苯乙烯聚合反应过程的灰箱建模  

Hybrid modeling of reaction process of styrene polymerization

作　　者：张利芹[1] 曹柳林[1] 

机构地区：[1]北京化工大学信息科学与技术学院,北京100029

出　　处：《电子测量技术》2012年第5期49-53,共5页Electronic Measurement Technology

基　　金：国家自然科学基金(60974031)

摘　　要：针对结构部分已知且参数未知的一类工业反应过程的建模问题,以苯乙烯聚合反应过程为背景展开研究。在该反应过程中,反应机理已知,包含链引发反应、链增长反应、链转移反应以及链终止反应等,而其中各基元反应中的反应速率方程的结构及参数不甚明了。针对此类问题,提出将已知的机理部分构建白箱模型,将未知的反应速率部分构建神经网络黑箱模型,并将黑箱模型与白箱模型连接构成整个反应过程的灰箱模型。该方法通过实验的验证,证明了其实用性及有效性。Aimed at the modeling problem of the kind of industrial reaction process that have the part of known structure and the unknown parameters, this paper use a styrene polymerization reaction as study background. In the reaction process, the reaction mechanism is known, including chain initiation reaction, chain growth reaction, chain transfer reactions chain termination reaction and so on, but in the kinds of the reactions, the Arrhenius equation parameters of the kinds of the reaction rates and the reaction order are unknown. In view of these problems,this paper presents that builds the white box model for the part of the known reaction mechanism and builds the black-box model for the unknown reaction rates using neural networks. Finally, the black box model and the white box model are connected to form the gray box model of the whole reaction process. The method is proved practicability and validity by the experiments in this paper.

关 键 词：工业反应过程 苯乙烯聚合反应过程 神经网络模型 机理模型 灰箱模型 

分 类 号：TP273.5[自动化与计算机技术—检测技术与自动化装置]