基于近红外光谱的汽油分子组成预测  

Prediction of Gasoline Molecule Composition Based on Near Infrared Spectrum

作　　者：蔡广庆 张莉 李春澎 胡益炯 王弘历[1] 杨诗棋 纪晔[1] CAI Guangqing;ZHANG Li;LI Chunpeng;HU Yijiong;WANG Hongli;YANG Shiqi;JI Ye(PetroChina Planning and Engineering Institute;Beijing Institute of Radio Metrology and Testing)

机构地区：[1]中国石油天然气股份有限公司规划总院 [2]北京无线电计量测试研究所

出　　处：《油气与新能源》2023年第1期111-116,共6页Petroleum and new energy

摘　　要：汽油分子信息解析耗时较长的问题制约着其在炼厂实时优化中的应用前景。依据汽油的近红外光谱及其对应的色谱分子组成数据,采用欧氏距离与多元线性回归方法拟合待测光谱,并把拟合参数代入对应的汽油分子数据库,建立了由近红外光谱快速预测汽油分子组成的模型。模型的预测值与实验值吻合较好,验证集的汽油分子组成预测平均绝对误差为0.0356,证明了此模型不但具有广泛适用性,而且满足炼厂汽油分子解析的精度要求。基于近红外光谱预测汽油分子组成的方法可以应用于炼厂的实时在线分析优化中,并且对炼厂反应过程模型和油品调和模型的建立具有重要意义。The time-consuming problem concerning gasoline molecular information analysis hampers its application prospect in refinery real-time optimization.Based on NIR(Near Infrared Spectrum)of gasoline and its corresponding molecular composition data,the paper used Euclidean distance and multiple linear regression method to fit the spectrum to be measured.Besides,the fitting parameters were substituted into the corresponding gasoline molecular database and the model consisted of NIR fast projection gasoline molecule was formed.The predicted values of the model agreed well with the experimental values and the mean absolute error(MAE)of gasoline molecule of validation set was 0.0356.Therefore,it proved that this model was not only quite applicable but also met the accuracy requirement of the analysis regarding gasoline molecule in refinery plant.The method concerning the prediction of gasoline molecular composition by near infrared spectroscopy could be applied in the real-time online analysis and optimization in refinery plant and was meaningful in terms of the establishment of Refinery reaction process model and oil blending model.

关 键 词：近红外光谱 汽油分子 欧氏距离 多元线性回归 在线分析 

分 类 号：TE622[石油与天然气工程—油气加工工程]