Forecast of steady-state voltage problems considering simulation and socio-environmental information

Abstract

This article presents the general panorama of the use of multiple sources of information to identify consumers who could receive voltage quality different from the appropriate. It is observed that only simulations of power flow, and consequently levels of voltage, are not sufficient for a good prediction of regions and consumers out of the appropriate levels of steady-state voltage. To work around this problem, adjustments and reforms are made to the distribution networks databases, improving the assertiveness of the simulations. However, the simulations still distance themselves from the ideal model of prediction. Using spatial statistics and machine learning techniques, together with data on voltage measurements of the ANEEL product quality campaign and socio-environmental information, there is a significantly higher prediction than the one obtained only by power flow simulations. The predictive results from each model is presented and compared.