Scaled Conjugate Back-Propagation Algorithm for Prediction of Phenol Adsorption Characteristics
Bishwarup Biswas1, Ayan Kumar Bhar2, Adwitiya Mullick3, Mahua Ghosh4, Monal Dutta5
1Bishwarup Biswas, Department of Chemical Engineering, Calcutta Institute of Technology, Howrah (West Bengal), India.
2Ayan Kumar Bhar, Department of Chemical Engineering, Calcutta Institute of Technology, Howrah (West Bengal), India.
3Adwitiya Mullick, Department of Chemical Engineering, Calcutta Institute of Technology, Howrah (West Bengal), India.
4Mahua Ghosh, Department of Chemical Engineering, Calcutta Institute of Technology, Howrah (West Bengal), India.
5Monal Dutta, Department of Chemical Engineering, Calcutta Institute of Technology, Howrah (West Bengal), India.
Manuscript received on 13 June 2017 | Revised Manuscript received on 20 June 2017 | Manuscript Published on 30 June 2017 | PP: 155-158 | Volume-6 Issue-5, June 2017 | Retrieval Number: E5029066517/17©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: In the present investigation the adsorption characteristics of phenol on the surface of chemically modified natural clay was predicted by using three-layer artificial neural network. The effect of various operational parameters on the adsorption process was determined by using scaled conjugate back-propagation algorithm. For this purpose, a feed forward network (5 – 11 – 1) with a learning rate of 0.02 was constructed. Various transfer functions such as, tangent sigmoid, saturated linear and positive linear were applied to hidden layer whereas pure linear transfer function was used in the output layer. The network performance was defined in terms of mean squared error (MSE) and validation error (VDE). The optimum number of neurons in the hidden layer was found to be 11 with “poselin” and “purelin” transfer functions in the hidden layer and output layer respectively. The MSE and VDE in this case were 2 × 10-5 and 5 × 10-5 Respectively
Keywords: Adsorption, ANN, MSE, VDE
Scope of the Article: Algorithm Engineering