Electricity demand prediction by multi-agent system with history-based weighting

Energy and load demand forecasting in short-horizons, over an interval ranging from one hour to one week, is crucial for on-line scheduling and security functions of power system. Many load forecasting methods have been developed in recent years which are usually complex solutions with many adjustable parameters. Best-matching models and their relevant parameters have to be determined in a search procedure. We propose a hybrid prediction model, where best exemplars from a possibly large set of different simple short-time load forecasting models are automatically selected based on their past performance by a multi-agent system with history-based weighting. The increase of prediction accuracy has been verified on real load data from the Polish power system.