The Use of the Artificial Neural Network (ANN) for Modeling of Thermogravimetric Curves of Tobacco Stalk Waste Exposed to Alkaline Treatment

Abstract

Tobacco is a rich source of cellulosic material and one of the most cultivated non-food plants in the world with great potential for incorporation in polymeric matrices. The use of tobacco residues as reinforcing filler requires chemical/physical fiber treatment aiming to maximize compatibility with the polymer. In this study, tobacco residues were treated with two concentrations of NaOH (10 or 15 wt.%) at two-time exposures (3 or 5 h). Four distinct heating rates were used for each condition. It was applied an artificial neural network to model the thermogravimetric curves. After, the fitted ANN curves were used to create a 3D surface response. The equations from 3D surface response allowed the creation of thermogravimetric curves in any heating rate situated between the minimum and maximum range tested.

Keywords

• tobacco residue
• alkaline treatment
• artificial neural network
• surface response methodology