Scholarlink

<p>The color and browning dynamics of fruits and vegetables during drying critically determine product quality and market value. Conventional surface colorimetry cannot detect subsurface browning, while NIR and LF-NMR offer complementary, non-destructive insights into both compositional changes and water dynamics during drying. A synergistic LF-NMR-NIR framework integrating moisture-background elimination and data fusion, augmented by machine learning algorithms (partial least squares, support vector machine, and backpropagation artificial neural networks), was developed for real-time browning prediction and control. This framework was validated using three representative crops (carrots, bananas, and <i>Pleurotus eryngii</i>) processed via hot-air, infrared, microwave-vacuum, and freeze-drying. Optimization of water background elimination and <i>T₂</i> relaxation parameters fusion reduced root mean square error (RMSE) of browning models by 46.23%, 25.72%, and 41.41% (<i>P</i> &lt; 0.05) for carrots, bananas, and <i>Pleurotus eryngii</i>, respectively. The multifunctional BP-ANN model demonstrated superior accuracy (<i>R</i><sup>2</sup> &gt; 0.8912, RMSE &lt; 0.0593), achieving precise control in practical applications: carrot browning degrees were constrained to 0.17 ± 0.02 (appearance control) and 0.27 ± 0.03 (burn prevention) with 28% and 37% efficiency gains, respectively. Similarly, banana drying maintained browning indices at 0.43 ± 0.06 and 0.65 ± 0.09 (22–25% efficiency improvements), while <i>Pleurotus eryngii</i> exhibited controlled browning (0.27 ± 0.04 and 0.33 ± 0.07) with 26–32% time savings. This work introduces the AI-enhanced system for simultaneous surface-internal quality monitoring in drying. By synergizing multimodal sensing and AI, this study advances dynamic drying optimization, demonstrating transformative potential for sustainable agro-industrial practices.</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

An Innovative Color Control Strategy of Fruits and Vegetables During Drying

  • 期刊论文

  • 发表日期：

    2025年7月8日

查看全文
  • Qing Sun,
  • Yanan Yuan,
  • Min Zhang,
  • Feiyue Xu,
  • Zhimei Guo,
  • Xiaona Wang,
  • Jun Ren,
  • Jiyong Shi,
  • Xiaobo Zou

摘要

The color and browning dynamics of fruits and vegetables during drying critically determine product quality and market value. Conventional surface colorimetry cannot detect subsurface browning, while NIR and LF-NMR offer complementary, non-destructive insights into both compositional changes and water dynamics during drying. A synergistic LF-NMR-NIR framework integrating moisture-background elimination and data fusion, augmented by machine learning algorithms (partial least squares, support vector machine, and backpropagation artificial neural networks), was developed for real-time browning prediction and control. This framework was validated using three representative crops (carrots, bananas, and Pleurotus eryngii) processed via hot-air, infrared, microwave-vacuum, and freeze-drying. Optimization of water background elimination and T₂ relaxation parameters fusion reduced root mean square error (RMSE) of browning models by 46.23%, 25.72%, and 41.41% (P < 0.05) for carrots, bananas, and Pleurotus eryngii, respectively. The multifunctional BP-ANN model demonstrated superior accuracy (R2 > 0.8912, RMSE < 0.0593), achieving precise control in practical applications: carrot browning degrees were constrained to 0.17 ± 0.02 (appearance control) and 0.27 ± 0.03 (burn prevention) with 28% and 37% efficiency gains, respectively. Similarly, banana drying maintained browning indices at 0.43 ± 0.06 and 0.65 ± 0.09 (22–25% efficiency improvements), while Pleurotus eryngii exhibited controlled browning (0.27 ± 0.04 and 0.33 ± 0.07) with 26–32% time savings. This work introduces the AI-enhanced system for simultaneous surface-internal quality monitoring in drying. By synergizing multimodal sensing and AI, this study advances dynamic drying optimization, demonstrating transformative potential for sustainable agro-industrial practices.