CAREER: Control Design of Complex Engineering Systems

复杂性日益成为高性能工程系统设计中的一个主要和核心问题。工程中大多数系统的巨大复杂性部分是由于长期以来认识到的事实，即即使是相对简单的物理系统也可以表现出非常复杂和微妙的行为，但更直接的原因是系统通过连接许多异质子系统而增长。在物理组件和计算能力快速发展的推动下，这种增长趋势似乎正在加速。随着系统变得越来越复杂，对系统性能水平的期望也越来越高。这对复杂系统工程提出了一个核心的设计挑战，因为这些系统的行为本质上是不确定的。除了处理复杂动力系统的更复杂和微妙的非线性行为外，当前和未来工程的控制和系统方法必须处理不完整或不确定的模型和信息的问题。同时，它们必须在计算上易于处理。PI研究计划的长期目标是开发直接适用于非线性复杂系统的先进控制和动态分析方法。该提案的研究计划描述了PI将在这个方向上采取的一些创新和实质性步骤，以及研究生和本科教育的综合计划。该计划的主要研究贡献将集中在三个主要领域：第一个领域是沿轨迹控制复杂非线性系统的分析和综合工具的构建，以及更一般的非线性控制方法。第二个领域是从实验数据中验证不确定非线性模型；这很重要，因为复杂系统的精确第一性原理建模是不可能的，因此与数据的直接比较是至关重要的。最后一个领域是对“由系统组成的系统”所需的专门控制策略的研究，这些策略可以随着新驱动技术的出现而得到更广泛的利用；具体地说，将采用分布式控制方法。该方案强调新的理论和分析方法的发展，但计算算法的创建和通用可重用软件的完成是该方案的主要目标；这将使该计划的想法和工具立即提供给从业者和行业。该计划将使用非线性，鲁棒和数值工具的组合。半确定规划将构成程序中大部分优化工作的基础，对特定领域优化的直接贡献被认为是程序的次要目标。为了补充建议的程序，计划的实验设施展示复杂的行为将用于批判性地评估和部分指导这个分析程序。该研究计划是在动力系统和控制理论，控制应用，数值模拟和优化的融合。研究生和本科教育都整合在研究项目中；参与该计划的学生将获得相当多的研究经验，使他们的工程背景更广泛，研究能力更强。该教育计划具体包括通过各种特别选择的项目，包括优化工具、基于网络的教育软件和实验建设项目，让本科生参与研究

ECS-9875244DullerudComplexity is increasingly a major and central issue in the design of high performance engineering systems. The enormous complexity of most systems in engineering is partially due to the long recognized fact that even relatively simple physical systems can exhibit very complex and subtle behavior, but is more directly attributable to the growth of systems by interconnecting many heterogeneous subsystems. This growing trend, fueled by the rapid development of physical components and computing capability, seems to be accelerating. While systems are becoming more and more complex, expectations on the level of system performance are increasing. This poses a central design challenge for complex systems engineering as the behavior of such systems are inherently uncertain. In addition to dealing with the more complicated and subtle nonlinear behavior of complex dynamical systems, control and systems methods for current and future engineering must necessarily contend with the issues of incomplete or uncertain models and information. At the same time they must be computationally tractable.The long term objective of the PI's research program is the development of advanced control and dynamical analysis methods that are directly applicable to nonlinear complex systems. The research plan of this proposal describes some innovative and substantial steps that the PI will take in this direction, together with an integrated program for both graduate and undergraduate education. The major research contributions of the proposed program will be in three main areas, The first area is construction of analysis and synthesis tools for the control of complex nonlinear systems along trajectories, and more general nonlinear control approaches. The second area is the verification of uncertain nonlinear models from experimentally obtained data; this is important because exact first principles modeling of complex systems is not possible, and so direct comparison with data is crucial. The final area is the investigation of specialized control strategies that are required for "systems comprised of systems" and can be exploited more generally with the appearance of new actuation technology; specifically distributed control methods will be pursued. The program proposed here emphasizes new theory and the development of analytical methods, however the creation of computational algorithms and the completion of general reuseable software are major objectives of the program; this will make the ideas and tools of the program immediately available to practitioners and industry. The program will use a combination of nonlinear, robust and numerical tools. Semi-definite programming will form a basis for much of the optimization work in the program, and direct contributions to domain specific optimization is considered a secondary aim of the program. To complement the proposed program, planned experimental facilities exhibiting complex behavior will be used to critically evaluate and partially direct this analytical program.The research program is at the confluence of dynamical systems and control theory, control applications, and numerical simulation and optimization. Both graduate and undergraduate education are integrated in the research program; students involved with the program will gain considerable research experience, making their engineering background broader and their research capabilities stronger. The education plan specifically includes the involvement of undergraduates in research through a variety of specially selected projects, including optimization tools, web-based education software, and experiment construction projects