近红外光谱用于原油快速评价的研究

摘要

基于345种原油建立原油近红外光谱数据库,通过偏最小二乘方法建立快速测定原油主要性质的分析模型.利用原油近红外光谱指纹特征提出原油种类精确识别方法——移动相关系数法,结合原油评价数据库可快速得到单种类原油的详细评价数据.对于混兑原油,采用库光谱拟合方法,可从原油近红外光谱数据库中解析出一组参与混兑的“伪原油种类”及其混兑比例,结合原油评价数据库可得到该混兑原油的详细评价数据.%Near infrared (NIR) spectroscopy calibration models for determining the main physical and chemical properties of crude oil were built using partial least square method based on NIR spectral library consisting of data from 345 crude oils. A method for rapid identification of crude oil type,named moving window correlation coefficient, was proposed, which can distinguish spectra of highly similar crude oils. For blended crude oils,library spectra fitting method was used to obtain pseudo-blended crude oil types and their corresponding blending proportions. Combined with crude oil assay database,detailed crude oil data of unknown crude oils can be rapidly provided by the above methods.