AI-Driven Optimization for Low-Profile Multi-Band Antennas for Direction Finding

This paper presents an AI-driven approach to optimize a low-profile, multi-band antenna for direction-finding applications. A Feed-Forward Back Propagation (FFBP) neural network was trained on 84 antenna configurations, simulated using a Method of Moments (MoM)-based tool, to predict resonant frequencies, VSWR, and gain across four frequency bands (433 MHz, 877.5 MHz, 2.4 GHz, and 5.8 GHz). The proposed method dramatically reduces computational cost while maintaining accuracy. Compared to a brute-force approach requiring over 262,000 full-wave simulations, our technique achieves similar precision with only 84 simulations, followed by rapid AI-based predictions. These results highlight the potential of machine learning to enhance antenna design, particularly for vehicular and electromagnetic source detection applications.