催化裂化原料和液相产物中氢含量的回归预测模型  

Regression Model of FCC Feedstock and Liquid Products Hydrogen Content

作　　者：崔守业[1] CUI Shouye(Research Institute of Petroleum Processing,SINOPEC,Beijing 100083,China)

机构地区：[1]中国石化石油化工科学研究院,北京100083

出　　处：《石油学报（石油加工）》2021年第3期662-668,共7页Acta Petrolei Sinica(Petroleum Processing Section)

摘　　要：以催化裂化原料和裂化液相产物(汽油、轻循环油(LCO)、油浆)的20℃密度、硫质量分数、30%馏出温度、50%馏出温度为4个基本回归因子,并根据其交互作用构造了10个交互回归因子,采用逐步回归法建立了4种馏分中氢质量分数的回归模型(汽油不考虑硫质量分数的影响,故为9个回归因子)。各模型回归结果表明,4种模型的校正决定系数R 2分别为0.868、0.727、0.949、0.973,标准化残差均基本沿0水平线上下分布,说明回归模型较为理想。采用文献数据验证模型预测效果,结果表明,4种模型预测结果的均方误差在0.00943~0.03150,均方根误差在0.0971~0.1780,模型预测效果较好。说明所建立的氢质量分数回归模型可用于催化裂化原料和产物汽油、轻循环油、油浆中氢质量分数的快捷有效预测。Stepwise regression method was used to construct regression models to characterize the hydrogen mass fraction suitable for FCC feedstock and liquid products by using density of 20℃,sulfur mass fraction,30%distillation temperature,50%distillation temperature and 10 interaction factors(9 regression factors for gasoline without considering the effect of sulfur mass fraction).The regression results of 4 hydrogen mass fraction regression models show that the adjusted R 2 of feedstock,gasoline,LCO and slurry models are 0.868,0.727,0.949 and 0.973,respectively.The standardized residuals are equally distributed around the zero horizontal line,indicating that the regression model is ideal.Furthermore,the testing results show that the mean square error of the prediction results are in 0.00943-0.03150,and the root mean square error are from 0.0971 to 0.1780,indicating that the model has a good prediction accuracy.Overall,the constructed model could be used to predict the hydrogen mass fraction of FCC feedstock and liquid products quickly and effectively.

关 键 词：催化裂化 氢 密度 馏程 硫 

分 类 号：TE624[石油与天然气工程—油气加工工程]