When we set about organizing EPIA 2003 in Porto during the APPIA meeting at the previous edition of the conference, EPIA 2001, it was decided that it would be organized by Fernando Moura Pires (Faj e) and myself. We chose Beja as the venue to host the conference, as it provided a good support infrastructure and Fernando had a good working relationship with several people at the Beja Polytechnic Institute. Shortly thereafter, Fernando came to know that he was ailing from a disease thatwastotakehislifeinMay2003.Aswithmanyotherprojectsinwhichhegot involved, Fernando clung to the organization of this conference with dedication and perseverance, even while knowing that he might not see the results of his work. EPIA 2003 is a tribute to his work. Taking up on the successful experience gained from EPIA 2001, we decided to structure EPIA 2003 as a set of ?ve distinct workshops, roughly re?ecting the panorama of AI research in Portugal. Special thanks are due to the organizers of each workshop, for the quality and timeliness of the work they carried out. The conference was all the more interesting because of the eight invited p- sentations and tutorials, by Alexander Bockmayr, Am ?lcar Cardoso, Dario F- reano, HaroldBoley, PedroDomingos, PieterAdriaans, VeronicaDahlandVitor Santos Costa. There are short one-page abstracts included in these proceedings for some of these presentations."

Forecasting electricity demand is a very important task for the agents of the Electricity Generation and Distribution Sectors. In this study we present a theoritical background about Brazil's Power Sector Conjuncture and also a review of Quantitative Methods used to estimate electricity consumption around the world. It is estimated a SARMAX model to the three main electricity consumption sectors with mensal data from January 2003 to September 2010.