基于近红外光谱的轻质油饱和烃定量分析研究  

Research on the Prediction Method of Saturated Hydrocarbon Content in Light Oil Based on Near-infrared Spectroscopy

作　　者：彭酉格 Peng Youge(College of Chemistry and Chemical Engineering,Xi'an Shiyou University,Shaanxi,710065)

机构地区：[1]西安石油大学化学化工学院,陕西710065

出　　处：《当代化工研究》2025年第5期72-74,共3页Modern Chemical Research

摘　　要：石油成分信息的分子管理已成为炼油行业的一个重要趋势。为实现轻质油品中饱和烃含量的快速检测,使用近红外光谱仪轻质油样品在4350~12500cm-1波段内进行近红外光谱采集,建立偏最小二乘模型(Partial Least Squares,PLS)。首先,探究了不同的预处理方式对模型预测性能的影响,为了评估这些方法对原始近红外光谱数据处理的有效性,采用了R2cv和RMSEcv(%)作为模型性能的评价指标,然后基于最优预处理后的光谱进行变量选择。结果表明,对于饱和烃,基于标准正态变化(Standard Normal Variate,SNV)和区间偏最小二乘回归(Iterative Penalized Least Squares,iPLS)的PLS模型具有最佳的预测性能,R2p为0.9889,RMSEp(%)为0.53。研究结果表明,近红外结合PLS是一种快速、准确的轻质油中饱和烃的定量分析方法。The molecular management of petroleum component information has become an important trend in the refining industry.To achieve rapid detection of saturated hydrocarbons in light oil products,near-infrared spectroscopy was used to collect spectra of light oil samples in the 4350~12500 cm-1 range,and a partial least squares(PLS)model was established.First,different preprocessing methods were explored for their impact on the predictive performance of the model.To evaluate the effectiveness of these methods in processing the original near-infrared spectral data,R2cv and RMSEcv(%)were used as evaluation metrics.Subsequently,variable selection was performed based on the optimally preprocessed spectra.The results showed that for saturated hydrocarbons,the PL S model based on standard normal variate(SNV)and iterative penalized least squares(iPLS)exhibited the best predictive performance,with an R2p of 0.9889 and an RMSEp(%)of 0.53.The study indicates that near-infrared combined with PLS is a fast and accurate quantitative analysis method for saturated hydrocarbons in light oils.

关 键 词：近红外光谱 轻质油 饱和烃 

分 类 号：TE622.1[石油与天然气工程—油气加工工程]