Novel use of artificial neural networks in predicting kinetic and equilibrium parameters for methylene blue adsorption on tomato stem

In this study, methylene blue (MB) adsorption onto tomato stem (TS) was comprehensively investigated, and the potential of TS as an effective biosorbent in wastewater treatment was examined. Additionally, the predictive power of the Artificial Neural Networks (ANN) model on the adsorption isotherms, kinetics, and thermodynamics was evaluated and compared with experimental calculations. The adsorption experiments were carried out in the batch system with MB solutions at various initial concentrations (20, 40, 60, 80, and 100 mg/L), temperatures (298, 308, and 318 K), and contact times (up to equilibrium). The adsorption kinetics were examined using nine different initial concentrations predicted by the ANN, and detailed kinetic analyses were performed at minute-level resolution. The ANN model successfully predicted the experimental results for untested concentrations and contact times. However, it was not sufficiently suitable for thermodynamic analysis. The Langmuir isotherm provided good agreement at 298 and 308 K, while the ANN model showed a stronger correlation with the Temkin isotherm.The novelty of this study is the first comprehensive assessment of MB adsorption by TS biosorbent by integrating the ANN model with both the adsorption equilibrium and kinetic analysis. This dual-perspective investigation enhances the understanding of adsorption mechanisms and supports the application of the ANN models to real processes.