工业流化床乙烯气相聚合中的在线测量与模型分析(I)

工业流化床乙烯气相聚合过程是一类典型的多变量复杂非线性系统。我们首先分析了若干主要过程变量(反应器温度等)对两个质量变量(树脂熔融指数和密度)的影响趋势，给出了两组变量间的非线性函数结构形式．然后运用多元统计投影原理和非线性函数的Taylor逼近原理建立了流化床乙烯气相聚合过程变量和质量变量之间的主元多项式非线性偏最小二乘软测量(Nonlinear PLS)模型，并对理论结果进行了推证。基于实际工业运行装置的采样数据对Nonlinear PLS模型进行了求取和验证，利用所求模型进行了树脂熔融指数和密度预测并将预测结果和误差分析结果与线性PLS及机理模型进行了比较。最后运用所建立的模型进行流化床乙烯气相聚合过程操作条件的模型预测分析.全文分为两部分，这是第一部分。

On-line Measurements and Model Analysis of Gas-phase Ethylene Polymerization Process in Fluidized Bed Reactor(I)

The process of gas phase ethylene polymerization in fluidized bed reactor (FBR) is one class of typical multi variable nonlinear systems. After discussing the interaction between process variables (reactor temperature, for example) and quality variables (melt index and density, for example), this paper developed a nonlinear partial least squares (PLS) regression model for on line resin qualities calculating based upon multivariable statistical projection technique and Taylor expansion principles. All the algorithms were proved. As a key step of this work, model verification is presented by using real operating data gathered from a industrial Unipol LLDPE FBR.The estimation of resin melt index and density by nonlinear PLS model were compared with by linear PLS and mechanical models. Finally, the relations between process variables and quality variables were analyzed by on line predicting of melt index and density.;This paper is divided into two parts, and this is the first part.