资源描述:
《棉子油分近红外光谱测定技术研究》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、中国农学通报2014,30(3):173-177ChineseAgriculturalScienceBulletin棉子油分近红外光谱测定技术研究韩智彪,郭小平(作物遗传国家重点实验室/华中农业大学植物科技学院,武汉430070)摘要:以近红外光谱分析技术无损、快速和高效等优点,建立棉籽油分含量的近红外分析模型,进行棉籽油分快速大批测定,提高分析水平和效率。218份陆地棉的脱绒光籽为定标样品,通过不同的光谱数学预处理和光谱反射方式,优化棉子含油量近红外光谱法所建立的回归方程。结果表明,采用(2,4,4,1)光谱数学预处理和散射校正(SN
2、V)得到的回归方程效果最好,回归决定系数(RSQ=0.9773)和验证决定系数(1-VR=0.9646)最高,回归标准误差(SEC=0.6977)和交互验证标准误差(SECV=0.8714)最小。对30份验证样品的化学测定值和近红外光谱预测值相关系数高达0.9882,平均偏差0.65%,回归标方程具有很好的预测效果,可以在实践中应用。长江流域棉花杂交种棉子含油量平均值为28.05%,含油量在26.50%~30.05%区间的品种居多,具有很大的改良潜力。关键词:棉子;近红外光谱;含油量;回归方程中图分类号:S562文献标志码:A论文编号:
3、2013-0505MeasurementTechniqueofOilContentinCottonseedbyNearInfraredReflectanceSpectroscopyHanZhibiao,GuoXiaoping(NationalKeyLaboratoryofCropGeneticImprovement/CollegeofPlantScienceandTechnology,HuazhongAgriculturalUniversity,Wuhan430070)Abstract:Thenear-infraredspectrosc
4、opyanalysismodelofcottonseedoilcontenthasbeensetup,whichisnon-destructive,fastandefficient,toassaytheoilcontentofcottonseedinlargequantitiesfasterandtoimprovethestandardandefficiencyofthisanalysis.Abetterregressionequationwasestablishedbyusingdifferentspectralmathematica
5、lpretreatmentandspectralreflectancemode,with218uplandcottonseedsastestsamplewhichisnakedcotton,abetterregressionequationforoilcontentbynearinfraredspectroscopy(NIRS)wasestablishedthroughthemethodsofspectramathematicpretreatmentanddifferentspectrareflectiontreatment.There
6、sultsshowedthebestregressionequationwasobtainedbythemathematicspretreatment“2,4,4,1”forspectroscopicdata,andtransformationofStandardNormalVariant(SNV).Ithadthehighestregressionsquaredofthecalibration(RSQ=0.9773)andverificationsquaredofthecalibration(1-VR=0.9646),andthesa
7、metime,ithadtheleastregressionstandarderrorofthecalibration(SEC=0.6977),andtheleastcross-validationstandarderrorofthecalibration(SECV=0.8714).Therewasacorrelationcoefficient0.9882betweenthedataofoilcontentobtainedbythechemicalandNIRSmethodby30verifiedsamples,onlywithadev
8、iationof0.65%.TheregressionequationbyNIRScouldmeasuretheoilcontentofcottonseedaccurately;itcouldbeappli