【作者】 皇才进； 韩鲁佳； 刘贤； 杨增玲；

【Author】 HUANG Cai-jin HAN Lu-jia LIU Xian YANG Zeng-ling College of Engineering,China Agricultural University,Beijing 100083,China

【机构】 中国农业大学工学院；

【摘要】 二维相关光谱是光谱分析中重要的研究方法之一,它可将光谱信号扩展到第二维上以提高光谱分辨率,简化含有许多重叠峰的复杂光谱。本文提出了利用二维相关光谱图中自相关系数来优选波段,建立秸秆样品热值的近红外预测模型。这种方法剔除了相关性弱的变量,从而显著减少了建模所用的变量数目,提高了模型的精度。得到的较优模型的校正集相关系数(r_cal²)与标准校正误差（SEC）分别为0.9512和176 J·g^-1,验证集的相关系数(r_val²)与标准预测误差（SEP）分别为0.973 0和139 J·g^-1。更多还原

【Abstract】 As an extensively applied method in spectral amlysis,two dimensional correlation spectroscopy is advantageous in en- hancing spectra resolution and simplifying overlapped spectra.This study investigated the method of selecting highly related wavelengths by examining autocorrelation coefficients in two dimensional correlation spectroscopy,to develop near infrared spec- troscopy models for predicting calorific value of straw.With less related varibles involved,the models presented improved accu- racy.r_cal² and SEC of the optimized model were 0.9512 and 176 J·g^-1;corresponding r_val² and SEP were 0.9730 and 139 J·g^-1,respectively.更多还原