Modeling of the electric perturbations in urban lighting networks by Machine Learning techniques

Abstract:

The project dealt with the elaboration of a mathematical algorithm for prediction of the perturbation generated in the electric currents in urban lighting networks by regulation devices.  These regulators are used to diminish the light intensity at late night hours. However, the theoretical savings in electric energy are limited by the electric perturbations produced by the regulators, that generate reactive power at high levels.

The purpose was to model the reactive power generated in terms of the characteristics of the electric distribution panels. Several Machine learning techniques were used to approximate the electric perturbation, determining the most appropriate in terms of accuracy. The real measurement data were provided by the company IMESAPI.
The prediction algorithm was delivered to IMESAPI.