My SciELO
Services on Demand
Article
Indicators
- Cited by SciELO
Related links
- Similars in SciELO
Share
Revista Universidad y Sociedad
On-line version ISSN 2218-3620
Abstract
GUTIERREZ LABRADOR, Julio César; PEREZ ONES, Osney and ZUMALACARREGUI DE CARDENAS, Lourdes. Artificial neural network to estimate physical, thermodynamic and equilibrium properties of ethanol-water mixtures. Universidad y Sociedad [online]. 2021, vol.13, n.6, pp. 514-525. Epub Dec 10, 2021. ISSN 2218-3620.
The purpose of this work is to development neural networks for the prediction of physical, thermodynamic and equilibrium properties of the ethanol-water mixture, based on the compilation of existing data in the literature consulted. The database is cleaned and processed using the criterion of the maximum heat of each variable. The multi-layered perceptron with three layers and the backward feedback for the training are used as a network type, with functions of sigmoidal and hyperbolic activation. The mean squared error (MSE), the regression coefficient (R), the mean relative error (MRE) and the Friedman test, are the criteria used to determine the most appropriate network. The relative importance of the independent variables in the prediction is determined by the use of the weights’ method. A customized user interface is designed, which unifies the selected neural networks.
Keywords : Artificial neural networks; physical properties; thermodynamics properties; equilibrium properties; user interface.