|
| 基于近红外光谱技术对纳米化番茄红素液态饮品多成分 同步快速分析 |
| Rapid and simultaneous analysis of multiple components in nanostructured lycopene liquid beverages based on near-infrared spectroscopy |
| 投稿时间：2026-03-30 修订日期：2026-08-06 |
| DOI： |
| 中文关键词: 近红外光谱 纳米化番茄红素液态饮品 预处理组合优化 偏最小二乘回归 |
| 英文关键词:Near-infrared spectroscopy Nanostructured lycopene liquid beverage Combined pretreatment optimization Partial least squares regression |
| 基金项目:新疆自治区重点研发计划项目 |
|
|
|
|
| 摘要点击次数: 3 |
| 全文下载次数: 0 |
| 中文摘要: |
| 目的? 利用近红外光谱（Near infrared spectroscopy, NIRS）技术结合化学计量学手段，对纳米化番茄红素液态饮品中的多种营养成分指标构建快速预测模型。 方法? 采用NIRS结合偏最小二乘回归（Partial Least Squares Regression, PLSR）算法，提出组合预处理策略优化模型参数。针对番茄红素液态原液、纳米化甘露聚糖番茄红素液态饮、纳米化右旋糖苷番茄红素液态饮三类样本，分别构建番茄红素、可溶性固形物、酸度及含水量等关键指标含量的快速定量分析模型。结合模型决定系数（Coefficient of Determination, R2）、均方根误差（Root Mean Square Error, RMSE）、预测残差平方和（Residual Predictive Deviation, RPD）及相对误差（Relative Error Ratio, RER）综合评估模型性能。 结果 经组合预处理优化后，三类纳米化番茄红素液态饮品的多指标定量模型均呈现良好预测性能。番茄红素液态原液模型整体最优，测试集决定系数R2最高达0.9796，RMSE为0.9279，RPD为5.3659，RER为18.3682；纳米化甘露聚糖修饰样品模型测试集R2为0.9085，RPD为4.7333；纳米化右旋糖苷修饰样品模型测试集R2为0.9434，RPD为3.4823。所有模型R2均大于0.90、RPD均大于3.0，说明模型预测精度高，可实现番茄红素、可溶性固形物、酸度、含水量多成分同步快速定量。结论? 综上，基于NIRS结合化学计量学手段可有效检测纳米化番茄红素液态饮品中的多个关键指标，能满足实际生产中的快速检测需求，为该类产品的质量控制提供高效技术支持。 |
| 英文摘要: |
| ABSTRACT:Objective To establish rapid prediction models for multiple nutritional component indexes in nanostructured lycopene liquid beverages by using near-infrared spectroscopy (NIRS) combined with chemometrics. Methods NIRS coupled with partial least squares regression (PLSR) was adopted, and a combined pretreatment optimization strategy was proposed. Rapid quantitative analysis models were constructed separately for key indexes including lycopene, soluble solids, acidity and moisture content in three types of samples: lycopene liquid stock solution, nanostructured mannan-modified lycopene liquid beverage, and nanostructured dextran-modified lycopene liquid beverage. Model performance was comprehensively evaluated by coefficient of determination (R2), root mean square error (RMSE), residual predictive deviation (RPD) and relative error ratio (RER). Results After optimized by combined pretreatment, the multi-index quantitative models of the three kinds of nanostructured lycopene liquid beverages all showed favorable predictive performance. The model of lycopene liquid stock solution was the best overall, with the highest test set R2 of 0.9796, RMSE of 0.9279, RPD of 5.3659 and RER of 18.3682; the test set R2 and RPD of the model for nanostructured mannan-modified samples were 0.9085 and 4.7333, respectively; the test set R2 and RPD of the model for nanostructured dextran-modified samples were 0.9434 and 3.4823, respectively. All models had R2 greater than 0.90 and RPD greater than 3.0, with high prediction accuracy and strong stability, which could realize the simultaneous and rapid quantification of lycopene, soluble solids, acidity and moisture content. Conclusion In summary, NIRS combined with chemometrics can effectively detect multiple key indexes in nanostructured lycopene liquid beverages, meet the demand for rapid detection in actual production, and provide efficient technical support for the quality control of such products. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
