Sustainable Performance Optimization of Natural Fiber Composites Using CRITIC-MEW and Generative Adversarial Networks

For systematic problem-solving and decision-making in the optimization of natural fiber-reinforced friction composites, the CRITIC-MEW methodology serves as a robust and objective multi-criteria decision-making tool. To overcome limitations associated with sparse experimental datasets and to enhance predictive modeling, Generative Adversarial Networks (GANs) were employed to generate synthetic data, thereby enriching the design space and strengthening the optimization process. This study investigates the synergistic integration of CRITIC-MEW and GANs to improve composite performance and presents a structured strategy for optimizing natural fiber composites with respect to mechanical performance and sustainability. The GAN-generated data effectively capture the underlying distributions and correlations among composite attributes, leading to more reliable and consistent predictions. Experimental results indicate that the percentage of jute fiber reinforcement significantly influences performance variations. While the J10K1 composition exhibits superior shear strength among the tested formulations, the CRITIC-MEW analysis, accounting for all physio-mechanical and physical criteria, identifies the composite with 5 wt.% jute and 15 wt.% kenaf (J5K15) as the most balanced and overall optimal formulation.