Predição de consumo de energia elétrica no campus Medianeira com o uso de métodos estatísticos

Abstract

The consumption projection is a prediction of future conditions with the purpose of crossing information, variables and trends that help to understand the reasons that lead an installation to have a certain profile of energy consumption. A reliable consumption projection can be a tool to propose energy strategies in order to minimize excessive expenses, being directly linked to the revenue projection, helping to determine expectations and avoid losses. On university campuses, the energy use environment is very physically dispersed, with different people and usage characteristics, so that without an effective measurement system it is difficult to identify actions that can improve the efficiency of the use of this much needed resource, and at the same time at the same time, establish a relationship with the institutions and their substantial costs associated with the annual budget. Under these conditions, this work proposes the development of the project of an electric energy consumption curve of the UTFPR Medianeira campus with 75% or more in the degree of prediction. Consumption data from the years 2014 to 2018, their respective monthly average temperatures and number of hours of use of all classrooms were used. The data were separated into two groups, with the years 2014 to 2017 being used to create a regression model, which was then tested on the data for the year 2018. From this, a model was arrived at that was able to predict consumption university students with an average absolute error of 18% during the months of the test year, as well as an error of –1.5% for the total annual consumption.