Integrated Research on Distributed Control for Large-Scale Dynamically-Interconnected Systems

工程设计的最新进展产生了各种大型系统，提供了巨大的经济效益。许多这样的系统由多个独立的单元组成，这些单元直接与邻居交互。例如，微型机电传感器和执行器阵列、电网和紧密耦合的飞机编队飞行。为大规模动态互联系统设计有效的分布式控制算法的几个限制因素包括：(A)与高状态维和大量输入和输出相关的复杂性；(B)子系统之间通过动态耦合和通信进行的交互；(C)子系统级别的异构性和建模不确定性。随着系统的复杂性和计算和通信中更强大的工具的出现，通信、控制和计算之间的强联系变得越来越重要。本项目旨在通过综合控制理论和优化技术的思想，开发用于大规模动态互联系统的分布式控制技术。具体地说，博弈论将提供一种通用的方法来开发子系统之间的动态交互，并改进分布式游戏框架中的控制设计。分布式控制策略将为互联系统提供简化的控制实现和提高的系统可靠性。此外，将在线优化应用于这类复杂系统将获得显著的性能改善。其他将产生重大影响的工程系统包括使用传感器和执行器阵列的对象运输和操纵的分布式操作、电信网络流量和大规模电网上的电力分配。创新教育方法的发展将通过为本科生和研究生提供宝贵的研究经验和动手实践技能来促进北卡罗来纳州立大学的教育计划。开发的大规模系统分析、分布式控制设计和仿真工具箱将作为向控制界传播研究成果的工具，并促进向工业从业者进行技术转移。

Recent advances in engineering design have generated a variety of large-scale systems that offer tremendous economic benefits. Many of these systems are composed of multiple individual units, which directly interact with their neighbors. Examples include micro electro-mechanical sensor and actuator arrays, electric power grid, and closely coupled aircraft formation flight. Several limiting factors in designing effective distributed control algorithms for large-scale dynamically-interconnected systems include: (a) the complexity associated with high state dimensions and large numbers of inputs and outputs; (b) the interaction between subsystems through both dynamic coupling and communication; (c) heterogeneity and modeling uncertainty at subsystem level.The strong connection among communication, control and computation has become increasingly important, as the systems grow in complexity and more powerful tools are made available in computing and communications. This project seeks to exploit these connections and develop distributed control techniques for large-scale, dynamically interconnected systems by synthesizing ideas from control theory and optimization techniques. Specifically, game theory will provide a general approach to exploit dynamic interaction between subsystems, and advance control design in the distributed game framework. The distributed control strategy will provide interconnected systems with simplified control implementation and improved system reliability. Moreover, significant performance improvement will be achieved by applying on-line optimizations to such complex systems.Other engineering systems on which the proposed research will have a significant impact include distributed operation of object transportation and manipulation using sensor and actuator arrays, telecommunication network flow, and power distribution over large-scale electric network.The development of innovative educational approach will enhance North Carolina State University educational program by providing undergraduates and graduate students valuable research experience and hands-on practical skills. The developed toolbox on large-scale system analysis, distributed control design, and simulation will serve as a vehicle to disseminate the research outcome to control community, and facilitate technology transfer to industrial practitioners.