不确定条件下汽油调和调度的鲁棒优化模型(英文)

A ROBUST OPTIMIZATION MODEL FOR GASOLINE BLENDING SCHEDULING UNDER UNCERTAINTY

汽油调和是炼油企业最重要的操作,也是企业控制产品质量、降低生产成本的关键一步。然而汽油调和生产过程中往往面临诸多不确定因素,如:产品需求存在不确定性,调和组分的物性会出现波动。这些不确定因素为汽油生产调度带来挑战。为了应对这种挑战,提出了一种鲁棒优化模型。在不确定条件下,该模型所产生的调度策略能保证产品需求和产品质量要求得到最大程度的满足。该鲁棒模型的特点是,能处理既有连续概率分布参数又有离散概率分布参数的情形。另外,该模型并不采用场景遍历的方式来处理连续概率分布参数,这样就减小了模型规模,增强其应用于实际问题的能力。最后通过实验验证了该模型的效果。

Gasoline blending and scheduling is the last and important operation in a refinery. It is also the last chance to improve the quality of products and to reduce the production costs. Product producing and blending are always faced with uncertainty, such as uncertain demand for products and fluctuation of properties of blending components. A robust optimization model is used for product blending and scheduling to ensure that the final product can meet the requirements of quality and demand, which can deal with the situation of uncertain parameters with both continuous probability distribution and discrete probability distribution. At the same time, the model can avoid enumerating scenarios for uncertain parameters of continuous probability distribution, which are the drawbacks of scenario-based approaches for practical problems with a large number of uncertain parameters. Finally, an example is used to demonstrate the performance of the presented model and algorithm.