Abstract
〈Vol.6 No.3(2013.5)〉
Titles
[Contributed Papers]
▲
■ On Construction of an H∞ Preview Output Feedback Law
Tokyo Metropolitan University, Tokyo・Kotaro HASHIKURA, Akira KOJIMA,
and Kyoto University・Yoshito OHTA
In preview control problems, efficient use of future reference or disturbance
signals is investigated to improve control performances. This paper addresses
a class of H∞ preview control problems where information on the state
variables and exogenous disturbance is only partially available, and reveals
the clear preview-feedforward structure of the controller. To that end,
the J-spectral factorization technique in the literatures is reviewed with
explicit consideration of the PDE realization of the time-delay element,
and a state-space interpretation is given by proposing new state-variable
transformations. The explicit solution of the control-type operator Riccati
equation is also constructed by the statevariable transformations.
▲
■ Stochastic Extremum-Seeking Algorithm for One-Dimensional and Multivariate
Static Systems
Hokkaido University・Kenta HOSHINO and Yuh YAMASHITA
The aim of this paper is to present a stochastic extremum-seeking algorithm
for one-dimensional and multivariate optimization of static systems. Extremum-seeking
algorithms estimate the optimum value of a function using perturbation
signals. The authors propose three schemes (a basic scheme, an annealing
parameter scheme, and a high-pass filter scheme) for the one-parameter
problem and one scheme (a high-pass filter scheme) for the multivariate
problem.These methods employ Wiener processes for the perturbation signals.
In this paper, the proposed methods are shown to converge by means of a
stability analysis of stochastic systems. The paper presents some numerical
examples to
demonstrate the effectiveness of the methods.
▲
■ A Convex Approach to State Feedback Synthesis for Polynomial Nonlinear Systems with Input Saturation
Meiji University・Hiroyuki ICHIHARA
Ignoring input limitation in control systems often causes closed-loop instability
or performance degradation. This paper proposes a sum of squares approach
to a problem of state feedback controller synthesis for polynomial nonlinear
systems with input saturation. A polytope representation of the saturation
function gives tractable sufficient stability conditions for the problem.
The design procedure follows two steps which are not iterative. In the
first step, the matrix sum of squares relaxation is directory applied to
the problem. In the second step, the annihilator of polynomials decreases
the conservativeness of the first design to expand an estimate of the domain
of attraction of the equilibrium point of the closed-loop system. The result
is an extension of a linear matrix inequality approach, which has been
developed for linear systems with input saturation, to polynomial nonlinear
systems. Numerical examples illustrate the proposed design procedure.
▲
■ Optimal Trajectory Generation for Linear Systems Based on Double Generating Functions
Nagoya University・Zhiwei HAO, Kyoto University・Kenji FUJIMOTO,
and Nagoya University・Yoshikazu HAYAKAWA
Recently, the generating function method for the optimal control problems
was proposed. It provides a family of optimal trajectories for different
boundary conditions. This paper proposes a method to compute a family of
optimal trajectories by using a pair of generating functions. The proposed
method reduces the on-line computational effort in calculating the numerical
integration required in the conventional generating function method. It
is useful for on-line trajectory generation problems.
▲
■ Speeding Up of Calculation for μ-Synthesis of Low Order H∞ Controllers
Hiroshima University・Takuma IWAMOTO, Masami SAEKI, Satoshi SATOH, and
Nobutaka WADA
In this paper, a numerical optimization problem for μ-synthesis with a
reduced order controller is studied. A local solution is searched for starting
from a reduced order controller, where μ-analysis and a local search method
for a reduced order H∞ controller are applied, alternately. A main different
point from the standard DK-iteration is that the procedure is all conducted
for frequency response data and no representation by state equations is
required. Therefore, not the plant order but the number of frequency gridding
mainly affects computation time. In order to develop a more efficient algorithm,
a method of reducing the number of LMIs in the calculation of a descent
direction is given. Numerical examples show that computation time can be
reduced drastically compared with the author’s previous algorithm.
▲
■ Integrity Enhancement of Real-Time Systems by Information-Theoretic
Cryptography
Toshiba Corporation・Shigeki KATAYAMA, Katsuhiko SEKIGUCHI, Kazuto FUKUSHIMA,
and Yokohama National University・Tsutomu MATSUMOTO
This paper describes a concept and a method of effectively applying information-theoretic
cryptography to real-time system communications. Information-theoretic
security, also known as unconditional security, is independent of the computing
power or time an opponent can bring to bear. These properties are suitable
for power systems which require both few computational resources and long
life time. This paper shows some case studies to apply this scheme into
several electrical power system applications, the experimental results
and the examinations of key management also.
▲
■ Investigating the Impact of Trading Frequencies of Market Makers : A
Multi-Agent Simulation Approach
The University of Tokyo・Chi WANG, The University of Tokyo・JST・Kiyoshi
IZUMI,
The University of Tokyo・SPARX Asset Management Co. Ltd.・Takanobu MIZUTA,
and The University of Tokyo・Shinobu YOSHIMURA
This research builds an artificial market, which was composed of stylized
traders and of market makers, to study the market impact under different
trading frequencies of market makers. In addition, the authors implemented
a learning algorithm for market makers to search the optimal trading frequency.
The results indicate that the market makers have a positive effect on reducing
the price volatility, but that, when the market is highly volatile, the
market making strategy can cause a crash of the market. The learning algorithm
may intensify the crash/spike.
▲
■ Automatic Calibration of Sensing Systems for Distributed Physical Fields
Keio University・Tomoyuki FUJINO and Satoshi HONDA
This paper proposes a new automatic calibration method for the sensor networks
which measure the distribution of physical fields, such as in-room thermal
temperature fields. In case of measuring the distribution of room’s temperature,
the regular calibration of the sensors is necessary for obtaining reliable
information. However, it is not an easy task in the case of a large scale
sensor network, because the manual calibration in such a system is time
consuming and costly. To solve the problem, this present study proposes
the new method which is based on the two concepts of regression analysis
and cross validation. The new method is explained and the efficient extensions
are also proposed, and the performance of the proposed methods is verified
by experiments and simulations.
|