Machine Learning Techniques for the Optimization of Polymer Composites Used in EMI Shielding

Abstract

The increasing use of electronic devices has led to the rising challenge of electromagnetic interference (EMI), necessitating efficient shielding materials. However, adding excessive fillers to polymer composites to enhance EMI shielding often results in higher costs, poor dispersion, and mechanical weakness. This research proposes a hybrid polymer composite using a Graphene substrate combined with conductive polymers like polyacetylene and multi-walled carbon nanotubes (MWCNTs), aimed at improving EMI shielding performance while maintaining mechanical integrity. The proposed design and optimization approach are assessed using ANSYS-HFSS software. To further enhance the prediction and optimization of EMI shielding effectiveness, a novel "bio-inspired predictive insight and optimization strategy" is introduced. This strategy employs "Bayesian-enriched genetic programming" to handle the complex, nonlinear relationships in polymer composites, addressing the multi-objective optimization challenges. Additionally, the "multi-objective dominant crowding seagull (MDCS) optimization" model is used to optimize conflicting objectives and EMI shielding simultaneously, demonstrating improved shielding efficiency and superior performance in terms of prediction accuracy (98.7%), faster training time (19 seconds), and quicker prediction time (6 seconds). The results suggest that this combined approach can significantly enhance the mechanical, electrical, and EMI shielding properties of polymer composites, offering a promising solution for modern electronic applications.