Short term electricity price forecasting using a new hybrid model based on two-layer decomposition technique and ensemble learning

Abstract

Research on forecasting electricity prices is of great significance to market participants. It is very difficult, however, to forecast the electricity price series because of its nonlinearity and high volatility. Considering that the existing studies directly ignore the important information contained in the residual term (Res.) after variational modal decomposition (VMD), this paper introduces a two-layer decomposition technique based on the combination of VMD technology and ensemble empirical modal decomposition (EEMD), carrying out EEMD decomposition on the residual term after VMD decomposition to improve the overall prediction accuracy of the model. At the same time, in order to address the defects of the existing hybrid model prediction methods—which use equal weights to reconstruct the prediction results—this paper draws on the idea of ensemble learning, combining the extreme learning machine (ELM) optimized by the differential evolution (DE) algorithm, introducing the DE-ELM meta-learner to optimize the reconstruction weights of the prediction components, and constructing a new hybrid model VMD-Res.-EEMD-DE-ELM-DE-ELM to obtain better prediction results. In order to verify the model's prediction performance, this paper uses electricity prices from Australian and Spanish electricity markets for empirical analysis. The results show that the hybrid model proposed in this paper has significant forecasting advantages.