玉米子粒粗蛋白近红外模型的建立  

Establishment of Near Infrared Model of Corn Kernel Crude Protein

作　　者：于翠红[1] 樊翠芹[1] 张丽[1] 徐俊杰[1] 刘朝芳[1] 蒋春志[1] 唐晓东[1] YU Cui-hong;FAN Cui-qin;ZHANG Li;XU Jun-jie;LIU Chao-fang;JIANG Chun-zhi;TANG Xiao-dong(Institute of Cereal and Oil Crops of Hebei Academy of Agriculture and Forestry Sciences,Crop Genetics and Breeding Laboratory of Hebei Province,Shijiazhuang 050035,China)

机构地区：[1]河北省农林科学院粮油作物研究所河北省作物遗传育种实验室

出　　处：《河北农业科学》2019年第6期101-105,共5页Journal of Hebei Agricultural Sciences

摘　　要：为找到一种快速、无破坏测定玉米子粒品质性状的方法,以适合黄淮海地区种植的具有代表性的598份普通玉米杂交种为试验材料,按照GB 5009.5-2016 (食品中蛋白质的测定)利用凯氏定氮仪测定粗蛋白含量,利用DA7200型近红外光谱仪采用近红外反射光谱法(NIRS)测定玉米子粒光谱值,采用偏最小二乘(PLS)回归法建立玉米子粒粗蛋白定标模型。对该模型进行预测准确性验证,定标集和验证集的预测值与化学测定值达到极显著正相关,相关系数分别为0.941 3和0.972 1,并具有较小的定标标准差(0.358 7)和预测标准差(0.313 7),所建定标模型的预测效果与真值具有较高的一致性。利用所建的近红外模型测定玉米子粒粗蛋白含量是可行的。该模型的建立可实现玉米品质育种过程中早代材料的快速、无破坏大批量检测,提高育种效率。To find a rapid,nondestructive determination method of corn kernel quality traits,598 normal corn hybrids suitable for cultivation in Huang-Huai-Hai area were used as the experimental materials,the content of crude protein was determined by Kjeldahl nitrogen determinator according to GB 5009.5-2016(Determination of Protein in Food),DA7200 near infrared spectrometer was used to determine the spectral value of corn kernel by near infrared reflectance spectroscopy(NIRS),and the partial least squares(PLS) regression method was used to establish the crude protein calibration model for corn kernel. The prediction accuracy of the model was verified,and there was remarkable positive correlation between the predicted values and the chemical values.The correlation coefficients to chemical values in calibration set and validation set were 0.941 3 and 0.972 1,respectively. The standard deviation of calibration was 0.358 7,and the deviation of validation was 0.3137,and the prediction effect of the calibration model established was highly consistent with the truth value. It is feasible to determine the crude protein content of corn kernel by this model. The establishment of this model can realize the rapid and nondestructive mass detection of early generation materials in corn quality breeding and improve the breeding efficiency.

关 键 词：玉米 粗蛋白 近红外光谱 定标模型 无损检测 偏最小二乘回归 

分 类 号：S513[农业科学—作物学]