Resumo

SANCHEZ, Roberto  e  BARRERA, Patricio. Short Term Demand Forecasting methodology for Power Decision Making Based on Markov Chain. Study Case - EEQ. Revista Técnica energía [online]. 2018, vol.15, n.1, pp.44-50. ISSN 2602-8492.  https://doi.org/10.37116/revistaenergia.v15.n1.2018.322.

This investigation is focused on the prediction of the electrical demand in short time. For this purpose, the “demand profiles” and the real time signal of the electrical demand of the Empresa Eléctrica Quito S.A. are used in order to determine which profile is expected to happen during the day. In this sense, this study uses the Hidden Markov Model for forecasting the electrical demand in short time. This approach first applies a learning/training process using data from the Sistema de Información Validada Operativa (SIVO). Later, a discovery process of demand profiles is performed in order to determine the most expected profile to happen during the day. This approach establishes an “expected demand area” that shall be a reference for the definitive behavior of the electrical demand. This methodology was applied over the EEQ system and evaluated during 30 days. The final tool successes 86% of the cases and the actual value of the electrical demand in real time is inside of the band of the expected demand area.The purpose of this work is to build an application that assist operators of the National Interconnected System, NIS, to make the decisions in short time, optimizing the resources for generation.

Palavras-chave : prediction of electrical demand; machine learning; hidden markov model; artificial intelligence.

        · resumo em Espanhol     · texto em Espanhol     · Espanhol ( pdf )