<![CDATA[Microstrip patch antennas are vital for Ka-band communication owing to their compact size and high performance. This study introduces a modified patch design at 28 GHz featuring two corner truncations and dual-slot integration to enhance impedance matching and broaden the operational bandwidth. The objective of this work is to investigate whether geometrical modifications combined with intelligent modelling can yield improved performance metrics while accelerating the performance evaluation phase through a data-driven surrogate model. The proposed antenna was developed through parametric optimization in Ansys HFSS, in which its structure was systematically varied to achieve stable resonance and improved radiation performance. The optimized prototype achieves a simulated return loss of −67.11 dB, a bandwidth of 3.8 GHz, a VSWR of 1.0009, a peak gain of 7.65 dB, and an input impedance of 50.01 Ω, all indicating strong simulated electromagnetic performance. The design demonstrates a deep resonance corresponding to a high quality (Q) factor, making it a suitable candidate for applications where precise frequency selectivity is paramount. To accelerate evaluation, a machine learning framework was integrated, using 65,682 simulated samples to train regression models for predicting return loss. Among the tested algorithms, the Random Forest Regressor demonstrated the highest accuracy with a mean absolute error of 0.0471 dB and an R² of 0.9995. The integration of electromagnetic simulation and ML-assisted performance prediction demonstrates a reliable pathway for rapid evaluation of Ka-band antennas, offering strong potential for next-generation satellite and wireless communication systems.]]>
Copyrights © 2025
