Scholarlink

<p>This paper proposes a unified framework that leverages biocompatible chitosan substrates and AI-driven optimization to advance terahertz (0.3 THz) Biomedical microstrip patch antennas. Unlike existing approaches that focus solely on electromagnetic performance, this study combines biocompatible material integration with graphene radiators and full SAR validation, hybrid photonic bandgap (PBG3) optimization using evolutionary algorithms (GA, PSO, CMA-ES), and machine learning-based rapid performance prediction. Comparative optimization in CST demonstrated that CMA-ES is the most effective method, achieving a peak gain of 6.11 dBi and 87.4% efficiency at 0.3 THz at H = 172&#xa0;µm. A Random Forest regression model trained on CMA-ES data achieved &lt; 2% prediction error for gain and efficiency, enabling accelerated design exploration. SAR simulations confirmed compliance with FCC and ICNIRP safety limits (0.03 W/kg). The proposed integration of biopolymer substrates, PBG optimization, and AI prediction establishes a novel and impactful pathway for THz biomedical antenna design.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

AI Driven Design of High-Gain THz Antennas Using PBG Structures: Combining Evolutionary Algorithms and Machine Learning for Implantable Biomedical Systems

  • 期刊论文

  • 发表日期：

    2025年12月4日

查看全文
  • Mustapha Reguig,
  • Mohammed Belkheir,
  • Allel Mokaddem,
  • Mehdi Rouissat,
  • Djamila Ziani

摘要

This paper proposes a unified framework that leverages biocompatible chitosan substrates and AI-driven optimization to advance terahertz (0.3 THz) Biomedical microstrip patch antennas. Unlike existing approaches that focus solely on electromagnetic performance, this study combines biocompatible material integration with graphene radiators and full SAR validation, hybrid photonic bandgap (PBG3) optimization using evolutionary algorithms (GA, PSO, CMA-ES), and machine learning-based rapid performance prediction. Comparative optimization in CST demonstrated that CMA-ES is the most effective method, achieving a peak gain of 6.11 dBi and 87.4% efficiency at 0.3 THz at H = 172 µm. A Random Forest regression model trained on CMA-ES data achieved < 2% prediction error for gain and efficiency, enabling accelerated design exploration. SAR simulations confirmed compliance with FCC and ICNIRP safety limits (0.03 W/kg). The proposed integration of biopolymer substrates, PBG optimization, and AI prediction establishes a novel and impactful pathway for THz biomedical antenna design.