具有不确定性的聚烯烃牌号切换过程动态优化

Optimal grade transition in polymerization processes under uncertainty

聚烯烃生产经常需要进行牌号切换操作,最优牌号切换策略能在保证生产安全的前提下减少切换过程中的经济损失。由于实际生产中存在不确定性,基于理论模型的最优解无法保证最优的切换结果,甚至使牌号切换失败。对此,有学者将最优性条件校正策略引入牌号切换问题中,在克服不确定性方面体现出较好效果。从最优输入轨迹中提取合理的解模型是校正策略实施的关键。目前,对于如何提取解模型缺乏系统的理论和公认的方法。本文提出了一种新的解模型提取方法,从校正系统对解模型的要求出发构造并求解系列子优化命题,以此为依据对最优牌号切换命题重新参数化求解以提取解模型。仿真结果表明,该方法能显著简化解模型结构,保证良好的校正性能,实现牌号在不确定性条件下的顺利平稳切换。

Grade transition is quite common in the polymerization process due to diversified market demands. Optimal grade transition could minimize the amount of off-specification polymer and transition time while meeting operational and safety requirements. Numerical optimization based on the nominal model is typically insufficient due to its uncertainty. A kind of close-loop approach based on tracking the necessary condition of optimality （NCO-tracking） was used in the polymerization process. The solution model generated from the nominal optimal solution was of great significance in NCO-tracking. This paper presents a new method for extracting the solution model. In this method the requirements from the decentralized control scheme was considered by solving a series of small-scale optimization problems. It was shown that a significantly simplified solution model could be generated and grade transition was successful in the presence of uncertainty in the forms of model mismatch and process disturbance.