基于矩值的聚合物分子量分布控制

分子量分布（MWD）可以全面反映聚合物微观特性进而决定聚合物性质和用途，因此能够提供某种分子链长平衡的分子量分布就成为聚合物微观质量控制领域的重要研究课题。本项目在分子量分布灰箱建模的基础上，探讨基于分布函数数字特征的MWD控制问题。利用灰箱模型可以建立有限阶次矩向量与MWD之间单射关系的特点，以矩值为直接被控对象，研究基于有限阶矩值对MWD函数进行部分反馈控制的通用解决方案。首先从分布函数及矩值的性质出发，提出被控矩向量降维准则并进行证明，为以有限阶矩值为直接被控目标实现MWD的精确跟踪提供理论基础。然后基于矩值的互相关性，以多变量统计过程控制方法为基础，探讨高维被控矩向量降维解耦的实现方法。最后，结合低阶矩可测/可推断的特点，提出加权控制目标函数，自控制目标中抽象出通用约束条件，实现MWD曲线的精确反馈控制。研究成果也可以应用到类似的分布参数系统中，实现建模与最优分布形状的精细控制。

The micromolecular architectures strongly influence properties and applications of the polymer, thus control of molecular weight distribution has become an important research target. Based on gray-box modeling result, the control of MWD through the numerical characters of distribution function will be investigated. Moments has been taken as direct control object to get a general control scheme to MWD function. The controlled moment vector dimensionality reduction will be proposed and testified based on the characters of distribution function and moments which will provide a solution for the accurate tracking of MWD. Based on the interdependency of moments, the decoupling method for high-dimension controlled moments will be discussed through the multivariate statistical process control method. Finally, a weighted control target function will be proposed combined with the character of lower-order moments are measurable or predictable. General limited constraints can be abstracted from the control object, and the accurate feedback control of MWD profile can be realized. Research results also can be used in other similar distribution systems and the accurate control of optimal distribution profile can be realized.