登录    注册    忘记密码

详细信息

近红外漫反射光谱结合偏最小二乘法对紫胶理化指标的快速测定     被引量:4

Rapid determination of physicochemical indexes in shellac using near infrared diffuse reflectance spectroscopy combined with PLS algorithm

文献类型:期刊文献

中文题名:近红外漫反射光谱结合偏最小二乘法对紫胶理化指标的快速测定

英文题名:Rapid determination of physicochemical indexes in shellac using near infrared diffuse reflectance spectroscopy combined with PLS algorithm

作者:唐保山[1,2] 李坤[1] 张雯雯[1] 史正军[2] 关庆芳[3] 徐涓[1] 马金菊[1] 刘兰香[1] 张弘[1]

第一作者:唐保山

机构:[1]中国林业科学研究院资源昆虫研究所,国家林业和草原局特色森林资源工程技术研究中心,云南昆明650233;[2]西南林业大学林学院,云南昆明650224;[3]安宁戴科精细化工有限公司,云南昆明650301

年份:2020

卷号:46

期号:18

起止页码:236-244

中文期刊名:食品与发酵工业

外文期刊名:Food and Fermentation Industries

收录:CSTPCD;;北大核心:【北大核心2017】;CSCD:【CSCD2019_2020】;

基金:中央级公益性科研院所基本科研业务费专项资金(CAFYBB2018SY024,CAFYBB2017MA012)。

语种:中文

中文关键词:近红外光谱;紫胶;理化指标;快速测定;区间偏最小二乘法

外文关键词:near infrared spectrum;shellac;physicochemical indexes;rapid determination;interval partial least square

分类号:O657.33;TQ317

摘要:采用化学法测定紫胶理化指标的化学值,应用傅里叶变换近红外光谱技术,采集紫胶的近红外光谱并使用光谱预处理方法消除噪声,组合区间偏最小二乘法选择特征波段,采用内部交互验证法筛选主成分数,最后通过偏最小二乘法建立回归模型,最终得到紫胶中灰分、水分、冷乙醇可溶物、热寿命、酸值和颜色指数的近红外光谱定量分析模型。灰分、水分、冷乙醇可溶物、热寿命、酸值和颜色指数校正集校正决定系数分别为0.968、0.982、0.945、0.821、0.873和0.946,交叉验证标准误差分别为0.054、0.081、1.050、0.359、1.230和1.880;验证集的决定系数分别为0.958、0.981、0.904、0.810、0.872和0.930,预测标准误差分别为0.039、0.039、0.039、0.234、0.700和0.618;相对分析误差值分别为5.58、7.65、3.30、2.51、2.82和4.31。结果表明,近红外光谱法对热寿命和酸值进行定量分析是可行的,但其精度有待进一步提高。对于紫胶中灰分、水分、冷乙醇可溶物和颜色指数,内部交叉验证和外部验证均证明,建立的近红外定量分析模型的准确度和预测性能良好,为紫胶理化指标的快速分析方法的研究提供了新的参考。
Chemical methods were used to determine the chemical values of physicochemical properties of shellac.The application of Fourier transform near-infrared spectrum technology was performed to collect near infrared spectra of shellac and the spectral noise eliminated by pretreatment method.And synergy interval partial least(SIPLS)was employed to choose the characteristics of the band and the internal interaction validation method screening principal components number.Finally,the index of physicochemical properties including ash,moisture,cold alcohol soluble,thermal life,acid value and color index of near infrared spectrum quantitative analysis model was established by partial least squares(PLS)regression algorithm.The results showed that the correction determination coefficients(R 2 c)of ash,moisture,cold alcohol soluble,thermal life,acid value and color index were 0.968,0.982,0.945,0.821,0.873 and 0.946,respectively.And the root mean square error of cross validation(RMSECV)was 0.054,0.081,1.050,0.359,1.230 and 1.880,respectively.Moreover,the determination coefficient(R 2 p)of validation sets were 0.958,0.981,0.904,0.810,0.872 and 0.930 for these indexes,and the prediction standard errors(RMSEP)were 0.039,0.039,0.039,0.234,0.700 and 0.618,respectively.Besides,the relative percent deviation(RPD)values were 5.58,7.65,3.30,2.51,2.82 and 4.31.Therefore,the near infrared spectroscopy was feasible for the quantitative analysis of thermal life and acid value.But its precision needs to be further improved.For ash,moisture content,cold alcohol soluble and color index of shellac,internal cross validation and external validation sets all proved that the accuracy and prediction performance of the established near-infrared quantitative analysis model were good,which provided a new reference for the research of rapid analysis method of physical and chemical indexes of shellac.

参考文献:

正在载入数据...

版权所有©中国林业科学研究院 重庆维普资讯有限公司 渝B2-20050021-8 
渝公网安备 50019002500408号 违法和不良信息举报中心