Prediction and optimisation of Cr (VI) removal by green and biodegradable adsorbent from aqueous solution using Deep machine learning (ANN and ANFIS)

This investigation aims to assess the performance of a green adsorbent in a laboratory-scale reactor for removing chromium (VI) from wastewater. Cellulose nanocrystals are excellent materials for removing heavy metal ions because of their biodegradability, availability and adaptability in dynamic and static adsorption processes. The green adsorbent was characterised using FTIR and SEM. A deep machine learning approach (artificial neural network and adaptive neuro-fuzzy inference system) was used to forecast the adsorption capacity of green and biodegradable adsorbent to remove chromium (VI) from wastewater. Four operational variables were input: pH, starting concentration of Cr (VI), contact duration, adsorbent dose, and the adsorption capacity output. The network was trained using feedforward propagation using the Levenberg–Marquardt algorithm (LM). ANN models with three algorithms (purelin, logsig, and transig) and ANFIS models were tested to optimise, develop, and forecast the chromium (VI) adsorption using a green and biodegradable adsorbent. The optimum conditions were pH 6, concentration 50 mg/L, time 120 min, and adsorbent dosage 10 g/ 100 mL. The findings show that artificial neural network models effectively predict chromium (VI) adsorption. In the training dataset, R² was 0.979, Mean Square Error (MSE), absolute average relative error (AARE) was 0.053, root mean square error (RMSE) was 0.077, and absolute average relative error (AARE) was 0.053 for the artificial neural network. For the adaptive neuro-fuzzy system, an RMSE of 0.021, AARE of 0.015, ARE of 0.01, MSE of 0.017, and R² of 0.998 were obtained.