Publications Repository - Gdańsk University of Technology

Design of modern antenna systems heavily relies on numerical opti-mization methods. Their primary purpose is performance improvement by tun-ing of geometry and material parameters of the antenna under study. For relia-bility, the process has to be conducted using full-wave electromagnetic (EM) simulation models, which are associated with sizable computational expendi-tures. The problem is aggravated in the case of global optimization, typically carried out using nature-inspired algorithms. To reduce the CPU cost, popula-tion-based routines are often combined with surrogate modeling techniques, frequently in the form of machine learning procedures. While offering certain advantages, their efficiency is worsened by the curse of dimensionality and antenna response nonlinearity. In this article, we investigate computational ad-vantages of combining population-based optimization with variable-resolution EM models. Consequently, a model management scheme is developed, which adjusts the discretization level of the antenna under optimization within the continuous spectrum of acceptable fidelities. Starting from the lowest practically useful fidelity, the resolution converges to the highest assumed level when the search process is close to conclusion. Several adjustment profiles are considered to investigate the speedup-reliability trade-offs. Numerical results have been obtained for two microstrip antennas and particle swarm optimizer as a widely-used nature-inspired algorithm. Consistent acceleration of up to eighty percent has been obtained in comparison to the single-resolution version with minor deterioration of the design quality. Another attractive feature of our methodology is versatility and easy implementation and handling.

Authors

• dr hab. inż. Anna Pietrenko-Dąbrowska link open in new tab ,
• prof. dr inż. Sławomir Kozieł link open in new tab ,
• Leifur Leifsson