Study of Dynamic Adsorption from Wastewater into Mineral Activated Carbon Using Artificial Neural Network | Chapter 9 | Current Approaches in Engineering Research and Technology Vol. 10

 

Water bodies are the most common natural resources that have been contaminated as a result of different human activities. Actually, in industries various methods of treatment of wastewater and many adsorbent materials used for purification of effluents have existed. An artificial neural network (ANN) was used in this study to predict the removal of sodium decanesulfonate using actived carbon obtained by the calcination of mineral biomass under different conditions. The structure of [3-3-1] was obtained and given a good correlation coefficient (R2 = 0.9965) with root mean squared error (RMSE = 0.0276). For the stage of interpolation and extrapolation, the results present a high correlation coefficient close to 1 which provides the robustness and the high capacity of ANN developed model.

 

Author(s)details:-

 

Sediri Meriem
Biomaterials and Transport Phenomena Laboratory (LBMPT), University of Médéa, 26000, Médéa, Algeria.

 

Hanini Salah
Biomaterials and Transport Phenomena Laboratory (LBMPT), University of Médéa, 26000, Médéa, Algeria.

 

Please See the book here :-  https://doi.org/10.9734/bpi/caert/v10/2974

Study on Dynamic Adsorption of Complex System In Solid-Liquid Phase Modelling Using Artificial Neural Networks | Chapter 2 | New Ideas Concerning Science and Technology Vol. 6

 The aim of this project is to create an ANN model that can predict the dynamic adsorption of a complex system of adsorbent-adsorbate in the solid-liquid phase based on various parameters using an adsorption column. In the input layer, nine neurons were used, while the first and second hidden layers used fourteen and ten neurons, respectively. In the output layer, only one neuron was used. The neural network was tested using a set of 2007 data points. For the hidden and output layers, the Levenberg Marquardt learning (LM) algorithm, logarithmic sigmoid transfer function, and linear transfer function were used, respectively. The ANN generated R2 = 0.9976 and 0.9969 for the total database and validation phase, respectively, between simulated data and those derived from the literature, with a root mean square error RMSE = 0.0268 and 0.0305 for the total database and validation phase, respectively.

Furthermore, Thomas and Bohart-Adams models were used to determine the best model. The neural network model outperformed the root mean square error (RMSE), sum of absolute error (SAE), Chi-square statistic test (X2), and correlation coefficient (R2) models. In general, the established model produces the highest [R2,,] agreement vector values with a root mean square error (RMSE) that is close to zero.

Author (s) Details

Meriem Sediri
Biomaterials and Transport Phenomena Laboratory (LBMPT), University of Médéa, Ain D’Heb 26000, Médéa, Algeria.

Salah Hanini
Biomaterials and Transport Phenomena Laboratory (LBMPT), University of Médéa, Ain D’Heb 26000, Médéa, Algeria.

Maamar Laidi
Biomaterials and Transport Phenomena Laboratory (LBMPT), University of Médéa, Ain D’Heb 26000, Médéa, Algeria.

Hakima Cherifi
Biomaterials and Transport Phenomena Laboratory (LBMPT), University of Médéa, Ain D’Heb 26000, Médéa, Algeria.

Siham Abbas Turki
Department of Electrics and Computing Engineering, University of Médéa, Ain D’Heb 26000, Médéa, Algeria.

View Book :- https://stm.bookpi.org/NICST-V6/article/view/168