连续聚合反应过程可控性分析与牌号切换循环调度一体化研究

The production of high-end polymers with agility and flexibility calls for the optimal production sequences and high-efficient strategy for grade transitions with desirable dynamics. Promoted by the balance between the economic performance and transition dynamics, the aim of this project is to conduct a strategy for integrating the controllability analysis and the cyclic scheduling of grade transitions for continuous polymerization processes based on the mixed integer nonlinear programming and classic control theory. Specifically, the research concentrates on the following aspects: A new data-driven modeling approach based on the integration of the generalized disjunctive programming and the symbolic regression is pursued to mathematically demonstrate the relationship between the bifurcation points of the controllability and the process design/operating variables, and such approach also stands for the mathematical conduction of the relationship between the transition time and the process design/operating variables. Incorporating the above set data-driven mathematical models, a mixed integer nonlinear programming (MINLP) model without any consideration of uncertainties is formulated. Furthermore, based on the hybrid robust optimization–stochastic programming approach, the updated MINLP with the consideration of uncertainties can improve the robustness and agility. The living/controlled radical styrene polymerization process is adopted to demonstrate the effectiveness of the proposed strategy. The results can be expected to lay the theoretical foundation for the simultaneous design, scheduling, and control of the multi-product polymerization processes. Moreover, this research will bridge the upper management/decision level with the lower control/operation level to accelerate the development of smart polymerization process manufacturing.

敏捷、柔性生产模态下的高端聚合产品的生产过程，需要探寻产品生产的优化序列和牌号切换过渡运行有效策略。本课题以混合整数非线性规划和经典控制理论为手段，开展多牌号连续聚合反应过程的可控性和切换循环调度集成方法的研究。建立可控性分析和牌号切换循环调度集成的混合整数非线性规划模型，解决聚合反应牌号切换过程动态性能和经济性同步协调的难题；提出融合广义析取解析和符号回归的精度可控数据驱动建模方法，数学表征所研究过程的可控性、切换过渡时间在关键设计、操作参数可行域内的变化规律；建立集成鲁棒优化和随机规划的不确定性混合整数非线性规划模型，提升聚合反应应对不确定性因素的敏捷性、鲁棒性；利用可控/活性自由基苯乙烯聚合过程验证并完善所提集成策略。研究成果将为聚合过程设计、调度与控制一体化奠定理论基础，同时有助于搭建聚合工业上层经营、运行决策和下层生产操作之间的集成平台，以提升聚合工业的智能优化制造水平。

本项目以混合整数非线性规划和经典控制理论为手段，针对多产品连续化工生产过程的鲁棒优化决策与调度优化，从基础理论和算法角度，针对如下几个方面进行了深入研究，并取得了系列成果：1. 综合考虑过程可操作性（可控性、稳定性），建立了多产品聚合过程牌号循环调度优化模型，有效平衡了复杂化工过程的安全性和经济性，为解决连续化工过程的调度问题提供了新的思路和方法。2. 提出了符号回归和数学规划混合驱动的代理模型构建方法，开发了相应的全局优化求解算法，形成了精度可控的数据驱动显性代理模型构建技术，为过程优化设计、操作优化、先进控制提供了模型保证。3. 提出了一种融合随机规划和鲁棒优化的不确定性优化模型，开发了相应的全局优化求解算法，在保证不确定性条件下的过程优化解的优化性能前提下，有效降低了不确定性条件下的优化解保守性。4. 提出了基于沃瑟斯坦距离的分布式鲁棒机会约束规划(WDRCCP)的不确定条件下的过程调度优化的新框架，有机融合了鲁棒优化和机会约束规划的特性，展示了不确定性优化框架良好的样本外性能即鲁棒性。研究成果将为多产品化工过程例如聚合过程优化决策和控制一体化奠定了坚实的理论基础。3年项目执行期间，共发表学术期刊论文8篇，会议论文1篇，参加国内外学术会议并作口头报告3次，在国内外学术会议作邀请报告4次，邀请国外知名学者来华访问4次。已毕业博士生1名，在读博士生4名，硕士生1名。

内容获取失败，请点击重试