Scholarlink

<p>The economic value of summer-autumn tea can be enhanced through fermentation. However, the real-time monitoring quality of fermentation production faces significant challenges. Therefore, an on-line visible-near infrared spectroscopy (VIS-NIR) taste compounds monitoring system was developed to assist the fermentation production. Linear and non-linear VIS-NIR calibration methods were evaluated, and a parallel weighted hybrid modeling (PWHM) strategy was first proposed to elucidate the spectral-composition mapping. The results demonstrated that model optimization significantly improved performance, with non-linear models generally outperforming linear ones. Specifically, the PWHM models achieved excellent predictive accuracy for total sugars, total acids, and L-Theanine, showing high ratio of performance to deviation (RPD) and low mean absolute error (MAE). The fundings showed that the proposed method can meet the detection requirements for the fermentation process of summer-autumn tea. This study provides a technical reference for digital transformation and efficient production in the food fermentation industry.</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

On-line perception of taste compounds during summer-autumn tea fermentation based on visible-near infrared spectroscopy: A linear and non-linear hybrid modeling strategy

  • 期刊论文

  • 发表日期：

    2026年1月30日

查看全文
  • Songguang Zhao,
  • Tianhui Jiao,
  • Zhen Wang,
  • Yi Xu,
  • Selorm Yao-Say Solomon Adade,
  • Qin Ouyang,
  • Quansheng Chen

摘要

The economic value of summer-autumn tea can be enhanced through fermentation. However, the real-time monitoring quality of fermentation production faces significant challenges. Therefore, an on-line visible-near infrared spectroscopy (VIS-NIR) taste compounds monitoring system was developed to assist the fermentation production. Linear and non-linear VIS-NIR calibration methods were evaluated, and a parallel weighted hybrid modeling (PWHM) strategy was first proposed to elucidate the spectral-composition mapping. The results demonstrated that model optimization significantly improved performance, with non-linear models generally outperforming linear ones. Specifically, the PWHM models achieved excellent predictive accuracy for total sugars, total acids, and L-Theanine, showing high ratio of performance to deviation (RPD) and low mean absolute error (MAE). The fundings showed that the proposed method can meet the detection requirements for the fermentation process of summer-autumn tea. This study provides a technical reference for digital transformation and efficient production in the food fermentation industry.