Automated Genetic Algorithm-Based Impedance Bandwidth Optimization of a Triangular U-Slotted Microstrip Patch Antenna Using Python and HFSS

Abstract

Microstrip patch antenna optimisation using full-wave electromagnetic simulations can become computationally expensive when multiple geometrical parameters are involved. This paper presents a Python–PyAEDT-based automated optimisation framework for enhancing the impedance bandwidth of a compact triangular U-slotted microstrip patch antenna. The antenna was initially designed for operation around 5 GHz on a 40 mm × 40 mm substrate and was subsequently modified to a compact 24 mm × 30 mm FR-4 substrate. The final configuration incorporates additional triangular and circular patch elements, a U-shaped slot, and a modified ground structure. Following the initial antenna modelling and parameterisation in Ansys High Frequency Structure Simulator (HFSS), the framework automatically updates six geometrical parameters, executes electromagnetic simulations, extracts the reflection coefficient (S₁₁), and calculates the continuous −10 dB impedance bandwidth. Genetic Algorithm (GA)-based optimisation and script-based random search were each evaluated using 100 HFSS simulations. The GA achieved an impedance bandwidth of 2357.65 MHz over approximately 3.202–5.559 GHz, with a minimum S₁₁ of −24.128 dB. In comparison, random search achieved a bandwidth of 2260.37 MHz over approximately 3.459–5.720 GHz, with a minimum S₁₁ of −25.852 dB. The GA-based solution provided an impedance bandwidth 4.30% greater than that of the random-search solution in the recorded runs, whereas random search produced a more negative minimum S₁₁. The recorded total wall-clock times were 32.42 minutes for GA and 50.34 minutes for random search. These results demonstrate the feasibility of automating antenna optimisation and comparing alternative search strategies under an equal simulation budget. However, because each method was evaluated in a single run and employed a different candidate-selection criterion, repeated trials are required to establish the robustness of the observed performance difference.