Scholarlink

<p>In order to comprehensively and effectively quantify the aflatoxin B1 (AFB1) content in wheat, this study proposes a method based on Raman spectroscopy and Fourier transform near-infrared spectroscopy (FT-NIR) combined with data fusion for the testing of moldy wheat samples. FT-NIR and Raman spectra were collected from wheat samples with different AFB1 contents using an FT-NIR spectrometer and a Raman spectrometer. Separate PLSR models were then constructed for each type of spectrum. Based on the single-spectrum model, the data-layer, feature-layer and decision-layer fusion of FT-NIR and Raman spectra were performed to establish the corresponding PLSR model aimed at accurately measuring AFB1 levels in wheat. The experimental results show that the data-fused PLSR models significantly improve the detection performance and generalization ability compared with the mono-spectral PLSR models, especially the feature-layer fusion PLSR model based on BOSS and the decision-layer fusion PLSR model based on VCPA. Among them, the VCPA-based decision-layer fusion PLSR model produced the best performance with <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\:{\text{R}}_{\text{p}}^{2}\)</EquationSource> </InlineEquation> of 0.92, RMSEP of 7.57, and RPD of 3.69. These results indicate that a data fusion scheme based on FT-NIR spectroscopy and Raman spectroscopy can be realized to effectively, comprehensively, and accurately quantify the AFB1 content in wheat. This study is of great importance in the field testing of food quality and safety.</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

High-precision detection of AFB1 content in wheat based on vibrational spectroscopy and data fusion strategy

  • 期刊论文

  • 发表日期：

    2026年5月26日

查看全文
  • Chengyun Zhu,
  • Yanheng Wang,
  • Zhanbo Ji,
  • Jingwen Zhu,
  • Wangfei Luo,
  • Hui Jiang

摘要

In order to comprehensively and effectively quantify the aflatoxin B1 (AFB1) content in wheat, this study proposes a method based on Raman spectroscopy and Fourier transform near-infrared spectroscopy (FT-NIR) combined with data fusion for the testing of moldy wheat samples. FT-NIR and Raman spectra were collected from wheat samples with different AFB1 contents using an FT-NIR spectrometer and a Raman spectrometer. Separate PLSR models were then constructed for each type of spectrum. Based on the single-spectrum model, the data-layer, feature-layer and decision-layer fusion of FT-NIR and Raman spectra were performed to establish the corresponding PLSR model aimed at accurately measuring AFB1 levels in wheat. The experimental results show that the data-fused PLSR models significantly improve the detection performance and generalization ability compared with the mono-spectral PLSR models, especially the feature-layer fusion PLSR model based on BOSS and the decision-layer fusion PLSR model based on VCPA. Among them, the VCPA-based decision-layer fusion PLSR model produced the best performance with \(\:{\text{R}}_{\text{p}}^{2}\) of 0.92, RMSEP of 7.57, and RPD of 3.69. These results indicate that a data fusion scheme based on FT-NIR spectroscopy and Raman spectroscopy can be realized to effectively, comprehensively, and accurately quantify the AFB1 content in wheat. This study is of great importance in the field testing of food quality and safety.