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

梁军

DOI: https://doi.org/10.3969/j.issn.1004-1699.2003.03.004

2003-01-01

Abstract: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.