COMPARISON OF PREDICTIVE MODELS FOR THE IDENTIFICATION OF CRITICAL VARIABLES IN THE ALKALINE TREATMENT OF SISAL FIBER COMPOSITES

This work investigates the influence of sodium hydroxide (NaOH) concentration and immersion time on the tensile strength of sisal fiber composites submitted to alkaline treatment. Three predictive models were compared: multiple linear regression, second-degree polynomial regression, and artificial neural network. The models were adjusted based on 48 specimens distributed in 16 experimental combinations, maintaining a constant 800 mL of distilled water and 30 g of fiber per beaker. The results showed a progressive increase in the coefficient of determination, from R² = 0.218 in the linear model to R² = 0.665 in the neural network, evidencing nonlinear behavior between the variables studied and the mechanical resistance. The three models pointed out the immersion time as the variable with the greatest influence on tensile stress, presenting a negative effect when it was longer than 4 hours.