Adaptive neuro-fuzzy inference system (ANIFS) and artificial neural network (ANN) applied for indium (III) adsorption on carbonaceous materials

Abstract

In this paper, we present an initial study relating the adsorption of indium (III) onto carbonaceous materials, namely the activated carbon (AC), multiwalled carbon nanotubes functionalized with OH (MWCNT–OH), and the multiwalled carbon nanotubes functionalized with COOH (MWCNT–COOH). The main objective of this study is the development of the adaptive neuro-fuzzy inference system (ANFIS) and an artificial neural network (ANN) for predicting the adsorption capacity in different operating conditions for different materials. Both models take into account the adsorbent type, adsorbent dosage (0.05, 0.25, 0.5, 1.0, 1.5, and 2.0 g L⁻¹), and the contact time (5, 20, 60, and 120 min) for predicting the adsorption capacity, which varied from 12.896 to 981.000 mg g⁻¹, a total record of 72 was used. Both modeling methodologies applied can represent the experimental data, taking into account the statistical values obtained. The ANFIS achieved the best performance when the hybrid method was selected, this leads into R of 0.9998, RMSE of 48,373 with 250 epochs. On the other hand, the ANN can represent the best performance when using the Levenberg–Marquardt algorithm, reaching an R of 0.9831, MSE of 0.0180 and 9 epochs. Considering the modeling and experimental aspects indicates that the increase of the adsorbent dosage diminished the adsorbent capacity. The increase of the contact time causes the effect to increase the adsorption capacity until its equilibrium. Lastly, it is possible to conclude that the MWCNT–COOH is the most suitable adsorbent to be used between the selected materials.

Keywords:

• Adaptive neuro-fuzzy inference system
• Adsorption
• Artificial neural network
• Carbon nanotubes
• Indium
• Modeling