研究会 (2011 年 09 月 30 日)
マイマクロ調和ダイナミクス設計論研究会, SICE 九州支部 制御理論と応用に関する研究会 協賛
日時: 09/30(金) 13:30〜18:00
場所: 九州工業大学天神サテライトキャンパス kyutech プラザ
http://www.ims.co.jp/hall/detail/?shopNo=041
http://www.kyutech.ac.jp/information/satellite_campus/index.html
講演1: The role of monotonicity for the stability of interconnected
nonlinear systems
(Dr. Bjoern Rueffer, University of Paderborn, Germany; 13:30〜15:00)
(ビヨン リュッフェル, パダーボルン大学)
講演2: Multi-objective decision-making problems for discrete-time
stochastic systems with state- and disturbance-dependent noise
(Dr. Hiroaki Mukaidani, Hiroshima University; 15:15〜16:30)
(向谷 博明, 広島大学)
講演3: First-ever solution to the problem of integral input-to-state
stability of time delay networks
(Dr. Hiroshi Ito, Kyushu Institute of Technology; 16:40〜17:40)
(伊藤 博, 九州工業大学)
懇親会: 福岡市中央区天神近辺
(18:00〜)
参加者: Bjoern Rueffer, 向谷(広島大), 西村(山口大), 松尾(大分大)
和田, 甲斐(以上九大), 伊藤(九工大)
(以上敬称略)
問合わせ先: 伊藤博 http://palm.ces.kyutech.ac.jp/~hiroshi
Abstract:
1. Interconnections of multiple nonlinear systems in fixed graph
structures appear in timely areas like formation control,
irrigation networks, or distributed model predicted control.
Recently, a number of tools to asses stability of such
interconnected systems has been developed in the literature.
This talk will review so-called small-gain and comparison
principle approaches to the stability problem and show how the
theory of monotone systems provides interesting insights into the
quantitative behavior of the resulting composite system.
Speaker's biographcal sketch:
Bjoern Rueffer (ビヨン リュッフェル) has eceived a PhD degree in
the area of mathematical systems theory at the Center for Applied
and Industrial Mathematics (ZeTeM) within the Department of
Mathematics and Computer Science at the University of Bremen,
Germany. From October 2007 to June 2009 he has been a member of
the Signal Processing Microelectronics (SPM) group and the School
of Electrical Engineering and Computer Science at the University
of Newcastle, Australia. From July 2009 to January 2011 he has
been a Research Fellow within the Department of Electrical and
Electronic Engineering at the University of Melbourne, Australia.
In early 2010 he has undertaken a three months research fellowship
at the Kyushu Institute of Technology, Japan, under the auspices
of the Japan Society for the Promotion of Science (JSPS). Since
February 2011 he is an "Akademischer Rat auf Zeit" within the
Institute of Electrical Engineering at the University of
Paderborn, Germany. Currently, he serves as an associated editor
for the journal Systems & Control Letters.
http://homepages.uni-paderborn.de/rueffer/
2. In this paper, we consider three types of infinite-horizon
multi-objective decision-making problems for a class of
discrete-time linear stochastic systems with state- and
disturbance-dependent noise. First, H_2/H_infinity control problem
with multiple decision makers is considered. Second, in order to
improve the transient response, the linear quadratic control under
the Pareto solution is investigated. Finally, the soft-constrained
stochastic Nash games are formulated in which robustness is
attained against disturbance input. The decision strategies for
the three types of problem are derived. It is found that the
conditions for the existences of these strategies are related to
the solutions of cross-coupled stochastic algebraic Riccati
equations (CSAREs). We develop some new algorithms based on linear
matrix inequality (LMI) to solve the CSAREs. Numerical example is
provided to verify the efficiency of the proposed decision
strategies.
3. Module-based methods are widely used in diverse areas of science
to analyze and build large-scale and complex systems. The distinct
advantage of the module approaches to analysis and design is that
we can deduce the overall behavior of complex systems from
independent observations of simple modules decoupled to each other.
Its superiority to intensive simulation and numerical approximation
is that the module approaches can provide us with qualitative
information as well as quantitative information, which is quite
useful for the purpose of design and understanding the complex
systems. In this talk, we pursue such a philosophy for networks of
nonlinear time-delay dynamical systems by making use of the notion
of integral input-to-state stability (iISS). We allow time-delays
to reside in both subsystems and interconnection channels, and the
time-delays may be both discrete and distributed. No assumption is
made on the interconnection topology of the network. This talk
presents the first solution to the challenging problem and shows
how to construct Lyapunov-Krasovskii functionals establishing the
stability of iISS time-delay networks.
Last modified: Wed Oct 5 11:43:41 JST 2011