ABSTRACT This study investigates the effects of poly(lactic acid) (PLA), montmorillonite (MMT), and castor oil (CO) on the water vapor permeability (WVP) and tensile strength (TS) of poly(butylene adipate‐co‐terephthalate) (PBAT)/PLA films. A Box–Behnken experimental design is employed to generate experimental data for predictive modeling using response surface methodology (RSM) and artificial neural networks (ANN), validated using independent test samples. The ANN models exhibit superior predictive performance for both WVP and TS (correlation coefficient, R > 0.98), whereas the RSM model shows limited predictive accuracy for WVP ( R < 0.88). ANOVA and sensitivity analysis identify CO as the most influential factor affecting WVP, while PLA is the primary contributor to TS. Response surface analysis indicates that optimal WVP occurs at high CO (1.5 wt%) and intermediate MMT (1 wt%), whereas maximum TS requires high PLA (28.57 wt%) and MMT (1.5 wt%) with low CO (0.5 wt%), reflecting a trade‐off between barrier and mechanical reinforcement mechanisms. Furthermore, the ANN models are integrated as objective functions into a multi‐objective genetic algorithm (MOGA) to simultaneously maximize TS and minimize WVP, generating Pareto optimal solutions. These results demonstrate the effectiveness of combining RSM and ANN modeling with multi‐objective optimization for designing biodegradable packaging films with balanced barrier and mechanical properties.
García‐Carrillo et al. (Thu,) studied this question.