论文部分内容阅读
以 精米粉为样品，研 究了不同光谱预处 理和回归 统计方法 在用近红 外反射光谱 分析稻米 表观直链淀粉含量（ Ａ Ａ Ｃ）时，对建立回 归方程的影响⒚结果 发现，光谱预处理对 校正结果影响较 小，不同光谱数学处理以一级衍生值较好，但“波段”和“间隙”长短对结果影响不大⒚回归统计方法对建立回归方程的影响最为明显，其中以修正的部 分最小平方法（ Ｍ Ｐ Ｌ Ｓ）建立的回归方 程效果最好⒚因此，在 建立以精米粉为样本进行稻 米 Ａ Ａ Ｃ 近红 外分析时，“光谱散 射校正／数学处理／回归统 计方法”组合以“标准正态变量转 换／１，５，５，１／ Ｍ Ｐ Ｌ Ｓ”最佳⒚用此组合建 立的回归方程测 定稻米 Ａ Ａ Ｃ 时，检验工作标准误 （ Ｓ Ｅ Ｐ）可小至 ０．８４，而决定系数可高达 ９４％ ⒚
Taking polished rice as sample, the effects of different spectral pretreatment and regression statistical methods on the establishment of regression equation when using near infrared reflectance spectroscopy to analyze the apparent amylose content (A A C) of rice were studied. The results showed that spectral pretreatment The results of the correction are less affected, and the derivative values of the first order of the spectrum processing are better, but the influence of the “band” and the “gap” has little effect on the results. The regression statistical method has the most obvious influence on the regression equation, The regression equation established by the partial least squares method (MPLS) works best. Therefore, the “Spectral Scatter Correction / Mathematical Processing / Regression Statistical Method” combination was established when rice AAC near-infrared analysis was established using polished rice flour as a sample “Standard Normal Variables Conversion / 1,5,5,1 / M P L S” Optimal ⒚ When using this combination to establish the regression equation for the determination of A A C in rice, the standard error of the test (S E P) can be as small as 0.84, and the decision factor can be as high as 94% ⒚