基于多类别的镉稻米近红外光谱识别分析
|
摘要
采用近红外(NIR)光谱结合化学计量学方法对不同镉污染程度的稻米进行鉴别。首先利用主成分分析(PCA)对样本的NIR光谱进行解析,再用有监督学习算法偏最小二乘识别分析(PLS-DA)、径向基人工神经网络(RBF-ANN)及支持向量机(SVM)对不同污染程度的镉稻米进行定性建模分析。本文还讨论了不同的光谱预处理方法以及建模方法对识别效果的影响。由于NIR光谱差异太小,所以PCA得分图重叠严重,类之间很难区分,PLS-DA、RBF-ANN与SVM模型的预测集鉴别准确率分别为77.1%,67.8%与67.2%,PLS-DA的识别率最高。近红外光谱技术与化学计量学方法虽难以获得预测准确率较高的识别模型,但其预测结果还是可用于超标镉稻米的初步筛查。
Near infrared(NIR) spectroscopy combined with chemometrics methods were used to discriminate different level of cadmium-polluted rice. The original spectra data were built qualitative models by means of unsupervised learning algorithm such as partial component analysis(PCA) and supervised learning recognition analysis including partial least squares discriminant analysis(PLS-DA), radical basis function-artificial neural network(RBF-ANN) and support vector machine(SVM). The optimization of spectral pretreatment methods and the comparison of the performance of different modeling methods were discussed. For different levels of cadmium-polluted rice, since the differences of near infrared spectra were small, which caused PCA score plots to overlap seriously. The discriminate rates of prediction set for PLSDA, RBF-ANN and SVM were 77.1%, 68.5% and 67.2%, respectively. The better results could be obtained by PLSDA. The results showed that it was difficult to obtain satisfactory accuracy rate for screening different levels of cadmiumpolluted rice by NIR spectroscopy and pattern recognition techniques based on multiple classifiers. However, it is possible to obtain acceptable classification models.
Near infrared(NIR) spectroscopy combined with chemometrics methods were used to discriminate different level of cadmium-polluted rice. The original spectra data were built qualitative models by means of unsupervised learning algorithm such as partial component analysis(PCA) and supervised learning recognition analysis including partial least squares discriminant analysis(PLS-DA), radical basis function-artificial neural network(RBF-ANN) and support vector machine(SVM). The optimization of spectral pretreatment methods and the comparison of the performance of different modeling methods were discussed. For different levels of cadmium-polluted rice, since the differences of near infrared spectra were small, which caused PCA score plots to overlap seriously. The discriminate rates of prediction set for PLSDA, RBF-ANN and SVM were 77.1%, 68.5% and 67.2%, respectively. The better results could be obtained by PLSDA. The results showed that it was difficult to obtain satisfactory accuracy rate for screening different levels of cadmiumpolluted rice by NIR spectroscopy and pattern recognition techniques based on multiple classifiers. However, it is possible to obtain acceptable classification models.
引文
[1]GUNDUZ S,SULEYMAN A.Investigation of arsenic and cadmium contents in rice samples in turkey by electrothermal atomic absorption spectrometry[J].Food Analytical Methods,2013,6(6):1693-1696.
[2]ROMKENS P F,GUO H Y,CHU C L,et al.Prediction of Cadmium uptake by brown rice and derivation of soil-plant transfer models to improve soil protection guidelines[J].Environmental Pollution,2009,157(8/9):2435-2444.
[3]ZHANG W B,YANG X A,XUE J J.Green method for the determination of Cd in rice and water samples based on electrolytic hydride generation and atomic fluorescence spectrometry[J].Journal of Analytical Atomic Spectrometry,2012,27(6):928-936.
[4]毕京翠,张文伟,肖应辉,等.应用近红外光谱技术分析稻米蛋白质含量[J].作物学报,2006,32(5):709-715.
[5]周子立,张瑜,何勇,等.基于近红外光谱技术的大米品种快速鉴别方法[J].农业工程学报,2009,25(8):131-135.
[6]唐绍清,石春海,焦桂爱,等.利用近红外反射光谱技术测定稻米中脂肪含量的研究初报[J].中国水稻科学,2004,18(6):563-566.
[7]邵咏妮,曹芳,何勇.基于独立组分分析和BP神经网络的可见/近红外光谱稻谷年份的鉴别[J].红外与毫米波学报,2007,26(6):433-436.
[8]KUMAGAI M,OHISA N,AMANO T,et al.Canonical discriminant analysis of cadmium content levels in unpolished rice using a portable near-infrared spectrometer[J].Analytical Science,2003,19(11):1553-1555.
[9]FONT R,VELEZ D,DEL RIO-CELESTINO M,et al.Screening inorganic arsenic in rice by visible and near-infrared spectroscopy[J].Microchimica Acta.2005,151(3/4):231-239.
[10]GONZALEZ-MARTIIN I,GONZALEZ-PEREZ C,HERNANDEZ-MENDEZ J,et al.Mineral analysis(Fe,Zn,Ca,Na,K)of fresh Iberian pork loin by near infrared reflectance spectrometry:Determination of Fe,Na and K with a remote fibre-optic reflectance probe[J].Analytica Chimica Acta,2002,468(2):293-301.
[11]FONT R,DEL RIO-CELESTINO M,VELEZ D,et al.Visible and near-infrared spectroscopy as a technique for screening the inorganic arsenic content in the red crayfish(Procambarus clarkii Girard)[J].Analytical Chemistry,2004,76(14):3893-3898.
[12]WU D,HE Y,SHI J H.Exploring near and midinfrared spectroscopy to predict trace iron and zinc contents in powdered milk[J].Journal of Agricultural and Food Chemistry,2009,57(5):1697-1704.
[13]FONT R,DEL RIO-CELESTINO M,VELEZ D,et al.Use of near-infrared spectroscopy for determining the total arsenic content in prostrate amaranth[J].Science of the Total Environment,2004,327(1/2/3):93-104.
[14]COZZOLINO D,MORON A.Exploring the use of near infrared reflectance spectroscopy(NIRS)to predict trace minerals in legumes[J].Animal Feed Science and Technology,2004,111(1/2/3/4):161-173.
[15]MOROS J,LLORCA I,CERVERA M L,et al.Chemometric determination of arsenic and lead in untreated powdered red paprika by diffuse reflectance near-infrared spectroscopy[J].Analytica Chimica Acta,2008,613(2):196-206.
[16]MARTINEZ-VALDIVIESO D,FONT R,GOMEZ P,et al.Determining the mineral composition in Cucurbita pepo fruit using near infrared reflectance spectroscopy[J].Journal of the Science of Food and Agriculture,2014,94(15):3171-3180.
[17]张龙,潘家荣,朱诚.基于近红外光谱的重金属汞、镉和铅污染水稻叶片鉴别[J].浙江大学学报(农业与生命科学版),2013,39(1):50-55.
[18]OUYANG A G,JIANG L X,LIU Y D,et al.Determination of copper and zinc pollutants in ludwigia prostrata roxb using near infrared reflectance spectroscopy[J].Applied Spectroscopy,2015,69(3):370-376
[19]刘燕德,施宇.近红外光谱快速检测香根草叶片铅含量[J].农业机械学报,2014,45(3):232-236.
[20]SHENG N,CAI W S,SHAO X G.An approach by using near-infrared diffuse reflectance spectroscopy and resin adsorption for the determination of copper,cobalt and nickel ions in dilute solution[J].Talanta,2009,79(2):339-343.
[21]SHAO X G,SHENG N,CAI W S.Quantitative analysis of chromium(VI)in dilute solution by using adsorption and diffuse reflectance near-infrared spectroscopy[J].Chinese Journal of Chemistry,2010,28(10):2009-2014.
[22]HUANG Z X,TAO W,FANG J J,et al.Multivariate calibration of on-line enrichment near-infrared(NIR)spectra and determination of trace lead in water[J].Chemometrics and Intelligent Laboratory Systems,2009,98(2):195-200.
[23]朱向荣,李高阳,黄绿红,等.近红外光谱与化学计量学方法用于镉污染稻米的定性鉴别[J].分析化学,2015,43(4):599-603.
[24]ZHU X R,LI G Y,SHAN Y.Prediction of Cadmium content in brown rice using near-infrared spectroscopy and regression modelling techniques.International Journal of Food Science and Technology,2015,50(5):1123-1129.
[25]田阳,魏帅,魏益民,等.大米淀粉提取工艺对淀粉产品镉含量的影响[J].中国粮油学报,2013,28(4):83-87.
[26]BALLABIO D,CONSONNI V.Classification tools in chemistry.Part 1:linear models.PLS-DA.Analytical Methods,2013,5(16):3790-3798.
[27]AMARI S,WU S.Improving support vector machine classifiers by modifying kernel functions[J].Neural Networks,1999,12(6):783-789.
[28]KEERTHI S S,LIN C J.Asymptotic behaviors of support vector machines with gaussian kernel[J].Neural Computer,2003,15(7):1667-1689.
[29]LIU H X,XUE C,ZHANG R S,et al.Quantitative prediction of logk of peptides in high-performance liquid chromatography based on molecular descriptors by using the heuristic method and support vector machine[J].Journal of Chemical Information and Computer Scienc,2004,44(6):1979-1986.
[30]葛哲学,孙志强.神经网络理论与MATLAB R2007实现[M].北京:电子工业出版社,2007:117-122,219-225.
[2]ROMKENS P F,GUO H Y,CHU C L,et al.Prediction of Cadmium uptake by brown rice and derivation of soil-plant transfer models to improve soil protection guidelines[J].Environmental Pollution,2009,157(8/9):2435-2444.
[3]ZHANG W B,YANG X A,XUE J J.Green method for the determination of Cd in rice and water samples based on electrolytic hydride generation and atomic fluorescence spectrometry[J].Journal of Analytical Atomic Spectrometry,2012,27(6):928-936.
[4]毕京翠,张文伟,肖应辉,等.应用近红外光谱技术分析稻米蛋白质含量[J].作物学报,2006,32(5):709-715.
[5]周子立,张瑜,何勇,等.基于近红外光谱技术的大米品种快速鉴别方法[J].农业工程学报,2009,25(8):131-135.
[6]唐绍清,石春海,焦桂爱,等.利用近红外反射光谱技术测定稻米中脂肪含量的研究初报[J].中国水稻科学,2004,18(6):563-566.
[7]邵咏妮,曹芳,何勇.基于独立组分分析和BP神经网络的可见/近红外光谱稻谷年份的鉴别[J].红外与毫米波学报,2007,26(6):433-436.
[8]KUMAGAI M,OHISA N,AMANO T,et al.Canonical discriminant analysis of cadmium content levels in unpolished rice using a portable near-infrared spectrometer[J].Analytical Science,2003,19(11):1553-1555.
[9]FONT R,VELEZ D,DEL RIO-CELESTINO M,et al.Screening inorganic arsenic in rice by visible and near-infrared spectroscopy[J].Microchimica Acta.2005,151(3/4):231-239.
[10]GONZALEZ-MARTIIN I,GONZALEZ-PEREZ C,HERNANDEZ-MENDEZ J,et al.Mineral analysis(Fe,Zn,Ca,Na,K)of fresh Iberian pork loin by near infrared reflectance spectrometry:Determination of Fe,Na and K with a remote fibre-optic reflectance probe[J].Analytica Chimica Acta,2002,468(2):293-301.
[11]FONT R,DEL RIO-CELESTINO M,VELEZ D,et al.Visible and near-infrared spectroscopy as a technique for screening the inorganic arsenic content in the red crayfish(Procambarus clarkii Girard)[J].Analytical Chemistry,2004,76(14):3893-3898.
[12]WU D,HE Y,SHI J H.Exploring near and midinfrared spectroscopy to predict trace iron and zinc contents in powdered milk[J].Journal of Agricultural and Food Chemistry,2009,57(5):1697-1704.
[13]FONT R,DEL RIO-CELESTINO M,VELEZ D,et al.Use of near-infrared spectroscopy for determining the total arsenic content in prostrate amaranth[J].Science of the Total Environment,2004,327(1/2/3):93-104.
[14]COZZOLINO D,MORON A.Exploring the use of near infrared reflectance spectroscopy(NIRS)to predict trace minerals in legumes[J].Animal Feed Science and Technology,2004,111(1/2/3/4):161-173.
[15]MOROS J,LLORCA I,CERVERA M L,et al.Chemometric determination of arsenic and lead in untreated powdered red paprika by diffuse reflectance near-infrared spectroscopy[J].Analytica Chimica Acta,2008,613(2):196-206.
[16]MARTINEZ-VALDIVIESO D,FONT R,GOMEZ P,et al.Determining the mineral composition in Cucurbita pepo fruit using near infrared reflectance spectroscopy[J].Journal of the Science of Food and Agriculture,2014,94(15):3171-3180.
[17]张龙,潘家荣,朱诚.基于近红外光谱的重金属汞、镉和铅污染水稻叶片鉴别[J].浙江大学学报(农业与生命科学版),2013,39(1):50-55.
[18]OUYANG A G,JIANG L X,LIU Y D,et al.Determination of copper and zinc pollutants in ludwigia prostrata roxb using near infrared reflectance spectroscopy[J].Applied Spectroscopy,2015,69(3):370-376
[19]刘燕德,施宇.近红外光谱快速检测香根草叶片铅含量[J].农业机械学报,2014,45(3):232-236.
[20]SHENG N,CAI W S,SHAO X G.An approach by using near-infrared diffuse reflectance spectroscopy and resin adsorption for the determination of copper,cobalt and nickel ions in dilute solution[J].Talanta,2009,79(2):339-343.
[21]SHAO X G,SHENG N,CAI W S.Quantitative analysis of chromium(VI)in dilute solution by using adsorption and diffuse reflectance near-infrared spectroscopy[J].Chinese Journal of Chemistry,2010,28(10):2009-2014.
[22]HUANG Z X,TAO W,FANG J J,et al.Multivariate calibration of on-line enrichment near-infrared(NIR)spectra and determination of trace lead in water[J].Chemometrics and Intelligent Laboratory Systems,2009,98(2):195-200.
[23]朱向荣,李高阳,黄绿红,等.近红外光谱与化学计量学方法用于镉污染稻米的定性鉴别[J].分析化学,2015,43(4):599-603.
[24]ZHU X R,LI G Y,SHAN Y.Prediction of Cadmium content in brown rice using near-infrared spectroscopy and regression modelling techniques.International Journal of Food Science and Technology,2015,50(5):1123-1129.
[25]田阳,魏帅,魏益民,等.大米淀粉提取工艺对淀粉产品镉含量的影响[J].中国粮油学报,2013,28(4):83-87.
[26]BALLABIO D,CONSONNI V.Classification tools in chemistry.Part 1:linear models.PLS-DA.Analytical Methods,2013,5(16):3790-3798.
[27]AMARI S,WU S.Improving support vector machine classifiers by modifying kernel functions[J].Neural Networks,1999,12(6):783-789.
[28]KEERTHI S S,LIN C J.Asymptotic behaviors of support vector machines with gaussian kernel[J].Neural Computer,2003,15(7):1667-1689.
[29]LIU H X,XUE C,ZHANG R S,et al.Quantitative prediction of logk of peptides in high-performance liquid chromatography based on molecular descriptors by using the heuristic method and support vector machine[J].Journal of Chemical Information and Computer Scienc,2004,44(6):1979-1986.
[30]葛哲学,孙志强.神经网络理论与MATLAB R2007实现[M].北京:电子工业出版社,2007:117-122,219-225.