摘要
氢气是催化重整过程的重要副产品,建立氢气纯度的软测量模型可以有效地指导再接触过程和下游用氢装置的生产操作。以某石化厂的催化重整装置脱氯前氢气为对象,提出了一种结合机理建模与辨识建模的混合建模方法。基于集总理论,提出了1种简单的三集总模型,根据再接触罐气液平衡原理构建了脱氯前氢气纯度的机理模型。使用最小二乘支持向量机对机理模型中的未知参数进行黑箱建模,并将该黑箱模型与机理模型连接构成脱氯前氢气的灰箱模型。利用生产数据对模型进行检验。结果表明所提出的混合建模方法可以较为准确地估算出脱氯前氢气的含量,优于传统的最小二乘支持向量机的建模方法。
Hydrogen is the by-product of catalytic reforming process, whose purity directly affects the operation of re-contacting section and downstream units. Aiming at modeling problem of hydrogen purity before dechlorination, a hybrid modeling method combining mechanism model and identification model was proposed. Based on the lumping theory and vapor-liquid equilibrium principle, a 3 components lumping model was developed for hydrogen purity of re-contacting process. Least Squares Support Vector Machines(LSSVM) was employed to build a black-box model for unknown parameter in the proposed mechanism model, which was connected with the mechanism model to form the gray box model of hydrogen purity. The proposed model was tested with production data. The results indicate the hybrid modeling method can predict the hydrogen purity with a satisfied precision, superior to LSSVM.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2014年第10期2465-2469,2475,共6页
Journal of System Simulation
基金
上海市科委基础研究重点项目(12JC1403400)
关键词
催化重整
氢气纯度
集总模型
最小二乘支持向量机
灰箱模型
catalytic reforming
hydrogen purity
lumping mode
least squares support vector machines
gray-box model