Estimação da produção de biogás na codigestão anaeróbia de resíduos orgânicos da indústria sucroalcooleira utilizando algoritmos de aprendizado de máquina

Abstract

The development of reliable predictive models that estimate biogas production in industrial-scale facilities has been a challenge. The use of a Machine Learning model combined with analytical tools can support decision-making and improve the performance of biogas plants. In this context, the present work aimed to develop Machine Learning models to estimate the production of biogas, using daily data from analysis and sensors of a plant producing biogas from residues from the sugar and ethanol industry. The methodology involved the use of Random Forest, Artificial Neural Networks, Support Vector Machine, and Least Absolute Shrinkage And Selection Operator Algorithms, being adjusted and compared using the R-squared, Mean absolute error (MAE) and Mean absolute percentage error (MAPE) metrics. The results showed that the Random Forest model obtained the best results, with a final R-squared of 0,83 for the test set. It was also found that the main attributes for the estimation of biogas production were the amount of waste dosed (solid and liquid) and pH. Finally, a web application was also developed to perform the estimations using the best model generated.