基于神经网络的自适应逆控制在CSTR中的应用  

Application of adaptive inverse control based on neural network for CSTR system

作　　者：王宏楠[1] 林丽君[1] 

机构地区：[1]辽宁石油化工大学信息与控制工程学院,辽宁抚顺113001

出　　处：《纺织高校基础科学学报》2009年第2期258-261,共4页Basic Sciences Journal of Textile Universities

基　　金：辽宁省教育厅科学研究计划资助(2004D031)

摘　　要：针对CSTR系统,提出一种基于神经网络的自适应逆控制方案.该方案将预测思想引入自适应逆控制中,采用E lman网络作为模型辩识器,利用它建立CSTR系统输出的预测模型;采用模糊神经网络作为自适应逆控制器,由预测误差进行参数的在线自适应寻优.仿真结果表明,该方案对于CSTR系统具有很好的跟踪效果和较强的鲁棒性.A scheme of adaptive inverse control based on neural network is proposed for the continuous stirred tank reactor （CSTR） system. The thought of prediction is introduced to adaptive inverse control. The predictive model of the CSTR system＇s output is estabIished by model identifier, which adopts Elman network. A fuzzy neural network is adopted for adaptive inverse controller, and the parameters of controller are trained by predictive error, so that they can have optimum searching on line. The simulation result on the CSTR indicates that the scheme has good traceability and robusmess.

关 键 词：自适应逆控制 ELMAN网络 模糊神经网络 CSTR 

分 类 号：TP271[自动化与计算机技术—检测技术与自动化装置]