Application of BP Neural Networks based on genetic simulated annealing algorithm for shortterm electricity price forecasting

摘要

BP Neural Network can forecast short-term electricity price, while it is necessary to explore technique to tune the back propagation learning algorithm either for better generalization, or for faster training. The paper proposed enhanced BP Neural Network to forecast electricity price, in which we replaced back propagation algorithm of BP Network with genetic simulated annealing algorithm (GSAA). It integrated GA's search performance and SA's strong local search performance, and has a better performance in terms of solution accuracy and convergence speed. Finally, a case study of New South Wales in Australia illustrates the feasibility and effectiveness of the proposed method