An Approximate Dynamic Programming Approach for Wind Power Dispatch in Wind Farms

With appropriate control based on predicted data of renewable energy sources, battery energy storage systems (BESS) can increase the dispatchability of wind and solar power. In a deregulated electricity market, proper charge and discharge decisions of BESS made at different price periods are likely to increase the profit for renewable power plants. In this paper, we propose an energy trading strategy for wind farms with BESS to control the actual power sold to the electric grid. To maximize the income and minimize the battery lifetime consumption, an approximate dynamic programming algorithm is produced to solve the optimal charge and discharge actions of the BESS within a future period. The rolling horizon approach is utilized, and the algorithm works with both short-term and long-term predicted data that are updated in every control step. The use of an approximate technique is to avoid the long computation time resulted from a large-scale BESS, and to meet time constraints in the rolling horizon approach. The algorithm is tested with practical wind power and electricity price data.