Metodologia para previsão de carga e geração no horizonte de curtíssimo prazo

Abstract

Load forecasting is a very important activity on electric power system operation and planning, because many studies on electricity sector depend on future behavior of the system, requiring the electricity demand forecast for its realization. The very short-term load forecasting has a horizon of few minutes to a few hours and it seeks to translate more accurately the instantaneous profile of load. There are several factors that should be considered in forecasting methods, climatic variables have a major influence on demand trends in the very short term, therefore, they should be incorporated into the projection model. In Brazil, has been growing use of electricity production through the photovoltaic generation, so, for this feature to be used efficiently, energy produced by the solar panels forecast is a tool that contributes to this type of energy act reliably. The main objective of this work is to develop a methodology for load and solar power generation forecasting in the very short-term considering the influence the climatic variables. The methodology for load, wind and solar power generation forecasting considers the climatic variables: temperature, relative humidity, wind speed, solar radiation and atmospheric pressure. The study presents data load for a typical year of a substation of the metropolitan region of Rio Grande do Sul, analyzed with data from a weather station in the region. For calculate the solar power generation forecasting the method uses a model that considers the solar radiation and the temperature to calculate the power produced by the photovoltaic module. The method for the forecast was performed using Excel VBA tool, by grouping the load and climate variables data of history and is based on multiple linear regression. The projection algorithm was tested and compared computationally, based on actual data, presenting significant results, because as it is projected to hours ahead, the data is updated with the actual data every hour, reducing forecast errors, confirming that the considered climatic variables are very important to refine load and generation forecasting methods, essential for system planning. Compared to other existing methods, the proposed method stands out by the fact to consider climatic variables for the projection, and uses the methodology to perform the projection of solar power generation.