EUVIMEDEUVIMED entdecken
Crossref

Inverse design of terahertz metasurface polarization converters based on generative adversarial networks

Linjing Liang, Wenrui Xue, Yue Zhang

Optics Continuum · 15.11.2026 · 5 (11) · 3647

DOI: 10.1364/optcon.601366 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Optics Continuum, online erschienen am 11.11.2026. Verfasst von Linjing Liang und 2 weiteren Autor:innen. Die Arbeit nennt 39 Literaturangaben. Gefördert von: National Basic Science Talents Training Fund of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

A reflection-type broadband terahertz metasurface polarization converter with a three-layer structure—comprising a patterned metal layer, a dielectric layer, and a metallic ground plane—is inversely designed based on a Wasserstein generative adversarial network with gradient penalty (WGAN-GP). By incorporating the Wasserstein distance and a gradient penalty term, the WGAN-GP effectively overcomes the training instability inherent in conventional generative adversarial networks, thereby enabling the construction of an end-to-end mapping model from target spectra to metasurface structural parameters. The results demonstrate that the WGAN-GP achieves efficient and high-accuracy inverse design, with a single inference time of only 30 ms and a mean absolute error of structural parameter prediction as low as 0.1263. The optimized metasurface polarization converter exhibits a consistently high polarization conversion ratio exceeding 0.8 over the 3–7.5 THz frequency range, corresponding to a relative bandwidth of up to 86%, and maintains excellent angular stability within an incidence angle range of 0° to 50°. This work not only verifies the efficiency and reliability of WGAN-GP in the inverse design of complex electromagnetic devices, but also provides an effective solution for the development of high-performance terahertz metasurfaces, with the potential to accelerate device innovation and applications of terahertz technologies in key fields such as communications, sensing, imaging, and spectrum security.

Förderung

  • National Basic Science Talents Training Fund of China

Zitieren

APA

Liang, L., Xue, W., & Zhang, Y. (2026). Inverse design of terahertz metasurface polarization converters based on generative adversarial networks. Optics Continuum, 5(11), 3647. https://doi.org/10.1364/optcon.601366

Vancouver

Liang L, Xue W, Zhang Y. Inverse design of terahertz metasurface polarization converters based on generative adversarial networks. Optics Continuum. 2026;5(11):3647. doi:10.1364/optcon.601366

BibTeX
@article{Liang2026,
  author = {Liang, Linjing and Xue, Wenrui and Zhang, Yue},
  title = {Inverse design of terahertz metasurface polarization converters based on generative adversarial networks},
  journal = {Optics Continuum},
  year = {2026},
  volume = {5},
  number = {11},
  pages = {3647},
  doi = {10.1364/optcon.601366}
}
Kostenlos starten →