Scholarlink

<p>Wheat moisture content is crucial for storage stability and processing quality, yet accurate detection in irregularly shaped bags and thickness variation remains challenging for conventional microwave reflection methods. This study innovatively presents a multi-region moisture detection method with meta-learning that fuses information between visual 3D reconstruction and microwave reflection. Surface morphology is first captured via 3D scanning to physically correct microwave incidence angles. A meta-learning framework then combines base learners optimized for different thickness levels, effectively decoupling thickness and moisture effects in microwave signals. The model achieves high prediction accuracy (<i>R</i><sup>2</sup> = 0.99) and reduces key error metrics by over 80% compared to conventional methods. In tests on irregular wheat bags, the mean absolute percentage error reached 2.10%, demonstrating improved practicality and reliability. This method offers an effective solution for online, non-destructive, multi-region moisture monitoring in grain storage and processing.</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

Multi-region Microwave Detection for Moisture Content of Irregular Wheat Bags Using Information Fusion

  • 期刊论文

  • 发表日期：

    2026年1月12日

查看全文
  • Shujin Guo,
  • Xu Mao,
  • Sisi Zhou,
  • Hao Huang,
  • Dong Dai,
  • Du Chen,
  • Fozilov Golibjon,
  • Komil Astanakulov

摘要

Wheat moisture content is crucial for storage stability and processing quality, yet accurate detection in irregularly shaped bags and thickness variation remains challenging for conventional microwave reflection methods. This study innovatively presents a multi-region moisture detection method with meta-learning that fuses information between visual 3D reconstruction and microwave reflection. Surface morphology is first captured via 3D scanning to physically correct microwave incidence angles. A meta-learning framework then combines base learners optimized for different thickness levels, effectively decoupling thickness and moisture effects in microwave signals. The model achieves high prediction accuracy (R2 = 0.99) and reduces key error metrics by over 80% compared to conventional methods. In tests on irregular wheat bags, the mean absolute percentage error reached 2.10%, demonstrating improved practicality and reliability. This method offers an effective solution for online, non-destructive, multi-region moisture monitoring in grain storage and processing.