Artificial neural network and multiple linear regression for modeling sorption of Pb²⁺ ions from aqueous solutions onto modified walnut shell

ABSTRACT

In the present work, for the first time, a new carboxylate-functionalized walnut shell (CFWS) was prepared via esterification of walnut shell (WS) with isopropylidene malonate. The characterization of the CFWS by different techniques approved that carboxylic groups were introduced onto the surface of WS. The performance of the modified adsorbent was studied for the removal of Pb²⁺ ions from aqueous solutions in a batch adsorption system. The analysis data showed that the Langmuir isotherm could satisfactorily explain the equilibrium data, and the maximum adsorption capacity for Pb²⁺ ions was found to be 192.3 mg g⁻¹ at 0.8 g L⁻¹ of the adsorbent, pH 5.5, and a temprature of 298 K. Two models, namely artificial neural network (ANN) and multiple linear regression (MLR), were used to construct an empirical model for prediction of the removal percentage of Pb²⁺ ions under different experimental conditions. These models were validated using a test set of 20 data. A comparison between the developed models shows that the ANN model is able to predict the removal percentage of Pb²⁺ ions more accurately. Consequently, the ANN model could be applied for the design of an automated wastewater remediation plan. Also it has to be noted that the used CFWS was recovered using EDTA-2Na, and employed for the removal of Pb²⁺ ions from aqueous solutions.

KEYWORDS:

• Modeling
• wastewater
• removal
• agricultural adsorbent

Acknowledgments

The authors are thankful to the Shahrood University of Technology Research Council for the support of this work.