Scholarlink

This study demonstrates a rapid and non-invasive method for classification and estimation of catechin and caffeine changes with storage time in tea sample by near-infrared (NIR) spectroscopy. The NIR spectra of 34 tea samples were used to evaluate the modelling and prediction performance of a combination of Optuna with three machine learning models. Result showed caffeine prediction with MSEP and R2 values of 0.27 and 0.98, respectively. However, smaller MSEP of 0.04 was found for catechin with R2 value of 0.96. These findings suggest that NIR spectroscopy based chemometric analysis can be used to predict the changes in biochemical content in tea with storage time.

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

NIR Spectroscopy Based Non-invasive Assessment of Tea Quality

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Onkar Sarma,
  • Kavya Dashora

摘要

This study demonstrates a rapid and non-invasive method for classification and estimation of catechin and caffeine changes with storage time in tea sample by near-infrared (NIR) spectroscopy. The NIR spectra of 34 tea samples were used to evaluate the modelling and prediction performance of a combination of Optuna with three machine learning models. Result showed caffeine prediction with MSEP and R2 values of 0.27 and 0.98, respectively. However, smaller MSEP of 0.04 was found for catechin with R2 value of 0.96. These findings suggest that NIR spectroscopy based chemometric analysis can be used to predict the changes in biochemical content in tea with storage time.