基于偏最小二乘法的近红外光谱分析应用

Near Infrared Spectral Analysis Based on Partial Least Squares

近红外光谱分析法是一种能够快速检测物质成分含量的分析方法。采用傅里叶变换的近红外光谱结合偏最小二乘法,对采样的126个玉米样本分别建立蛋白质、脂肪和纤维素的偏最小二乘模型,并进行定量分析。结果显示,模型的决定系数均大于0.95,定标集的均方误差均小于0.14。该方法与常规化学检测方法相比,检测简单、速度快、精度较高。

Near infrared spectral analysis technique is a kind of analysis method that can detect material component quickly.This paper adopts near infrared spectrum through Fourier transform. And it combines with the partial least squares to conduct about 126 samples on the modeling analysis. The result is that the model of decision coefficient is greater than 0.95. And the mean square error of calibration set is less than 0.14. This method is more simple,more fast and higher precision than the conventional chemical detected method.