中文摘要：

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，司。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。

英文摘要：

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,