Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
690203 | Journal of the Taiwan Institute of Chemical Engineers | 2016 | 8 Pages |
•This study presents a model for predicting the refractive index of binary ionic liquid system containing alcohol using the artificial neural network (ANN) algorithm.•The 6-6-9-1 neural network architecture was found to be the best network using two hidden layers Diffusion coefficient.•A total of 752 data points were used to train the network and to obtain the optimum neural network parameters.•The obtained correlation satisfactorily represents the experimental refractive index data tested.
In recent years, a new class of solvent called ionic liquids had successfully demonstrated potential applications in industrial chemistry and chemical technology due to its desirable properties. To this end, understanding their physico-chemical properties is of high importance. The current study presents a model for predicting the refractive index of binary ionic liquid system containing alcohol (methanol or ethanol or 1-propanol) using the artificial neural network (ANN) algorithm. The refractive index data were correlated as function of temperature, mole fraction, number of carbon atoms in the cation, number of atoms in the anion, number of hydrogen atoms in the anion and number of carbon atoms in the alcohol. Refractive index data from ThermoIL Database were used. Using ANN, a total of 752 data points were used in the calculation and to obtain the optimum neural network parameters. The 6-6-9-1 neural network architecture was found to be the best network using two hidden layers as shown by mean absolute error of 0.00783 and an overall average percentage error of 0.55%. The obtained correlation satisfactorily represents the experimental refractive index data and can be reliably used to predict the refractive index of other binary systems containing the considered cation and anions and the studied alcohols.
Graphical abstractArtificial neural network (ANN) algorithm with two hidden layers. The architecture consists of one input layer followed by hidden layer/s and an output layer being the last layer.Figure optionsDownload full-size imageDownload as PowerPoint slide