Granular load forecast by clustering techniques
Daniel Sica, L.H. de Macedo, Jacques Szczupak, Leontina M. V. G. Pinto, Robinson Semolini, Marcia Inoue · International Conference on Circuits · 2010
This paper addresses the granular load forecast problem - where it is necessary to produce consistent and coordinated forecasts for a wide range of small, localized loads. We focus the feeder load forecast, comprising near a thousand of electrical points which, in turn, are composed by up to eight different consumer's classes. We propose a novel solution based on a combination between Hilbert Spaces and Cluster Techniques, able to combine speed, simplicity, consistency and coherence. The potentiality of the described model may be observed through a case study with a real Brazilian utility.