find the optimal split direction during the partitioning procedure. Moreover, it selects the most relevant input variables with the least redundancy, and a MI input selection technique is utilized. To validate the proposed approach, wind generation data in the Irish grid for 2013 are used. To present a detailed investigation, the last weeks in all 12 months in 2013 are forecasted and the results are compared to several commonly used forecasting techniques. The numerical results indicate the promising performance of the LNF model with HS optimization for wind power forecasting applications.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call