稻米蛋白质含量快速活体测定的定标研究

Equation development of protein content in living rice

应用近红外透射光谱技术（NITS），采用偏最小二乘法（PLS）建立重庆地区稻米活体蛋白质含量（PC）定量分析数学模型。结果表明，糙米和精米蛋白质含量预测数学模型的定标标准误偏差（SEC）、交叉检验标准误差（SECV）、定标相关系数（RSQ）和交叉验证相关系数（1-VR）分别为0．252、0．247；0．256、0．278；-0．953、0．946；0.951、0.940；近红外预测值与化学值误差范围分别为-0．61～0．18、-0．39～0．46，相关系数分别为0．984、0．978，均达到极显著相关。利用该模型能够对育种材料的蛋白质含量进行快速非破坏性活体测定．可大大提高育种选择效率。

With the technique of near infrared transmittance spectroscopy （NITS）, the mathematical model for quantitative analysis of protein content in living rice about Chongqing area was established by Partial Least Square（PLS）. The results indicated that the standard error of calibration （ SEC ）, standard error of cross-validation （ SECV ）, regression squared （RSQ） and 1 minus the variance ratio （ 1 -VR ） of brown and milled rice protein were 0. 252 ,0. 247 ;0.256,0.278 ;0.953,0.946 ;and 0. 951,0.940,respectively, the range of predictive errors about predicted value of Protein Content （ PC ） and lab reference value PC was - 0. 61 - 0. 18 and - 0. 39 - 0. 46,while correlation coefficient was 0. 984 and . 978 , both of which reached significant level. The model could detect quickly and nondestructively PC of breeding materials, so the selective efficiency of breeding was greatly improved.