论文部分内容阅读
采用偏最小二乘(PLS)算法结合衰减全反射(ATR)校正提取近红外光谱的特征信息,建立红茶含水率、茶多酚、游离氨基酸和酚氨比的预测模型。与原始光谱建模相比,在保证预测精度的前提下,采纳的主成分数更少,降低了预测模型的复杂性。红茶含水率、茶多酚、游离氨基酸和酚氨比预测集相关系数(R_p)和预测均方差(RMSEP)分别为0.994、0.960、0.944、0.922和0.182、0.523、0.184、0.556。通过配对双边t-检验,该方法与标准法测定结果无显著性差异。结果表明,偏最小二乘法结合ATR校正能够有效地简化红茶含水率、茶多酚、游离氨基酸和酚氨比的预测模型,实现红茶品质成分的快速检测,为评估红茶质量优劣提供了新的方法。
Partial least squares (PLS) algorithm combined with attenuated total reflectance (ATR) was used to extract the characteristic information of near infrared spectroscopy to establish a prediction model of moisture content, tea polyphenols, free amino acids and phenol to ammonia ratio in black tea. Compared with the original spectral modeling, the number of principal components adopted is smaller and the complexity of the prediction model is reduced, while ensuring the prediction accuracy. The correlation coefficient (R_p) and predicted root mean square error (RMSEP) of black tea moisture content, tea polyphenols, free amino acids and phenols ammonia were 0.994,0.960,0.944,0.922 and 0.182,0.523,0.184,0.556, respectively. By paired bilateral t-test, there is no significant difference between this method and the standard method. The results showed that partial least square method combined with ATR correction could effectively simplify the prediction model of water content, tea polyphenol, free amino acid and phenol to ammonia ratio in black tea and achieve the rapid detection of the quality components of black tea, thus providing a new method for assessing the quality of black tea method.