摘要
采用凯氏定氮法测试了216份小麦子粒的粗蛋白含量,用近红外仪采集数据,选择113份建立了数学模型。结果:最佳主成分数(Rank)=6,内部交叉验证均方差(RMSECV)=0.377,决定系数(R2)=96.89。为了验证模型的可靠性,对预测集样品进行预测,结果粗蛋白的预测均方差(RMSEP)=0.950,相对偏差(RSEP,%)=8.40。
Bruker fourier transform near-infrared spectroscopy （FT-NIRS） used 113 seeds to establish math model of crude protein. The result on this model showed that the Rank was 6 and the RMSECV was 0.377 and R2 was 96.89. 54 samples were used to test this model,The result on model validation showed that the RMSEP was 0.950 and the RSEP（%） was 8.40.
出处
《安徽农学通报》
2015年第3期22-23,共2页
Anhui Agricultural Science Bulletin
基金
国家小麦产业技术体系绵阳综合试验站(CARS-3-2-40)
强优势杂交种绵杂麦168示范推广与产业化(12CGZHZX0753)
四川省"十二五"育种攻关资助项目(2011YZGG-3)
作者简介
雷加容（1969-），女，硕士，高级农艺师，从事小麦品质分析和栽培研究工作。
