Prediction of Lead (II) Removal from Aqueous Solution via the Adsorption Process: Convolutional Neural Network and Artificial Neural Network Approach

This work used modified cellulose nanocrystals (CNCs), which are environmentally sustainable and biodegradable adsorbents, to remove lead (II) from aqueous solution. A deep learning technique, such as a convolutional Neural Network and an artificial neural network, was employed to predict the efficiency of biodegradable and environmentally friendly adsorbents. A maximum adsorption capacity of 270 mg/g with a percentage removal of 97% was achieved at pH 5, a contact time of 300 minutes, an adsorbent dosage of 20 mg, and a pollutant concentration of 315 mg/L. The LM algorithm was used to train the artificial neural network and Convolutional Neural Network. It consists of 5 inputs, 2 outputs, and 10 hidden layers. The outputs of the ANN and CNN models were evaluated against real data via R² and MSE. The ideal artificial neural network (ANN) model produced both accurate and dependable findings, with R² values of 0.996 and an error of 0.012. The Langmuir model most accurately explained the adsorption of lead (II). The model showed very high accuracy, with an R² value of 0.994 and very small errors, including a mean percentage standard deviation (MPSD) of 0.0011, a root mean square error (RMSE) of 0.025, and a mean squared error (MSE) of 0.005. The pseudo-second-order model describes the adsorption process, whereas the Dubinin‒Radushkevich model suggests that a chemical reaction guides the process.