Evaluating the impact of wind power probabilistic forecasting on very-short-term generation scheduling for Wind-Storage Combined Generation System

Incorporating Energy Storage System (ESS) with wind farms to build up Wind-Storage Combined Generation System is a promising solution to improve the dependability of wind power. Wind power forecasting precision plays a vital role to provide reliable generation scheduling on the combined generation system with smaller ESS. This type of generation scheduling (time interval: 15 minutes) makes wind generation a favorable player in power market trading. In this paper, wind power forecasting models with determined value and probabilistic interval are both established based on Radial Basis Function (RBF) neural network and nonparametric estimation method, respectively. According to the probabilistic model, the Plan Deviation (PD) index is proposed to estimate the forecasting impact to real-time system operation as well as adjusting generation plan in the scheduling time duration. After case study, the proposed model is proven to improve the reliability of generation scheduling along with increasing the participation of wind farm in power market trading.