燕麦干草品质的近红外光谱定量分析

应用近红外漫反射光谱(NIRS)分析技术,采用偏最小二乘回归法(PLS),建立了适合不同品种类型和不同生长发育时期的NIRS测定燕麦全株干草的粗蛋白(Crude Protein,CP)、秸秆中性洗涤纤维(Neu-tral Detergent Fibet,NDF)和酸性洗涤纤维(Acid Detergent Fiber,ADF)含量的稳定校正模型.结果表明,采用二阶导数(2st Deriv) 平滑处理(Norris)、多元散射校正(MSC) 二阶导数(2st Deriv) 平滑处理(Norris)、多元散射校正(MSC),分析谱区为9668～4 518,9 550～5 543,8 943～4 042 cm-1建立粗蛋白、中性洗涤纤维和酸性洗涤纤维的校正模型,其校正和预测效果最佳.其中CP与NDF的建标决定系数(r2cal)和交叉检验的决定系数(Rk)均在0.95以上,各项误差均小于3%,RPD值均大于3,逼近了化学分析的精确度,具有较好的预测效果.ADF的建模效果较CP与NDF差,其建标决定系数和交叉检验决定系数分别为0.912 0,0.855 3,建标误差(RMSEC)和检验误差(RMSECV)分别为2.33%,2.62%,接近了化学分析的精确度,且RPD值大于2.5,说明所建的ADF模型也可用于近红外预测.

Near Infrared Determination of Dry Hay Quality in Oats

In the present paper, the analysis of the content of CP, NDF and ADF in the whole dry hay of oats was carried out by using near infrared reflectance spectroscopy (NIRS) technique, and in combination with the partial least square (PLS) regression algorithm the calibration analysis was performed at the same time . The results showed that the calibration models developed by the spectral data pretreatment of the second derivative Norris smoothing, the multivariate scattering correction second derivative Norris smoothing, and the multivariate scattering correction were the best for CP, NDF and ADF with the same spectral regions (9 668-4 518, 9 550-5 543, 8 943-4 042 cm-1). A1l these models yielded coefficients of determination of calibration (r2cal) for CP and NDF that are both higher than 0.95, and each error lower than 3%, approached the chemical analysis precision. Moreover, the values of (RPD) of CP and NDF were both higher than 3.0. The results of these studies indicate that the contents of CP and NDF can be used to measure various samples in screening and evaluating quality constituents of dry hay in oats. While the effect of ADF modelling was poorer, the coefficients of determination of calibration (r2cal) and cross validation (r2CV)for ADF were 0.912 0, 0.855 3 respectively. The root mean square error of calibration, root mean square error of cross validation, and root mean square error of prediction ( RMSEE, RMSECV and RMSEP) for ADF were 2.33%, 2.62% and 1.91% respectively, and the precision is near the precision of the chemical analysis. The models of ADF can be used to measure various samples in screening and evaluating quality constituents of dry hay of oats also. This study has proved that NIRS technique can be applied to detect the contents of CP, NDF and ADF in the whole dry hay of oats.