Modeling of vanillin adsorption from aqueous solution using resin H103 by artificial neural network
Vanillin adsorption onto resin H103 was modelled using artificial neural network (ANN) approach and the best ANN algorithm was determined in this work. The first step of ANN modeling was ANN set up, followed by the optimization of ANN. The parameters for the input layers are contact time, initial va...
Saved in:
Main Authors: | , , , , , |
---|---|
Format: | Conference or Workshop Item |
Language: | English |
Published: |
IOP Publishing Ltd
2019
|
Subjects: | |
Online Access: | http://umpir.ump.edu.my/id/eprint/35839/1/Modeling%20of%20vanillin%20adsorption%20from%20aqueous%20solution%20using%20resin%20H103%20by%20artificial%20neural%20network.pdf http://umpir.ump.edu.my/id/eprint/35839/ https://doi.org/10.1088/1757-899X/702/1/012048 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | Vanillin adsorption onto resin H103 was modelled using artificial neural network (ANN) approach and the best ANN algorithm was determined in this work. The first step of ANN modeling was ANN set up, followed by the optimization of ANN. The parameters for the input layers are contact time, initial vanillin concentration, resin dosage, pH, and temperature while the response is residual vanillin concentration. The neural network was trained using backpropagation (BP) algorithm. The result shows that the Levenberg-Marquardt algorithm was best suited the training function and the optimized ANN involved seven neurons at the hidden layer. This model can produce a correlation of determination value of 0.9999 with the mean square error (MSE) value of 0.0277. The best adsorption efficiencies for each factor were 98.11%, 96.03%, 98.14%, 98.2%, and 98.10% at 2.0 g of adsorbent dosage, 30 min of contact time, 100 mg/L of initial vanillin concentration, pH 5, and 25 °C, respectively. The outcomes of this work proved that ANN is excellent in predicting experimental data of vanillin adsorption by resin H103. |
---|