Forecasting Electricity Consumption with Intelligent Hybrid Model

In this paper, an Extreme Learning Distributed Adaptive Neuro-Fuzzy Architecture (ELDANFA) model has been presented. It has been tested with real data for predicting energy consumption at an electrical substation in the South-Central region, near Plovdiv, Bulgaria. The main goal of this hybrid intelligent structure is to reduce the computational burdens of neuro-fuzzy models and to keep prediction error to a minimum. The obtained results prove that the proposed model predicts accurately electricity consumption. It is also suitable for real-time applications due to the reduced number of fuzzy rules and the small number of parameters updated during the learning procedure.