Scholarlink

<p>This article presents a compact dual-band microstrip bandpass filter, realized through coupling-matrix synthesis and experimentally validated as a microwave sensor for precise, direct-contact detection of water contamination in petroleum fluids. The asymmetric layout integrates a rectangular resonator with asymmetrically coupled open-ended stubs for the lower passband (centered at 3.48&#xa0;GHz) and a pair of gear-shaped resonators with interlocking peripheral teeth for the upper passband (centered at 12.05&#xa0;GHz), achieving a footprint of 13.2&#xa0;mm × 9.9&#xa0;mm with highly selective responses. Direct-contact measurements on gasoline, crude oil, and diesel mixtures, prepared across the full 0–100% water range in 5% volumetric steps, exhibit consistent, monotonic downward resonance shifts driven by the large dielectric contrast between hydrocarbons and water. The sensor demonstrates high sensitivity, with an average value of approximately 524 MHz/ε<sub>r</sub> across both passbands, enabling excellent resolution even in low-contamination regimes. To convert the dual-band frequency displacements into accurate water-content estimates, a Group Method of Data Handling (GMDH) neural network is employed, which autonomously constructs interpretable polynomial models tailored to each fuel. The approach delivers outstanding test-set performance with RMSEs of 0.78% (gasoline), 0.92% (crude oil), and 0.85% (diesel), MAEs of 0.61%, 0.74%, and 0.68%, respectively, and strong correlation with measurements (R² = 0.992). Over 85% of absolute errors fall within ± 1.2% in a near-symmetric distribution, confirming the method’s robustness and its broader potential for multiphase dielectric sensing in industrial and environmental applications.</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

Coupling-matrix-designed dual-band sensor for high-accuracy water quantification in petroleum mixtures with self-organizing GMDH prediction

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月10日

查看全文
  • Farhad Vahdani Dehkalani,
  • Mohsen Hayati,
  • Ashkan Horri,
  • Mohammad Bagher Tavakoli

摘要

This article presents a compact dual-band microstrip bandpass filter, realized through coupling-matrix synthesis and experimentally validated as a microwave sensor for precise, direct-contact detection of water contamination in petroleum fluids. The asymmetric layout integrates a rectangular resonator with asymmetrically coupled open-ended stubs for the lower passband (centered at 3.48 GHz) and a pair of gear-shaped resonators with interlocking peripheral teeth for the upper passband (centered at 12.05 GHz), achieving a footprint of 13.2 mm × 9.9 mm with highly selective responses. Direct-contact measurements on gasoline, crude oil, and diesel mixtures, prepared across the full 0–100% water range in 5% volumetric steps, exhibit consistent, monotonic downward resonance shifts driven by the large dielectric contrast between hydrocarbons and water. The sensor demonstrates high sensitivity, with an average value of approximately 524 MHz/εr across both passbands, enabling excellent resolution even in low-contamination regimes. To convert the dual-band frequency displacements into accurate water-content estimates, a Group Method of Data Handling (GMDH) neural network is employed, which autonomously constructs interpretable polynomial models tailored to each fuel. The approach delivers outstanding test-set performance with RMSEs of 0.78% (gasoline), 0.92% (crude oil), and 0.85% (diesel), MAEs of 0.61%, 0.74%, and 0.68%, respectively, and strong correlation with measurements (R² = 0.992). Over 85% of absolute errors fall within ± 1.2% in a near-symmetric distribution, confirming the method’s robustness and its broader potential for multiphase dielectric sensing in industrial and environmental applications.