Keywords: artificial neural networks, ANN, multivariate nonlinear regression, MNLR, operational parameters, adsorption efficiency, modelling, Pb(II) adsorption, lead adsorption, industrial wastewater, ostrich bone char, wastewater treatment, Iran
Modelling Pb(II) adsorption based on synthetic and industrial wastewaters by ostrich bone char using artificial neural network and multivariate non–linear regression
In this study at first, the most suitable networks of ANN and MNLR models were constructed based on experimental datasets from a laboratory batch mode. Five ANN and MNLR models comprising various combinations of operational parameters were developed. The results showed that ANN5 model characterised by Levenberg–Marquardt learning algorithm and tangent activation function, which used all input parameters was the most accurate (MSE = 0.0002 and R² = 0.98). The results also showed that MSE and R² of MNLR5 that all operational factors are used were 0.18 and 0.89, respectively. Afterwards, 100 industrial wastewater samples obtained from different manufactories that are located in Isfahan province were selected, and treated by adsorption under optimal conditions. Finally, the removal percentage of Pb(II) ions estimates have been achieved by a suitable ANN model for the industrial wastewater samples. The findings reported in this paper indicated characterisation of flexibility and extensibility of ANN model.