Author(s): Martins, Gilberto ; Duarte, Maria Salomé Lira ; Oliveira, P. ; Dias, A. ; Marcondes, Francisco Supino ; Pereira, M. A. ; Novais, Paulo
Date: 2024
Persistent ID: https://hdl.handle.net/1822/93692
Origin: RepositóriUM - Universidade do Minho
Subject(s): Decision support system; Anaerobic digestion; Artificial intelligence
Anaerobic digestion processes are often implemented by wastewater treatment plants (WWTPs) to stabilize and decrease the organic content of primary and secondary sludges. Additionally, this process allows to increase the energetic efficiency of WWTPs. Data from a digestor of a WWTP from the North of Portugal was used to study and design two machine learning models, LSTM-based and GRU-based, to predict energy production from biogas generated through Anaerobic Digestion in the WWTP. By analyzing the results obtained, it was possible to conclude that the best candidate model, in terms of RMSE, was LSTM-based with an error value of 347.9 kWh. Based on these results, a new decision support system was carried out using the machine learning model (LSTM-based) to be used for the accurate prediction of biogas and electricity produced in the co-generator, and also the mechanistic anaerobic digestion model no.1 to be used to process understanding and trend prevision.
info:eu-repo/semantics/publishedVersion
