Projecte fet en col.laboració amb l'Instituto Superior Tecnico. Universidade Técnica de Lisboa

Load forecasting is vitally important for the electric industry in the deregulated economy. It has many applications including energy purchasing and generation, load switching, contract evaluation and infrastructure development. Because of this, a large variety of mathematical methods have been developed for load forecasting. In addition, the large-scale integration of wind power, nowadays, represents a challenge for power system operations planning, because wind power cannot be dispatched in a classical sense. Moreover, its output varies as weather conditions change. This warrants the investigation of alternative short-term power system planning methods capable of better coping with the nature of wind generation. This paper focuses on carrying out an accurate short-term load forecasting method, in a mixed scenario with uncertain wind power generation (WPG) and load, improving forecasting errors obtained with different algorithms tested. It focuses only in regression models, although other methods have already been developed, such as systems of artificial intelligence.