第 17 卷第 3 期计算机与应用化学 Vol. 17, No. 3 2000 年 5 月 28 puters and Applied Chemistry May , 2000 FCC 油品质量指标智能监测系统的数据挖掘与修正技术 徐强何小荣陈丙珍 ( 清华大学化工系北京 100084) 摘要: 为使 FCC 油品质量指标智能监测系统更加有效地实时预测催化裂化生产过程中的汽、柴油重要质量指标, 本文利用模糊匹配技术和计算机网络技术, 建立了基于稳定历史工况的数据挖掘及修正子系统, 并在实际生产 装置中稳定运行, 结果表明该子系统能更加有效、准确地对稳定汽油蒸汽压、粗汽油干点和轻柴油凝点进行实 时修正, 从而提高了 FCC 油品质量指标智能监测系统的预测精度和运用水平。 关键词: 人工神经网络; 在线监测; 干点; 凝点; RVP; 数据挖掘 中图法分类号: TP 311, TQ 019 文献标识码: A 文章编号: 1001- 4160 ( 2000) 03- 210-214 Data Mining and Modifying Technology of Intellect Monitoring System on FCC Oil Qualities XU Qiang HE Xiao-rong CHEN Bin- zhen ( Department of Chemical Engineering , Tsinghua University , Beijing 100084) Abstract: A new sub- system was proposed to improve the monitoring results of an intellect monitoring system for FCC oil qualities which were Rador vapor pressure ( RVP) of steady gasoline, dry point of crude gasoline and freezing point of light diesel oil. One kind of data mining and modifying technology based on historical steady data samples was used in the sub- system. It was proved to be quite useful to increase the monitoring accuracy of the whole system in real time practice on FCCU . Key words: ANN, on- line monitoring, dry point, freezing point, RVP, data mining 1 前言 先进控制在实际生产中得以有效实施的一个重要前提是实时而准确地获知待控产品的质量指标。例如 FCC 油品质量指标智能监测系统就是对催化裂化装置( FCCU) 生产过程中的重要汽柴油质量指标( 分馏塔粗 汽油干点、轻柴油凝点和稳定汽油蒸汽压) 实施动
