Artificial neural network model to predict thermodynamic properties of low molar mass protic ionic liquid
- Author Miguel Iglesias
- Co-Author Rebecca S. Andrade, Robson Magalhaes,
- DOI
- Country : Brasil
- Subject : Chemical Engineering
Keywords : Artificial neural networks; protic ionic liquid; temperature; density; ultrasonic velocity
Abstract
This paper presents a model based on artificial neural networks (ANNs) to predict density and ultrasonic velocity of short aliphatic chain protic ionic liquids. An experimental database was used for developing the model, where the input variables in the network were temperature, number of carbon, hydrogen, nitrogen and oxygen atoms into each compound, as well as, the number of specific functional groups. The learning task was done through a nonlinear activation function of sigmoid and hyperbolic tangent nature. Correlation coefficients of 0.9783–0.9830 and mean squared error (MSE) of 1.2328·10-4 and 2.5783·10-6 were obtained for density and ultrasonic velocity, respectively, which suggests that the proposed ANNs model shows robust and accurate character for prediction of physical properties of these new promising chemicals.
Comments
No have any comment !