Comparative Modeling and Prediction of the Mechanical Strength of Chemically Modified Bamboo Fibers Using Artificial Neural Networks and Response Surface Methodology

Artificial neural networks (ANN) in artificial intelligence (AI) have been utilized to solve an array of engineering challenges. At the same time, response surface methodology (RSM) is a modeling strategy to maximize the response or responses when two or more quantitative components are involved. This inquest presents a comparative investigation between ANN and RSM modeling and prediction of the tensile strength (TS) of chemically modified natural fibers (NF). In this investigation, ethanol (C₂H₆O) was utilized to modify bamboo fibers (BmF) at varied durations, followed by potassium permanganate (KMnO₄) in acetone solution at different concentrations and time ranges. ANN and RSM via Box-Behnken design (BBD) were employed to predict the impact of these chemical treatment parameters on the TS of the modified BmF. The treatment variables significantly affect the TS of BmF as indicated by the analysis of variance (ANOVA), which reveals the statistical significance of the model at a level of p <.0001. The calculated coefficients of R² (correlation coefficient), RMSE (root mean square error), and MAE (mean absolute error) for each output parameter were used to compare the prediction abilities of the ANN and RSM. The outcomes show that a well-trained ANN is superior to the RSM-BBD model. The current investigation has exhibited the efficacy of the ANN and RSM-BBD modeling methodology in promptly attaining suitable mechanical property values, curtailing manufacturing expenses, and preserving resources.