GOALI: Robust Model Predictive Control Using Cost Function Constraints

摘要- Badgwell - 9714950 模型预测控制（MPC）是指一类通过使用显式过程模型来控制未来工厂行为的算法。 MPC技术用于整个石油精炼和化学工业。 典型的MPC控制器每分钟调整一次工厂输入，以优化未来的工厂性能。 工业MPC算法使用各种性能目标、输入参数化和过程模型。 MPC算法的强大之处在于它们对要控制的过程进行了数学描述。 然而，在实践中，该模型只是真实工厂行为的近似值，当模型误差变得显著时，可能会出现问题。 这个项目的目标是开发和实施一个新的家庭强大的MPC算法，其中工厂的不确定性直接纳入控制配方。 理论和计算问题将得到解决;原型算法将在Matlab编程语言中开发，并将在两个模拟过程中进行测试。 此外，还计划在埃克森化学公司的高纯度蒸馏塔和聚合物反应系统上进行现场试验。 经过验证的算法将被白杨科技公司纳入下一代MPC产品。 这是一个赠款机会学术联络与产业（GOALI）赠款。

Abstract - Badgwell - 9714950 Model Predictive Control (MPC) refers to a class of algorithms that control future plant behavior through the use of an explicit process model. MPC technology is used throughout the petroleum refining and chemical industries. A typical MPC controller adjusts plant inputs once a minute in order to optimize future plant performance. Industrial MPC algorithms use a variety of performance objectives, input parameterizations, and process models. MPC algorithms owe their power to the fact that they have a mathematical description of the process to be controlled. In practice, however, the model is only an approximation of true plant behavior and problems can arise when model errors become significant. The objective of this project is to develop and implement a new family of robust MPC algorithms in which plant uncertainty is incorporated directly into the control formulation. Theoretical and computational issues will be addressed; prototype algorithms will be developed in the Matlab programming language and will be tested on two simulated processes. In addition, field tests are planned on a high purity distillation columns and a polymer reaction system at Exxon Chemical Inc. The verified algorithms will then be incorporated into next-generation MPC products by Aspen Technology, Inc. This is a Grant Opportunities for Academic Liaison with Industry (GOALI) grant.