摘要
对一类工业固定床反应器出口转化率提出了一种基于改进混合模型的推断估计。该推断估计器包括过程简化机理模型，改进型径向基函数网络（ＲＢＦＮ）及在线自校正三个部分。仿真结果表明，该推断估计器具有良好的静态、动态特性，较高的估计精度和较好的实时性，所需训练样本少，易于理解和研究，外推性能良好等优点。
Based on improved hybrid models an inferential estimator for the output conversion of fixedbed reactors is presented.It includes three parts:the reduced mechanism models comprising the process prior knowledge,the improved RBFN to estimate the catalyst activity which is difficult to be modelled and the online selfcompensator to dynamically correct the estimator.Simulation results show that the estimator provides good steady and dynamic characteristics,high accuracy,good swiftness and generalizaiton performacnce.Besides,it needs less training samples and is easy to be understood and interpreted.
出处
《华东理工大学学报（自然科学版）》
CAS
CSCD
北大核心
1998年第3期354-360,共7页
Journal of East China University of Science and Technology
基金
国家自然科学基金
关键词
固定床反应器
混合模型
推断估计器
动态校正
ANN
radial basis function network (RBFN)
fixedbed reactor
catalyst activity
hybrid model
inferential estimator
dynamic compensation
