Deep Learning Approach for the Surface Resistivity of a Multilayer Structure

Novinfard E., ÇINAR G.

2026 Symposium on Microwave, Antenna, and Propagation, SMAP 2026, Jakarta, Endonezya, 16 - 17 Ocak 2026, ss.180-185, (Tam Metin Bildiri) identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/smap67528.2026.11439383
  • Basıldığı Şehir: Jakarta
  • Basıldığı Ülke: Endonezya
  • Sayfa Sayıları: ss.180-185
  • Anahtar Kelimeler: deep learning, electromagnetic modeling, multilayer structures, Surface resistivity
  • Eskişehir Osmangazi Üniversitesi Adresli: Evet

Özet

Accurate prediction of surface resistivity is important for analyzing multilayer electromagnetic structures, but classical multilayer solvers become computationally expensive for wide parameter sweeps. This work proposes a deep learning surrogate that predicts the effective surface resistivity from material parameters, layer thicknesses, incidence angle, and frequency. Training data are generated using exact multilayer formulations with the surface transition condition. The resulting neural network achieves high accuracy across diverse configurations, capturing smooth and resonance-dominated responses. The model provides a fast alternative to traditional multilayer solvers for design and optimization tasks.