利用近红外光谱法测定玉米籽粒含油量的研究

Studies on the Determining of Oil Contents for Maize Kernels Using Near Infrared Spectroscopy

用近红外光谱(NIRS)分析技术和偏最小二乘法(PLS)建立了玉米籽粒含油量分析数学模型,并对模型预测结果的准确性进行了评价。结果表明,定标集、检验集的预测值与化学测定值间均呈极显著正相关,相关系数分别为0.958和0.957,定标标准差和预测标准差分别为0.757和0.745。利用该技术能测定玉米籽粒含油量。

Based on NIRS technique and partial least squares （PLS） algorithm, a calibration model was established to analyze oil contents for maize kernels. The result indicated that there was remarkable positive correlation between the prediction values and the chemical values. The correlation coefficients were 0. 958 and 0. 957, respectively. The standard error of calibration （SEC） was 0. 757, and the error of validation （SEP） was 0. 745. NIRS can be used to determine oil contents for maize kernels.