Advanced Optimal Control Methods for Non-Linear and Distributed-Parameter Processes

为了在竞争日益激烈、利润率下降、消费者对产品质量的期望不断提高以及对能源和环境保护的担忧日益加剧的情况下实现最高效率，现代工业对控制方法提出了越来越高的要求。在处理化学过程时，这是一个主要的挑战，因为化学过程是普遍存在的非线性和非均匀的，并且由复杂的数学模型来描述。 尽管控制研究、高速计算和计算机硬件的进步有助于更好地控制工业过程，但仍有相当大的空间来开发计算更高效、响应时间延迟明显更小的控制方法，从而提供对与工业高度相关的复杂过程的更有效的实时控制。 拟议的研究计划结合了发展先进的最优控制方法的两个主要举措。第一项倡议涉及开发和测试高性能的最优反馈控制方法，大大减少了计算延迟，用于非线性、分布参数过程的实时控制。这一举措包括开发基于串联变换和同伦延续的快速反应模型预测控制策略。还包括周期最优控制策略的设计，以提高连续过程操作的性能。第二项倡议利用人工智能，包括人工神经网络、强化学习和进化计算，结合所提出的控制策略开发极其高效和有弹性的实时控制方法。 通过开展上述工作，拟议的研究计划将有助于开发复杂、非线性和非均匀过程的先进控制方法，这些过程在工业中比比皆是。这些方法将提高能源和资源的利用率，降低对环境的影响，并支持生产质量持续更高的产品。 最后但并非最不重要的一点是，该研究计划将为高级最优控制和人工智能领域的两名博士和两名硕士学生提供宝贵的培训机会。这些学生将发展行业对知识和技能的高需求。

To achieve top efficiencies amid increasing competition, reduced profit margins, increasing product quality expectations from consumers, and rising concerns to protect energy and the environment, modern industry demands an increasingly higher performance from control methods. This is a major challenge when dealing with chemical processes, which are ubiquitously non-linear and non-uniform, and are described by sophisticated mathematical models. Although the progress in control research, high-speed computing and computer hardware has helped in better control of industrial processes, there is considerable scope to develop control methods that are computationally more efficient, respond with significantly less time delay, and thus offer more effective real-time control of complex processes, which are highly relevant to industry. The proposed research program incorporates two major initiatives in the development of advanced optimal control methods. The first initiative involves the development and testing of high-performance optimal feedback control methods with greatly reduced computational delays for real-time control of nonlinear, distributed-parameter processes. This initiative incorporates developing fast-acting model predictive control strategies based on series transformation and homotopy continuation. The design of periodic optimal control strategies is also included to enhance the performance of continuous process operations. The second initiative exploits Artificial Intelligence including artificial neural networks, reinforced learning, and evolutionary computation to develop extremely efficient and resilient real-time control methods in conjunction with the proposed control strategies. By working on the above initiatives, the proposed research program will contribute to the development of advanced control methods for complex, non-linear, and non-uniform processes, which abound in industry. These methods will enable improved utilization of energy and resources, lower environmental impacts, and support production of consistently better quality products. Last but not least, the research program will provide valuable opportunities for the training of two PhD and two Master's students in the field of advanced optimal control and Artificial Intelligence. These students will develop knowledge and skills in high demand by industry.