Development of Robust Tools for Simultaneous Design and Control and Multiscale Modeling and Control

本研究旨在设计新的数值工具，以整合动态系统的设计与控制，以及以多尺度建模方法描述系统的可控性。设计与控制的整合：本研究将发展一种新的最优设计与控制方法，该方法以最坏情况发生的概率为目标，即由于扰动的变化而导致输出的最大偏差。由此产生的方法将对工程实践的不同领域产生影响，因为它将被用作优化设计和控制大规模工业应用的实用和有效的工具。例如，这种方法将应用于加拿大自然资源公司CanmetEnergy运营的二氧化碳捕获试点工厂。因此，这项研究有可能通过重新设计这个试点工厂来提高工厂的二氧化碳捕获能力，或者提出一个新的设计，作为设计商业规模二氧化碳捕获工厂的基础。多尺度建模和控制：将设计新的健壮的灵敏度工具，能够控制基于多尺度建模的系统。多尺度模型是一组相互关联的模型，它们描述了系统在不同时间和长度尺度上的行为。所提出的控制策略将同时考虑细尺度模型参数引起的不确定性和用于测量过程中细尺度上发生的事件的传感器的当前能力。因此，由此产生的控制技术将提供一种更现实的方法来控制基于多尺度建模的系统，这些系统通常出现在半导体行业的应用中。例如，目前正在设计化学气相沉积(CVD)系统的一家当地公司有兴趣对这一基于多尺度建模的过程实施实时控制。因此，这项研究将促进加拿大对这一新兴领域的了解，因为本研究中提出的控制策略将有可能实现对薄膜生产的相关工业应用的在线控制，例如CVD工艺。

The present research aims to design new numerical tools for the integration of design and control of dynamic systems and for the controllability of systems described with a multiscale modeling approach.Integration of design and control: This research will develop a new optimal design and control methodology that is subject to a probability of occurrence of the worst-case scenario, i.e., the largest deviation in the outputs due to changes in the disturbances. The resulting methodology will impact on different areas of engineering practice since it will be used as a practical and efficient tool to optimally design and control large-scale industrial applications. For example, this method will be applied to a CO2 capture pilot plant operated by CanmetEnergy, a division of Natural Resources Canada. Thus, this research has the potential to increase the plant's capacity to capture CO2 by redesigning this pilot plant or to propose a new design that can be used as a basis to design commercial-scale CO2 capture plants.Multiscale modeling and control: New robust sensitivity tools that enable the control of multiscale modeling-based systems will be designed. A multiscale model is a set of interconnected models that describe the behaviour of systems at different time and length scales. The proposed control strategies will account for both, the uncertainty arising from the fine scale model parameters and the current capabilities of the sensors used to measure the events occurring at the fine scales in a process. Thus, the resulting control technologies will provide with a more realistic approach to control multiscale modeling-based systems, which usually emerge in applications in the semiconductor industry. For example, a local company that is currently designing a Chemical Vapour Deposition (CVD) system is interested to implement real-time control for this multiscale modeling-based process. Thus, this research will advance Canada's knowledge on this emerging field because the control strategies proposed in this research will have the potential to be implemented for the on-line control of relevant industrial applications for thin films production such as CVD processes.