An evaluation of conventional and computational intelligence methods for medium and long-term load forecasting in Algeria

Abstract

Electric load forecasting is one of the most important areas in electrical engineering, due to its main role for economic and reliable operation in power systems. In particular, accurate medium and long-term forecasts have significant effect on grid expansion planning and future generating capacity scheduling. This paper uses the Algerian electricity demand observations to evaluate methods for medium and long-term predictions. We consider methods designed to capture the trend and the seasonal cycle in the data as well as computational intelligence techniques. Among the variety of methods considered, satisfactory results were obtained by the adaptive neuro-fuzzy inference system and the autoregressive integrated moving average based approaches.