An Economic Model Predictive Control Framework for Smart Manufacturing

背景：过程系统是涉及耦合材料、能量和信息流的复杂系统。流程系统的有效运行包括许多不同的任务。根据每个任务执行的相对频率，任务通常分为以下分层功能层：计划，调度，实时优化（RTO）（其中执行经济优化以找到最佳操作条件），高级控制（其中模型预测控制（MPC）通常用于驱动过程遵循由RTO层确定的操作条件）和监管控制。虽然这种不同任务的分层分离极大地简化了任务的设计和实现，并在工业中取得了成功，但它也将层次结构中每个层可用的数据分开，不适合下一代智能制造范式（或工业4.0）。在智能制造范式中，数据将在不同任务之间无缝共享，决策将以分布式和协作的方式做出。近年来，经济MPC作为一种非常有前途的先进控制（决策）框架出现了，它消除了RTO和先进控制之间的分离，在一个统一的框架中进行经济优化和控制。它已被公认为智能制造的核心技术之一。研究建议：经济货币政策的发展还处于初级阶段，有许多问题需要解决。基于我们之前在经济MPC方面的工作，在本研究中，我们将继续我们的长期目标，为智能制造开发一个灵活的经济模型预测控制框架。具体而言，在建议的工作中，我们将考虑以下重要问题：(a)通过区域跟踪处理经济MPC中的多个目标（经济优化和控制）；(b)计算效率高的控制不变集逼近算法；(c)发展一种有效的方法来表征经济MPC的（瞬时和无限时间）性能；(d)发展一个分布式的经济MPC框架，保证经济和控制性能；(e)发达的经济MPC框架在不同过程系统中的应用，包括油砂处理过程、废水处理厂、生物医学系统、灌溉过程。影响：本研究的成功将产生一个灵活的经济MPC框架和实用的绩效评估方法。同时，一些HQP将接受培训，掌握将经济MPC框架应用于不同系统的重要第一手经验。经济MPC框架和HQP将有助于加拿大制造业向智能制造范式的过渡。

Background: Process systems are complex systems that involve coupled materials, energy and information flows. The efficient operation of process systems includes many different tasks. According to the relative frequency with which each task is performed, the tasks are typically classified into the following hierarchical functioning layers: planning, scheduling, real-time optimisation (RTO) (in which economic optimisation is performed to find the best operating conditions), advanced control (in which model predictive control (MPC) is typically used to drive the process to follow the operating conditions determined by the RTO layer), and regulatory control. While this hierarchical separation of different tasks greatly simplifies the design and implementation of the tasks and has been successful in industry, it also separates data available to each layer in the hierarchy and is not suitable for the next generation smart manufacturing paradigm (or industry 4.0). In the smart manufacturing paradigm, data will be shared seamlessly between different tasks and decisions will be make in a distributed and cooperative manner. In recent years, economic MPC has emerged as a very promising advanced control (decision making) framework that removes the separation between RTO and advanced control, and performs economic optimisation and control in a unified framework. It has been recognised as one of the core techniques in smart manufacturing. Proposed research: The development of economic MPC is still in its early stage and there are many issues that need to be addressed. Based upon our previous work on economic MPC, in this proposed research, we will continue our long term goal in the development of a flexible economic model predictive control framework for smart manufacturing. Specifically, in the proposed work, we will consider the following important issues: (a) handling multiple objectives (economic optimization and control) in economic MPC through zone tracking; (b) computationally efficient algorithms for control invariant set approximation; (c) the development of an effective approach to characterise the (transient and infinite time) performance of economic MPC; (d) the development of a distributed economic MPC framework that gives guaranteed economic and control performance; (e) the applications of developed economic MPC framework to different process systems including oil sands processing processes, wastewater treatment plants, biomedical systems, irrigation processes. Impacts: The success of this proposed research will results in a flexible economic MPC framework with practical performance assessing methods. At the same time, a few HQP will be trained with important firsthand experience in applying the economic MPC framework to different systems. The economic MPC framework and the HQP will contribute to the transition of Canadian's manufacturing industries to the smart manufacturing paradigm.