Control and estimation in distributed actuator/sensor arrays with application to micro-systems

空间分布动力系统的大阵列在现代技术应用中变得越来越普遍。这些系统可以从宏观的，从编队飞行器、无人机到微镜、微悬臂梁阵列，微机电系统的研究是这类系统研究的重要推动力。其中，构建非常大的传感器和致动器阵列现在是可行的和经济的。人们普遍认为，该领域的下一个前沿之一是这种MEMS的系统级设计，以及它们的集成分布式反馈控制和估计。从系统的角度来看，这些大型执行器/传感器阵列是具有大量输入和输出的耦合动力学系统。传统的控制和估计技术不能直接应用于这些系统，主要原因有两个：（a）由于大规模的计算复杂性是令人生畏的，以及（B）关键的设计考虑是控制和估计的分布式性质，即哪个传感器应该与哪个致动器通信，以及对控制器的本地化和分散化要求。第二个问题，就其一般形式而言，被广泛认为是一个非常困难的问题，在分散控制领域有大约三十年的相关研究。因此，很可能只有利用某些空间分布控制和估计问题所固有的特殊结构才能取得进展。PI最近的工作对于具有某些类型的不变性和对称性的空间分布系统，阐明了解决上述两个问题的有效方法。这些方法也为可能不具有这些对称性的系统提供了指导和基准。我们建议在建模中进行理论和实验相结合的研究计划，分布式控制器的设计和估计的大型网络的传感器和执行器在各种相关的设计约束。这些想法和设计技术将被应用到多微，我们已经在UCSB测试了分布式鲁棒控制系统。所提出的研究的智力价值是双重的。第一是发展具有结构约束的分布式鲁棒控制系统的系统设计方法。这些方法包括：约束包括传感器和控制单元之间传递的信息的本地化。这种架构约束是完全集中式和完全分散的控制，并因此被期望是易于处理的。拟议的研究的另一个重要方面是MEMS问题和控制理论之间的协同作用。集成系统设计正在成为MEMS中越来越重要的问题。我们的工作特别将在两个方面对此作出贡献。第一是使用分布式估计作为简化MEMS阵列中传感器设计的方法，第二个是反馈的使用（电子）简化机械MEMS设计。更广泛的影响：该项目是基于控制理论和MEMS之间的协同作用。控制工程的思想被用来设计从根本上更简单的MEMS系统，MEMS系统中的结构问题提出了新的控制问题。参与这项研究的学生将有兴趣，在动力学和控制以及MEMS方面的学科培训。我们预计，我们的研究结果将使设计高效的MEMS系统，有望成为科技经济的重要组成部分。

Large arrays of spatially distributed dynamical systems are becoming increasingly prevalent in moderntechnological applications.These systems can range from the macroscopic,such as vehicular platoons orUnmaned Aerial Vehicles (UAVs)in formation .ight to the microscopic such as micro-mirror or micro-cantilever arrays.A signi .cant impetus for research on these systems has come from the .eld of Micro-Electro-Mechanical Systems (MEMS)where the construction of very large arrays of sensors and actuators isnow feasible and economical.It is widely recognized that one of the next frontiers in this area is the systemsleve design of such MEMS,as well as their integrated distributed feedback control and estimation.From the systems point of view,these large actuator/sensor arrays are coupled dynamical systems witha large number of inputs and outputs.Traditional control and estimation techniques are not immediatelyapplicable to these systems for two main reasons:(a)the computational complexity due to the large scale isdaunting,and (b)a crucial design consideration is the distributed nature of control and estimation,i.e.whichsensors should communicate with which actuators,and the localization and decentralization requirements onthe controllers.The second problem,in its general form,is widely recognized to be a very di .cult one,withabout three decades of related research in the area of decentralized control.It is thus likely that progress willonly come from exploiting special structures inherent to certain spatially distributed control and estimationproblems.Recent work by the PI has elucidated e .ective methods for addressing both of the above issues forspatially distributed systems with certain types of invariances and symmetries.The methods also provideguidelines and benchmarks for systems that may not posses those symmetries as well.We propose to carry out a combined theoretical and experimental research program in the modelling,distributed controller design and estimation for large networks of sensors and actuators under a variety ofrelevant design constraints.These ideas and design techniques will be applied to the multi-micro-cantileversystems we have been testing at UCSB.The intellectual merits of the proposed research are twofold.The .rst is the development of systematicdesign methodologies for distributed robust control systems which have architectural constraints.These con-straints include the ocalization of information passing between sensors and control units.Such architecturalconstraints are a compromise between fully centralized and fully decentralized control,and are thus expectedto be tractable.Another important aspect of the proposed research is the synergy between MEMS problemsand control theory.Integrated systems design is becoming an increasingly important issue in MEMS.Ourwork in particular will contribute to this in two ways.The .rst is in using distributed estimation as a methodfor simplifying sensor design in a MEMS array,and the second is in the use of feedback (electronically)tosimplify the mechanical MEMS design.Broader impacts:This project is based on a synergistic interaction between control theory and MEMS.Ideas from control engineering are used to design radically simpler MEMS systems,and architectural issuesin MEMS systems are posing new control problems.Students involved in this research will have inter-disciplinary training in both dynamics and controls and in MEMS.We anticipate that our results willenable the design of e .cient MEMS systems which are expected to become an important component of thetechnological economy.