Scholarlink

<p>In this study, a self-developed K-band and KA-band microwave detection device was used to detect aflatoxin B1 (AFB1) in wheat with high accuracy. The microwave transmission spectra of 108 wheat samples with different mould levels were acquired, smoothed and preprocessed using Savitzky-Golay and moving average filters, and feature selection was performed using embedded algorithms such as Competitive Adaptive Re-weighted Sampling (CARS) and Multi-Feature Elimination using LASSO (MFE-LASSO). Based on the optimised feature wavelengths, Partial Least Squares Regression (PLSR) and Support Vector Machine Regression (SVR) models were constructed and the feature wavelengths were quadratically optimised by the Spotted Kingfisher Optimisation (PKO)、Hiking Optimisation Algorithm (HOA). The results showed that the CARS_PKO_PLSR model performed best on the prediction set with an RMSE of 3.4202 and an R<sup>2</sup> of 0.9843, which demonstrated that microwave detection based on the K-band and the KA-band was able to accurately predict the AFB1 content of wheat and demonstrated the potential of millimetre and centimetre waves in food safety detection.</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

Improvement of aflatoxin detection accuracy in wheat using microwave and key frequency optimized by metaheuristic algorithms

  • 期刊论文

  • 发表日期：

    2025年8月7日

查看全文
  • Fanzhen Meng,
  • Jihong Deng,
  • Leijun Xu,
  • Hui Jiang

摘要

In this study, a self-developed K-band and KA-band microwave detection device was used to detect aflatoxin B1 (AFB1) in wheat with high accuracy. The microwave transmission spectra of 108 wheat samples with different mould levels were acquired, smoothed and preprocessed using Savitzky-Golay and moving average filters, and feature selection was performed using embedded algorithms such as Competitive Adaptive Re-weighted Sampling (CARS) and Multi-Feature Elimination using LASSO (MFE-LASSO). Based on the optimised feature wavelengths, Partial Least Squares Regression (PLSR) and Support Vector Machine Regression (SVR) models were constructed and the feature wavelengths were quadratically optimised by the Spotted Kingfisher Optimisation (PKO)、Hiking Optimisation Algorithm (HOA). The results showed that the CARS_PKO_PLSR model performed best on the prediction set with an RMSE of 3.4202 and an R2 of 0.9843, which demonstrated that microwave detection based on the K-band and the KA-band was able to accurately predict the AFB1 content of wheat and demonstrated the potential of millimetre and centimetre waves in food safety detection.