基于近红外光谱的红茶干燥中含水率无损检测方法

摘要

Moisture is an important index of tea drying effect and quality. To understand rapid detection of moisture in black tea, a nondestructive testing method was proposed based on near infrared spectroscopy (NIR). The diffuse reflectance spectra of 226 tea samples were scanned in the range of 1000-1799 nm. These samples were from 6 drying processes. Moisture contents of samples were immediately measured after spectral scanning. The original spectrum data were proposed by the Standard Normal Variate Transformation (SNVT). Two regression algorithms including Partial Least Square (PLS) and Synergy Interval Partial Least Square (siPLS) were used to develop models for determination of moisture contents respectively. The result showed that both models had high accuracy, but the performance of model with siPLS was better. It contained 13 spectral intervals combined with 4 subinterval and 6 principal component factors. The root mean square for prediction (RMSEP) and the correlation coefficient (Rp) reached 0.0395 and 0.9593, respectively. It showed that it is feasible to measure moisture content of black tea during drying process.%为实现工夫红茶干燥中含水率的快速检测,提出了基于近红外光谱红茶干燥中含水率无损检测方法.随机抽取6次干燥处理中的226个样本,进行波长1000~1799 nm近红外光谱扫描后按照国标法测定含水率.对原始光谱数据进行标准正态变量变换(SNVT)预处理,利用全局偏最小二乘法(PLS)、联合区间偏最小二乘法(siPLS),分别构建水分近红外预测模型并验证.结果表明:用两种方法检测含水率,其准确度都可靠,但利用siPLS法将全光谱划分为13个区间,联合4个区间用6个主成分数构建的水分预测模型效果更优,其预测集的相关系数R和预测均方根误差RMSEP值分别为0.9593和0.0395,说明模型预测精度高,可以实现红茶干燥中含水率的快速无损检测.