重整汽油在线拉曼分析系统开发与工业应用  

Development and industrial application of online Raman analysis system for catalytic reforming gasoline

作　　者：王拓[1] 戴连奎[1] 

机构地区：[1]浙江大学工业控制技术国家重点实验室,杭州310027

出　　处：《仪器仪表学报》2015年第6期1201-1206,共6页Chinese Journal of Scientific Instrument

基　　金：国家863计划(2012AA10A503)项目资助

摘　　要：针对催化重整装置的优化需要,研制开发了一套新的在线拉曼分析系统。系统硬件由采样单元、分析主机以及连接两者的远距离光纤构成;系统软件包括光谱获取、非线性模型计算和数据通信等部分。该系统已应用于某一工业重整装置,以实现对重整汽油研究法辛烷值(RON)与其他关键参数的在线分析。通过连续九个月的现场运行表明系统运行良好,包括RON在内的分析数据可以快速反映重整反应器温度和其他操作条件的变化。RON在线分析的重复性误差为0.015个辛烷值;非线性模型预测复相关系数为0.990 9,标准预测误差为0.31个辛烷值,平均绝对值误差为0.26个辛烷值。运行结果表明:拉曼分析系统可以为工艺装置的操作优化提供准确而重要的检测参数。Aiming at the optimization requirement of catalytic reformer,a new set of online Raman analysis system is developed. The hardware system is composed of a sampling unit,host analyzer and two long-distance fibers connecting both of them; the software consists of spectral acquisition,nonlinear model calculation and data communication. The system has been applied to a certain industrial reformer to detect the research octane number（ RON） and other key parameters of reformed gasoline. Nine-month continuous field application shows that the online system performs well. The obtained analytical data,including RON,can rapidly reflect the change of reforming reaction temperature and other operating conditions. The repeatability error of RON is 0. 015 octane number. The nonlinear model prediction correlation coefficient is 0. 990 9. The standard prediction error and the mean absolute error are 0. 31 and 0. 26 octane number,respectively. Application results show that the Raman spectral analyzer can accurately provide important detection parameters for the process optimization.

关 键 词：催化重整汽油 拉曼光谱 最小二乘支持向量机 研究法辛烷值 

分 类 号：TP2[自动化与计算机技术—检测技术与自动化装置] TH7[自动化与计算机技术—控制科学与工程]