抹茶品质指标的可见近红外光谱检测研究

Detection of quality indicators in matcha using visible-near infrared spectroscopy

中文关键词:  可见近红外光谱,抹茶,化学计量学,变量筛选,无损检测

英文关键词:visible-near infrared spectroscopy  matcha, chemometrics  variables selection  nondestructive detection

中文摘要:

      目的 建立适用于抹茶品质的可见近红外(visible-near infrared, Vis-NIR)光谱快速无损检测模型以实现多种品质指标的定量分析。方法 通过Vis-NIR获取抹茶样本的光谱数据, 使用一阶导数(first derivative, 1st)光谱预处理方法, 最后采用自助软收缩法(bootstrapping soft shrinkage, BOSS)、迭代变量子集优化法(iterative variable subset optimization, IVSO)和竞争性自适应重加权采样法(competitive adaptive reweighted sampling, CARS)筛选光谱特征变量, 构建抹茶品质指标的偏最小二乘(partial least square, PLS)预测模型, 探究光谱信息与茶多酚、游离氨基酸、酚氨比、咖啡碱和可溶性糖之间的定量关系。结果 构建的Vis-NIR的CARS-PLS预测模型在抹茶品质指标含量预测方面均获得了最佳结果, 预测相关系数(correlation coefficient in the prediction set, Rp)分别为0.9227、0.8906、0.9243、0.9381和0.9522; 预测均方根误差(root mean square error in the prediction set, RMSEP)分别为0.867、0.337、0.557、0.216和0.440。结论 本研究采用的Vis-NIR光谱技术综合了可见光、短波近红外和长波近红外的优势, 在快速无损预测多种抹茶品质指标方面具有良好应用潜力, 为抹茶品质的快速无损高效检测提供理论依据和技术支撑。

英文摘要:

      Objective To establish a rapid non-destructive detection model of matcha quality based on visible-near infrared (Vis-NIR) spectroscopy for the assessment of matcha quality, enabling quantitative analysis of multiple quality indicators. Methods The spectral data of matcha samples were used to Vis-NIR, the spectral were preprocessed by first derivative (1st), and then bootstrapping soft shrinkage (BOSS), iterative variable subset optimization (IVSO) and competitive adaptive reweighted sampling (CARS) feature variable to screen the spectral characteristic variables, and construct partial least square (PLS) model to explore the relationship between spectral information and tea polyphenols, free amino acids, phenol-ammonia ratio and caffeine and soluble sugars. Results The CARS-PLS prediction models constructed by Vis-NIR obtained the best results in predicting matcha quality indicators with correlation coefficient in the prediction set (Rp) of 0.9227, 0.8906, 0.9243, 0.9381 and 0.9522, respectively, and the root mean square error in the prediction set (RMSEP) was 0.867, 0.337, 0.557, 0.216 and 0.440, respectively. Conclusion This study combines the advantages of visible light, short-wave NIR and long-wave NIR, which has good potential for rapid and nondestructive prediction of a variety of matcha quality indicators, and provides a theoretical basis and technical support for rapid and nondestructive and efficient testing of matcha quality.