www.ietdl.org
Published in IET Control Theory and Applications
Received on 5th January 2012
Revised on 27th June 2012
doi: 10.1049/iet-cta.2012.0007
ISSN 1751-8644
New analysis and synthesis conditions for continuous
Markov jump linear systems with partly known
transition probabilities
M. Shen
1
G.-H.Yang
1,2
1
College of Information Science and Engineering, Northeastern University, Shenyang, Liaoning 110004,
People’s Republic of China
2
State Key Laboratory of Synthetical Automation for Process Industries, Northeastern University, Shenyang,
People’s Republic of China
E-mail: mouquanshen@gmail.com; yangguanghong@ise.neu.edu.cn
Abstract: In this study, the stability analysis and synthesis problems for continuous Markov jump linear systems with partly
known transition probabilities are investigated. The partly known transition probabilities cover the cases that some elements
are known, some are unknown with known lower and upper bounds and some are completely unknown. By making full use
of the continuous transition probability matrix property, that is, the sum of transition probabilities is 0 for each row, a new
method for the analysis and synthesis is presented in terms of solvability of a set of linear matrix inequalities. Compared
to the existing results in the literature, it is shown that the proposed method is more effective to deal with the considered
transition probabilities. Numerical examples are given to show the validity of the proposed method.
1 Introduction
Markov jump systems (MJSs), a special class of stochastic
hybrid systems, are suitable mathematical models to describe
systems subject to abrupt variation in their structures or
parameters, which are caused by sudden environmental
changes, component and interconnection failures, parameters
shifting etc. Much effort has been devoted to their widely
practical applications in manufacturing systems, power
systems, aerospace systems and networked control systems
etc. [1, 2].
Owing to the fact that transition probabilities are crucial
to the behaviour of MJSs, it is always assumed that the
transition probabilities are completely known. Based on
this assumption, many important issues have been studied
for this kind of system, such as stability analysis [3–5],
stabilisation [5–7], estimation and optimal control [8–10],
H
2
and H
∞
control [11–22], sliding mode control [23],
sampled-data control [24] and references therein. However,
in practice, transition probabilities may not be measured
exactly or only part of transition probabilities are available,
which may lead to instability or at least degraded
performance just as uncertainties in system matrices do.
No matter in theory or in practice, it is necessary to
consider more general jump systems with partial information
on transition probabilities [25–30]. Specially, the robust
controller design problem for systems involving parameter
uncertainties in the transition probability matrix is presented
by Xiong et al. [27]. In order to guarantee MJSs with
uncertain transition probabilities to remain stable, sufficient
conditions are established to provide perturbation bounds
on the transition probabilities in [26]. The stability and
stabilisation problems are addressed for MJSs with partly
unknown transition probabilities in [28, 29]. Based on [28],
the free-connection weighting matrix method is adopted
and a less conservative stability criterion of MJSs with
the partly transition probability is proposed by He and
co-workers [30]. Introducing a lower bound of unknown
diagonal element, necessary and sufficient conditions for
the stability analysis and stabilisation synthesis problems
are derived for both continuous-time and discrete-time cases
in [29]. Furthermore, the results developed in [29] are
extended to the case that the underlying systems are involved
with non-deterministic switching dynamics [31–34]. An
equivalence between convex polytope uncertainties and
the partly unknown transition probabilities is established
by Goncalves and co-workers [25] for discrete MJSs,
which cannot be employed to deal with continuous MJSs.
Moreover, the result in [29] is not applicable to unknown
transition probabilities with known lower and upper bounds,
and the result in [30] is conservative because of such case
is considered as completely unknown. Although the bound
information is made full use of in [35] for discrete MJSs,
the method cannot directly be applied to continuous MJSs.
This is the motivation for us to study the control problem
for MJSs with the considered partly known transition
probabilities.
The stability analysis and synthesis problems for
continuous MJSs with partly known transition probabilities
are concerned in this paper. The partly known transition
2318 IET Control Theory Appl., 2012, Vol. 6, Iss. 14, pp. 2318–2325
© The Institution of Engineering and Technology 2012 doi: 10.1049/iet-cta.2012.0007