Artificial Neural Networks for Modeling of Adsorption

Artificial Neural Networks are applied to the literature data pertaining to adsorption batch studies to develop and validate a model that can predict the Pollutant Removal Efficiency (PRE). Back-Propagation Network (BPN) is used with one hidden layer. The BPN model is systematically trained with 440 data points and is validated with 73 data points from the database. Three different combinations of learning parameters for number of neurons in hidden layer, learning rate, number of epochs, and error tolerance are attempted. Standard Deviation based on the test data is calculated to validate the accuracy of the output. The optimum values of learning parameters that are giving encouraging results for the present problem are identified using Elimination Technique.