The water absorption (WA) behavior of hybrid biocomposites reinforced with Syagrus romanzoffiana palm fibers (SrPFs) and Syagrus romanzoffiana palm waste (biochar, SrPW) shows a nonlinear relationship with immersion time and biochar content. This study explores the use of response surface methodology (RSM) and artificial neural networks (ANN) for multi-objective optimization and predictive modeling of this behavior. A hybrid model combining ANN with a genetic algorithm (GA) and RSM was developed to predict WA over immersion periods from 24 to 720 hours and SrPW contents from 0.5% to 2%. The model was further enhanced using a Multi-Criteria Decision-Making (MCDM) approach with the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS). The ANN-GA model predicted WA with reasonable accuracy (2.72% at 672 hours), closely matching experimental results, as confirmed by validation tests. At 615 hours, RSM optimization resulted in a slightly lower WA (2.68%). These findings demonstrate the reliability of the proposed modeling framework in reducing experimental work and guiding material design. The biocomposite exhibits potential for eco-friendly applications, particularly in the automotive sector.
Ghernaout et al. (Tue,) studied this question.