https://doi.org/10.15255/KUI.2019.024
Published: Kem. Ind. 69 (3-4) (2020) 111–127
Paper reference number: KUI-24/2019
Paper type: Original scientific paper
Download paper:  PDF

A Comparison of “Neural Networks and Multiple Linear Regressions” Models to Describe the Rejection of Micropollutants by Membranes

Y. Ammi, L. Khaouane and S. Hanini

Abstract

A rejection process of organic compounds by nanofiltration and reverse osmosis membranes was modelled using the artificial neural networks. Three feed-forward neural networks based on quantitative structure-activity relationship (QSAR-NN models) characterised by a similar structure (twelve neurons for QSAR-NN₁, QSAR-NN₂, and QSAR-NN₃ in the input layer, one hidden layer and one neuron in the output layer), were constructed with the aim of predicting the rejection of organic compounds. A set of 1394 data points for QSAR-NN₁, 980 data points for QSAR-NN₂, and 436 data points for QSAR-NN₃ were used to construct the neural networks. Good agreements between the predicted and experimental rejections were obtained by QSAR-NN models (the correlation coefficient for the total dataset were 0.9191 for QSAR-NN₁, 0.9338 for QSAR-NN₂, and 0.9709 for QSAR-NN₃). Comparison between the feed-forward neural networks and multiple linear regressions based on quantitative structure-activity relationship “QSAR-MLR” revealed the superiority of the QSAR-NN models (the root mean squared errors for the total dataset for the QSAR-NN models were 10.6517 % for QSAR-NN₁, 9.1991 % for QSAR-NN₂, and 5.8869 % for QSAR-NN₃, and for QSAR-MLR models they were 20.1865 % for QSAR-MLR₁, 19.3815 % for QSAR-MLR₂, and 16.2062 % for QSAR-MLR₃).

This work is licensed under a Creative Commons Attribution 4.0 International License

Keywords

neural networks, multiple linear regressions, models, rejection, micropollutants, membranes