A Meta-heuristic Learning Approach for Short-term Price Forecasting

Abstract

Energy management is essential for both economy and environment, and proper management needs an efficient forecasting tool. Energy forecasting plays a significant role in balancing the generator-distributor loads. In this paper, a hybrid model has been proposed for accurate and effective electricity price forecasting. To improve the forecasting results of vanilla artificial neural network (ANN), it has been combined with different meta-heuristic algorithms to train its network parameters. The meta-heuristic algorithms evaluated here include follow the leader (FTL), cuckoo search optimization (COA), fruit fly optimization (FFOA), and particle swarm optimization (PSO). The performance of the hybrid models is evaluated on the New Pool, England dataset to predict the short-term electricity price. Results show that the FTL-ANN algorithm outperforms the other algorithms, including traditional ANN architecture.