This paper proposes a structure for long-term energy demand forecasting. The\r\nproposed hybrid approach, called HPLLNF, uses the local linear neuro-fuzzy (LLNF)\r\nmodel as the forecaster and utilizes the Hodrickââ?¬â??Prescott (HP) filter for extraction of the\r\ntrend and cyclic components of the energy demand series. Besides, the sophisticated\r\ntechnique of mutual information (MI) is employed to select the most relevant input features\r\nwith least possible redundancies for the forecast model. Each generated component by the\r\nHP filter is then modeled through an LLNF model. Starting from an optimal least square\r\nestimation, the local linear model tree (LOLIMOT) learning algorithm increases the\r\ncomplexity of the LLNF model as long as its performance is improved. The proposed\r\nHPLLNF model with MI-based input selection is applied to the problem of long-term\r\nenergy forecasting in three different case studies, including forecasting of the gasoline,\r\ncrude oil and natural gas demand over the next 12 months. The obtained forecasting results\r\nreveal the noteworthy performance of the proposed approach for long-term energy demand\r\nforecasting applications.
Loading....