東京大学工学部計数工学科 システム情報談話会
日時:2007年6月12日(火) 15:00-16:30
場所:東京大学工学部6号館 3階セミナー室A (本郷キャンパス)
(アクセス) http://www.i.u-tokyo.ac.jp/map/index.shtml#hongo
講師:Pramod P. Khargonekar 教授
(Dean, College of
Engineering, University of Florida)
題目:Stabilization and Output Regulation of Discrete-Time Switched Linear Systems
概要:
A switched system consists of a finite number of
subsystems that are switched according to the time variation of the system's
mode of operation. Such systems can be used to model complex hybrid systems,
where continuous dynamics and discrete state transitions coexist and depend on
each other, and arise in many different contexts such as nonlinear control,
supervisory control, networked control, distributed networks, etc.
This talk consists of two parts. In the first part, we present a
generalization of some of the most important system theoretic concepts to
switched systems under constrained switching. We show that a stabilizing
controller has to depend not only on the current state and mode pair but also to
a certain number of past modes. This shows the limitation of the usual
mode-dependent control approaches, and necessitates the need for path-dependent
notions of detectability and stabilizability. We present new results on a
generalization of the duality concept, path-dependent detectability and
stabilizability concepts, and a separation principle for path-dependent dynamic
output feedback stabilization.
The second part of the talk presents the
extension of the stabilization results to output regulation problems such as the
minimization of the peak output variance and the infinite-horizon linear
quadratic Gaussian control. These two output regulation problems are connected
via a receding-horizon control type problem, where the zero horizon length leads
to the former and the limit as the horizon length approaches infinity yields the
latter. We present exact linear matrix inequality based conditions for the
output regulation performance problem. We will also discuss exact linear matrix
inequality based synthesis conditions for suboptimal controllers. Finally, we
discuss the computational complexity for optimizing the
performance.
This talk is based on joint work with Dr. J.-W. Lee.
連絡先: 原 辰次 (shinji_hara (at)
ipc.i.u-tokyo.ac.jp)
小島千昭 (chiaki_kojima (at) ipc.i.u-tokyo.ac.jp)