A Very Short-term Forecasting Approach for Virtual Power Plant Using a Self-adaptive Hybrid Algorithm

Virtual power plant (VPP) emerges as a new concept to promote the efficient utilization of renewable energy resources (RES) and flexible loads. RES forecasting approach is an important item to ensure the stability and economy in VPP dispatching and bidding. In this paper, considering multienvironmental factors, a very short-term photovoltaic (PV) forecasting approach based on a self-adaptive simulated annealing hybrid genetic algorithm (SA-GA) and backpropagation neural network (BP) algorithm is proposed. Numerical studies illustrate that the proposed approach achieves a satisfactory forecasting accuracy and offers a high computation efficiency, which indicates its promising application value in RES forecasting for VPP.