【打印本页】   【下载PDF全文】   查看/发表评论  下载PDF阅读器  关闭
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 3239次   下载 0
分享到: 微信 更多
采用近红外光谱快速测定小米硒含量
王浩, 于港华, 侯颖, 侯思宇, 韩渊怀, 李红英, Guofang
山西农业大学农学院
摘要:
谷子是山西主栽杂粮作物之一,在自然条件下具有良好的富硒能力。谷子籽粒硒含量是评价谷子富硒能力强弱的重要指标,然而目前籽粒硒含量的测定需要依赖国标法,检测过程繁长冗杂,污染大,近红外光谱技术具有分析速度快,适用性广,样品无需前处理、不用试剂、对环境友好、操作简单、成本低等优点,还可实现全自动操作、减少人为误差,具有较高的精密度和重现性,因此,利用近红外光谱技术快速检测谷粒中的硒含量具有重要的现实意义。本实验以73个脱壳谷子样品为研究对象,采用国标法测定其籽粒总硒含量,采用德国布鲁克光谱仪器公司生产的MPA傅立叶变换近红外光谱仪采集光谱信息。将样品分为校正建模集和验证集,其中校正集51个,验证集22个,通过变量标准化SNV和Detrend光谱预处理方法和PLSR(偏最小二乘法)建模方法建立脱壳谷子-小米总硒含量的定量模型,用Workflow调用模型来实现小米总硒含量的快速检测,结果表明,通过PLSR建立模型具有较高的预测精度,小米总硒含量的内部交叉验证的相关系数为84.5%,RMSEC、RMSEP(均方根误差)分别为0.0396和0.0892,均接近0,说明小米总硒含量的近红外预测值接近化学测定值,本研究中RPD为5.478,大于AACC、ICC等国际分析组织提出的质量控制,本实验中预测集和建模集标准误差的比值(SEP/SEC,代表模型的稳健程度,一般SEP/SEC≤1.2)为1.0730,,小于1.2,因此所建立的模型具有较好的稳健程度。可实现对小米总硒含量的快速检测。
关键词:  近红外光谱分析技术  定量模型  快速检测  小米总硒含量
DOI:
分类号:
基金项目:山西省重点研发计划项目(201803D221008-4),国家自然科学基金资助项目(32070366),晋中市重点研发计划项目(Y192011)
Quantitative and rapid determination of selenium content in millet using Near Infrared Spectroscopy
wanghao, yuganghua, houying, sousiyu, hanyuanhuai, lihongying, Guofang
Shanxi Agricultural University
Abstract:
Foxtail millet is one of the minor crops in Shanxi, and it has good selenium-enrichment capacity under natural conditions. The content of selenium in grains is an important indicator to evaluate the selenium-enrichment ability of foxtail millet. A new method based on Near Infrared Spectroscopy (NIRS) was proposed to identify the selenium content of the seeds, in order to provide technical support for foxtail millet quality improvement breeding. In this experiment, 73 naked millet samples were used as the research object. The total selenium contents were measured by the national standard method. The MPA Fourier Transform Near-infrared Spectrometer from the German Bruker Spectrometer Company was used to collect the spectral information. The 73 samples were divided into calibration modeling set and validation set, of which 51 were calibration set and 22 were validation set. The total amount of millet seed was established through variable standardization SNV and Detrend spectral preprocessing method, and PLSR (Partial Least Squares) modeling method, quantitative model of selenium content, and Workflow was used to establish the model to detect the total selenium content of millet powder rapidly. The results showed that the model established by PLSR has high prediction accuracy. The internal cross-validation correlation coefficient of the total selenium content of millet powder is high to 84.5%; RMSEC and RMSEP (root mean square error) are both low, indicating that the near-infrared predicted value of the total selenium content of millet powder is close to the chemical measurement value. In this study, the RPD is 5.478 (The detection limit), which meets the acceptable quality proposed by international analysis organizations such as AACC and ICC Control, SEP/SEC reflects the robustness of the model, and SEP/SEC≤1.2. In this experiment, SEP/SEC is 1.0730, which is less than 1.2, so the established model has a good degree of robustness. So, the rapid detection of the total selenium content of millet we established in this experimentwas necessary, feasible and effectivein selenium content measurement.for foxtail millet.
Key words:  Near- infrared Spectroscopy Analysis Technology  quantitative model  rapid detection  total selenium content of millet
引用本文: