Distributed moving horizon estimation and cooperative predictive control

由于全球竞争的加剧和对经济利益的追求，大型复杂过程在现代制造业中越来越普遍。一个大规模的复杂过程是由多个相互关联的单元组成的，这些单元通过物质流、能量流和信息流紧密结合在一起。现代制造过程的平稳运行高度依赖于状态估计和过程控制系统。一种状态估计系统提供了对不容易测量的过程变量的估计。过程控制系统利用过程变量的估计以自动方式驱动或维持过程在期望的操作点。传统上，大规模过程的状态估计和过程控制主要是在分散框架内研究的。然而，由于分散式框架完全忽略了子系统之间的互连，这可能会导致性能恶化，甚至失去闭环稳定性。这一事实导致了强烈呼吁分布式状态估计和协同控制系统的发展，能够实现改善的闭环性能，同时保持分散框架的灵活性。本研究建议致力于开发高效的分布式状态估计和过程控制方法的大规模过程。这一建议的成功将导致一个有效的分布式状态估计和更实用的分布式合作预测控制方法的非线性过程，是现在还没有。这些技术将大大提高加拿大制造业的经济效率，包括化学品、制药、石油化工和矿物加工。

Because of increasing global competition and the pursuit of economic profit, large-scale complex processes are becoming more and more common in modern manufacturing industries. A large-scale complex process is composed of multiple interconnected units that are tightly integrated through material, energy and information flows. The smooth operation of modern manufacturing processes highly depends on state estimation and process control systems. A state estimation system provides estimates of process variables that can not be easily measured. A process control system takes advantage of the estimates of process variables to drive or maintain the process at desired operating points in an automatic fashion. Traditionally, state estimation and process control of large-scale processes has been primarily studied within the decentralized framework. However, because the interconnections between subsystems are totally neglected in the decentralized framework, it may lead to deteriorated performance or even loss of closed-loop stability. This fact has led to strong calls for the development of distributed state estimation and cooperative control systems which are able to achieve improved closed-loop performance while preserving the flexibility of the decentralized framework.This research proposal is dedicated to the development of efficient distributed state estimation and process control methods for large-scale processes. The success of this proposal will result in an efficient distributed state estimation and a more practical distributed cooperative predictive control methodology for nonlinear processes that is not available now. These techniques will greatly improve the economic efficiency of Canadian manufacturing industries ranging from chemicals, pharmaceuticals, petrochemicals and mineral processes.