Prediction of Biomethane Production of Cheese Whey by Using Artificial Neural Networks

The search for new energy sources has intensified today worldwide. In Argentina, bioenergy continues to be the energy with minor participation in the energy matrix. Therefore, research should focus on searching for new substrates that increase their production without compromising agricultural syste...

Full description

Saved in:
Bibliographic Details
Published in:Revista IEEE América Latina 2023-09, Vol.21 (9), p.1032-1039
Main Authors: Cordoba, Veronica Elizabeth, Mussi, Jorgelina, De Paula, Mariano, Acosta, Gerardo Gabriel
Format: Article
Language:English
Subjects:
Algorithms
Anaerobic digestion
Artificial neural network
Artificial neural networks
Back propagation networks
Biogas
Biological system modeling
Biomethane
Cheese
Cheese whey
Complex systems
Dairy products
Goodness of fit
Industrial wastes
Inoculum
Kinetics
Mathematical models
Methane
Modelling
Neural networks
Predictive models
Production
Regression coefficients
RNA
Statistical analysis
Statistical tests
Substrates
Whey
Citations: Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The search for new energy sources has intensified today worldwide. In Argentina, bioenergy continues to be the energy with minor participation in the energy matrix. Therefore, research should focus on searching for new substrates that increase their production without compromising agricultural systems. This work analyzed the cumulative methane production by anaerobic digestion of cheese whey, a food industry waste, using three substrate/inoculum ratios in volatile solids units. An artificial neural network (ANN) was developed to model the process based on the characterization data (input data) and methane production obtained in the laboratory (output data). The training algorithm used for the ANN was backpropagation. This model's validity was analyzed using statistical parameters such as the regression coefficient (R2) and the mean square error (MSE). The results obtained by the ANN were compared with those obtained by conventional kinetic models, such as the First-order and Gompertz models. The experimental biomethane production was in the range 2325 to 38222 NmLCH4/g VS. The proposed network could predict the experimental data with an R2 of 0.9992 and a training MSE of 2.7598. The statistics used to compare the goodness of fit between the models (R2 and RMSE) were higher for the network, demonstrating its ability to model a complex system such as anaerobic digestion.
ISSN:1548-0992
1548-0992
DOI:10.1109/TLA.2023.10251810