摘要
提出一种新的汽油干点观测方法———机理模型与神经网络相结合的汽油干点观测方法。这一方法是先将Luenberger观测器理论应用到常压塔顶这一具有不可测输入的非线性系统 ,构造Luenberger观测器观测出一些不可测量但可观测的内部物理量 ;然后将这些内部物理量与其它物理量一起作为输入信号、汽油干点作为输出信号构成神经网络观测汽油干点 ,并用改进的学习算法训练神经网络 ;通过与其它方法进行比较 。
A new observing method of gasoline endpoint,which is based on mechanism model and ANN,is presented in this paper.This mothod applies Luenberger theory of observer into the upper section of crude unit,which is a nonlinear system and has unmeasured inputs,then constructs a Luenberger observer to observe some inner physical signals that can’t be measured but can be observed.Next,an ANN whose input is inner physical signals oberved and other physical signals measured,output is gasoline endpoint,will be construct to observe gasoline endpoint.The ANN is trained with a modified BP algorithm presented by this paper.After comparing with other methods,results show that the observing method obtained has higher accuracy and better robustness.
出处
《化工自动化及仪表》
CAS
北大核心
2001年第3期15-18,共4页
Control and Instruments in Chemical Industry
基金
中国石油天然气总公司石油科技中青年创新基金资助项目 !(编号 :科字 1998年第 3号 )
