Aplicação de técnicas de regressão linear, regressão do processo Gaussiano e de redes neurais artificiais para previsão do consumo de energia de uma agroindústria

Abstract

This work seeks to compare time series analysis techniques to predict the electricity consumption of an agroindustry, located in the southwest of the state of Goiás. The database presents a historical series of monthly energy consumption in the period between january/2016 and december/2021, totaling 72 observations. Forecast models, Linear Regression (LR), Gaussian Process Regression (GPR) and Multilayer Perception Neural Networks (MLP), provided by WEKA software, were used to predict energy consumption. Results, obtained from the three models, were compared using the Mean Absolute Percentage Error (MAPE). It was verified, for a horizon of four months, that the MLP model presented, in relation to the LR and GPR models, a better performance.