不同偏最小二乘法的近红外光谱技术测定大米中水分的研究  

Determination of Moisture Content in Rice by Near Infrared Spectroscopy with Different Partial Least Squares

作　　者：苗雪雪[1] 苗莹 龚浩如[1] 陶曙华[1] 陈英姿[1] 陈祖武[1] MIAO Xue-xue;MIAO Ying;GONG Hao-ru;TAO Shu-hua;CHEN Ying-zi;CHEN Zu-wu(Hunan Provincial Rice Research Institute, Hunan Academy of Agricultural Sciences,Key Laboratory of Indica Rice Genetics and Breeding in the Middle and Lower Reachesof Yangtze River Valley, Ministry of Agriculture, Changsha 410125;College of Mathematics and Informatics, South China Agricultural University,Guangzhou 510642)

机构地区：[1]湖南省农业科学院水稻研究所农业部长江中下游籼稻遗传育种重点实验室,湖南长沙410125 [2]华南农业大学数学与信息学院,广东广州510642

出　　处：《分析科学学报》2019年第5期643-649,共7页Journal of Analytical Science

基　　金：湖南省农业科学院科技创新项目(No.2017QN06);“十三五”国家重点研发计划(No.2018YFD0301004)

摘　　要：通过近红外光谱法对大米中含水量进行分析,运用Kennard-Stone法对校正集及预测集样本进行选取,利用Range Normalization法、二阶导数和多元散射校正加一阶导数法,分别对近红外光谱进行预处理优化,并采用偏最小二乘法(PLS)、组合区间偏最小二乘法(SiPLS)和移动窗口偏最小二乘法(MWPLS)分别建立了定标模型。结果显示,相较于全谱建模,2种变量优选方法都能在有效减少建模所用的变量数,同时提高模型性能。其中采用MWPLS优选变量所建的大米水分定量模型的性能最优,决定系数为0.9525,校正集均方根误差为0.4093。利用40个验证集样本对定标模型进行了验证和配对t检验,预测相关系数达0.9617,相对分析误差为3.64,模型预测值与标准方法测定值没有显著性差异,说明模型具有良好的预测能力。基于MWPLS的近红外光谱技术能够实现大米中水分含量的快速检测。The near-infrared(NIR) spectrum was used to detect the moisture content in rice samples.The calibration set and prediction set of samples were selected by using Kennard-Stonemethod, then the range normalization method, second derivative method and MSC+FD method were used as pre-treatment method for raw spectrum before establishing the calibration models by using partial least square(PLS), synergy interval partial least square(SiPLS) and moving window partial least square(MWPLS) methods. Results showed that compared with the model built by full spectrum, the two methods were effective in reducing the variable numbers and improving the performance of models. Between them, the optimal model was established by MWPLS. The correlation coefficient and root mean square error of cross validation(RMSECV) of the model were 0.9525 and 0.4093, respectively. 40 samples were used to forecast the moisture contents of rice, and the correlation coefficient in prediction set was 0.9617,the relative prediction deviation(RPD) was 3.64. There was no significant difference between the standard method and NIR quantitative analysis in the experiment of paired t-test. In a word, this rapid, accurate, loss less and eco-friendly method can be applied to the analysis of moisture content in rice.

关 键 词：近红外光谱 大米 水分 组合区间偏最小二乘法 移动窗口偏最小二乘法 

分 类 号：O657.33[理学—分析化学]