University Of Tehran Press Pollution 2383-451X 10 1 2024 01 01 Application of Artificial Neural Network and Multiple Linear Regression for Modelling Adsorptive Removal of Pb (II) ions over Cedrus deodara Bark Powder 528 549 95597 10.22059/poll.2023.365500.2076 EN Anurag Samson Lall Department of Chemistry, Sam Higginbottom University of Agriculture, Technology and Sciences, Prayagraj, Uttar Pradesh, 211007, India 0000-0002-7961-232X Avinash Kumar Pandey Department of Chemistry, GLA University, Chaumuhan, Mathura, Uttar Pradesh- 281406, India Jyoti Vandana Mani Department of Chemistry, Sam Higginbottom University of Agriculture, Technology and Sciences, Prayagraj, Uttar Pradesh, 211007, India Journal Article 2023 09 17 Cedrus deodara is a coniferous tree native to Himalayan region. Its wood is a valuable resource for the timber industry; however, its bark is typically discarded as a waste material. The present study examines the performance of Cedrus deodara bark powder (CD) as an inexpensive adsorbent for elimination of Pb (II) ions. In addition to this multiple linear regression (MLR) and artificial neural network (ANN) models were developed for modelling the adsorption process and prediction of Pb (II) removal efficiency. The structural and chemical properties of CD were explored using Field Emission Scanning Electron Microscope (FE-SEM), Energy Dispersive Spectrometer (EDS), X-Ray Diffractometer (XRD) and Fourier Transform Infrared Spectroscopy (FTIR). Batch experiments were conducted to investigate the influence of factors including pH, contact time, initial Pb (II) concentration and temperature on Pb (II) adsorption. The adsorption followed pseudo-second-order kinetic and Langmuir isotherm models with maximum monolayer uptake capacity 77.52 mg/g. Based on the thermodynamic criteria, the process was endothermic and spontaneous with enthalpy change (ΔH = 8.08 kJ/mol), free energy change (ΔG = -2.44 kJ/mol) and entropy change (ΔS = 0.03 kJ/K/mol). Statistical comparison of MLR model (R2 = 0.817, RMSE = 8.954, MAPE = 17.379 %) and ANN model (R2 = 0.993, RMSE = 1.777, MAPE = 2.054 %) confirmed that ANN model was far more accurate in predicting removal efficiency. https://jpoll.ut.ac.ir/article_95597_96e780727281bc3d109b130bfe7f41dd.pdf