CAJ | 학술논문

基于近红外光谱检测牛奶中蛋白质和脂肪含量是当前的一个研究热门，主要可以利用近红外反射光谱，近红外漫反射光谱和近红外透射光谱的光谱信息来分析牛奶的主成分含量。本文是利用近红外透射光谱原始数据，通过数学方法转换，从吸光值、吸光比和吸收度三种形式，采用偏最小二乘法（PLS）回归，分别建立了其与牛奶主成分浓度之间的校正模型。对比三种形式所建立的数学模型的优劣，探讨了如何提高PLS校正模型预测精度的相关问题。
기우근홍외광보검측우내중단백질화지방함량시당전적일개연구열문，주요가이이용근홍외반사광보，근홍외만반사광보화근홍외투사광보적광보신식래분석우내적주성분함량。본문시이용근홍외투사광보원시수거，통과수학방법전환，종흡광치、흡광비화흡수도삼충형식，채용편최소이승법（PLS）회귀，분별건립료기여우내주성분농도지간적교정모형。대비삼충형식소건립적수학모형적우렬，탐토료여하제고PLS교정모형예측정도적상관문제。
Testing the amount of protein and fat contained in the milk based on the near infrared spectroscopy is a popular research, that near infrared reflectance spectroscopy, the spectral information of near infrared diffuse reflectance spectroscopy and near infrared transmission spectroscopy is the main method. In this paper the models about main components concentration of the milk and the light absorption value, absorption ratio, absorbance were built respectively using PLS(partial least squares) by mathematical transformation of the original data. The issue about improving the precision of the PLS models was discussed, after the three models with different ways were compared.