基于Kriging函数的聚合反应过程分子量分布的回归建模与优化  

Modeling and optimization of polymer molecular weight distribution based on Kriging method

作　　者：刘剑峰[1] 占志良[1] 邵之江[1] 陈曦[1] 

机构地区：[1]浙江大学控制科学与工程系,浙江杭州310027

出　　处：《计算机与应用化学》2012年第1期31-36,共6页Computers and Applied Chemistry

基　　金：国家自然科学基金项目(61074148);中央高校基本科研业务费专项资金项目

摘　　要：以分子量分布作为聚合物产品的质量指标越来越受到关注,针对分子量分布机理建模较复杂的问题,提出了利用克里金插值方法获得了基于数据的分子量分布模型。首先,建立了分子量分布的离散化模型,可以较为准确地估计特定链长处的分子量,从而实现对分子量分布曲线的离散化表征。进一步,提出了一种多维插值建模方法实现了对分子量分布的整体建模,可以对于任意链长处的分子量进行估计。最后将基于数据的分子量分布模型替代机理模型,实现了对分子量分布的静态优化,获得了较好的效果。Molecular weight distribution （MWD） is an important quality index for polymer products. As the first principle model of MWD is very complex and unsuitable for online usage, the Kriging method is used to develop the molecular weight distribution model of polyolefin based on data obtained from a rigorous model. First, a number of composition models with specified chain length axe built with the Kriging method, which can accurately predict the composition of the polymer with given chain length. MWD can be further constructed by using the outputs of those models. Next, a unified model independent on the chain length is also developed which includes chain length as one input together with the other process variables. Finally, optimization on the polymer production with given MWD is conducted based on the developed model,

关 键 词：聚烯烃 分子量分布 建模与优化 克里金插值 

分 类 号：TQ015.9[化学工程] TP391.9[自动化与计算机技术—计算机应用技术]