Efficient thermal management stays a critical requirement in aerospace systems, high-power electronics, and compact heat exchangers, where excessive temperature rise can degrade performance and reliability. Conventional solid fins are limited by boundary-layer development and structural weight, motivating the use of perforated fins to enhance convective heat transfer while reducing material usage. This study develops a
computational framework that combines high-fidelity computational fluid dynamics (CFD), Radial Basis Function Neural Network (RBFNN) surrogate modeling, and multi-objective optimization to investigate the thermo-fluid performance of perforated fins under forced convection. The aims are to maximize heat transfer performance, represented by the Nusselt number (Nu), and minimize hydraulic losses, represented by
pressure drop (ΔP), while considering perforation number, perforation diameter, and airflow velocity as design variables. The RBFNN model was trained using CFD-generated data and confirmed against independent test datasets, showing high predictive accuracy (R² > 0.98) with low prediction error. It is important to note that validation in this study is limited to comparison with CFD results and previously published experimental correlations [4] and does not involve new experimental measurements. The multi-objective optimization, performed using NSGA- II, produced a Pareto frontier illustrating the trade-off between thermal enhancement and pressure loss. The optimized configurations achieved up to 16% improvement in heat transfer compared to a solid fin baseline, along with approximately 24% reduction in material usage. Sensitivity analysis using Sobol indices found airflow velocity as the dominant parameter, contributing approximately 60% to performance
variation. The proposed framework provides a computationally efficient approach for exploring perforated fin designs and supports engineering decision-making under competing thermal and hydraulic aims.