光谱数据对油页岩含油率近红外光谱分析PLS建模精度的影响

目前油页岩关键的评价参数——含油率的检测方法均无法实现原位测量，无法满足油页岩资源的勘查和开采中样品检测的要求。便携式近红外光谱分析技术，为实现油页岩含油率的原位检测提供了可能性。由于光谱数据的不同形式与样品的成分含量值之间有不同的相关关系，样品不同成分的吸收特性表现在不同的近红外波段上，因此利用合成样品，针对反射率、吸光度、K-M函数等三种不同的光谱数据表示形式和四种不同的建模区间，研究它们对油页岩含油率PLS模型精度的影响情况。结果表明：对于合成样品，进行PLS建模的最佳光谱数据形式是反射率，最佳建模区间是组合特征区间，即适当的光谱数据形式及建模区间可提高模型的精度。

Effect of Near Infrared Spectrum on the Precision of PLS Model for Oil Yield from Oil Shale

It is impossible to use present measurement methods for the oil yield of oil shale to realize in-situ detection and these methods unable to meet the requirements of the oil shale resources exploration and exploitation. But in-situ oil yield analysis of oil shale can be achieved by the portable near infrared spectroscopy technique. There are different correlativities of NIR spectrum data formats and contents of sample components, and the different absorption specialities of sample components shows in different NIR spectral regions. So with the proportioning samples, the PLS modeling experiments were done by 3 formats (reflectance, absorbance and K-M function) and 4 regions of modeling spectrum, and the effect of NIR spectral format and region to the precision of PLS model for oil yield from oil shale was studied. The results show that the best data format is reflectance and the best modeling region is combination spectral range by PLS model method and proportioning samples. Therefore, the appropriate data format and the proper characteristic spectral region can increase the precision of PLS model for oil yield form oil shale.