摘要
利用近红外光谱技术建立甘薯块根淀粉和糖类化合物含量的预测模型,实现其快速鉴定与分析。以218份不同类型甘薯块根样品为定标样品材料,30份作为验证样品材料,运用各种光谱数据预处理技术和偏最小二乘法(PLS)建立甘薯块根淀粉、蔗糖、葡萄糖和果糖等指标的预测模型。结果显示,建立的淀粉、蔗糖、葡萄糖和果糖预测模型校正决定系数(R2cal)分别为0.998、0.992、0.994和0.993,内部交叉验证决定系数(R2CV)分别为0.997、0.991、0.990和0.994,验证样品预测值与化学测定值相关系数达0.990以上。表明该类模型具有较好检测性能,在最佳光谱区范围内检测的数据是可靠的,可用于甘薯块根淀粉和糖类化合物含量的快速预测。
A rapid predictive method based on near-infrared spectroscopy （NIRS） was introduced to predict sweet potato root starch and sugar contents. 218 sweet potato root samples were selected as the calibration and 30 as validation sample materials from the National Sweet potato Germplasm Bank in Xuzhou. Some techniques for pre-treating data and par- tial least square（PLS） as a statistical method were used to establish the NIRS models for measuring the contents of starch, sucrose,glucose and fructose. The results revealed that the R2cal of NIRS models for starch, sucrose, glucose and fructose were 0. 998, 0.992, 0.994 and 0.993, respectively, and R2CV were 0.997, 0.991 , 0. 990 and 0. 994, respectively. The correlation coefficients between predicted value and measured value of 30 sweet potato samples were more than 0. 99, indi- cating that the models had good performance and reliabity, and could be used to fast and correctly predict the contents of starch and sugar compounds in sweetpotato root.
出处
《江苏农业学报》
CSCD
北大核心
2013年第6期1260-1265,共6页
Jiangsu Journal of Agricultural Sciences
基金
科技部"863"计划项目(2012AA101204)
农业部国家甘薯现代产业体系建设项目(CARS-11-07B)
关键词
近红外光谱
甘薯块根
淀粉
糖类化合物
偏最小二乘法
near-infrared spectroscopy （ NIRS ）
sweet potato root
starch
sugar
partial least square（PLS）
