This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS: SYSTEMS 1
Two-Phase Attractors for Finite-Duration
Consensus of Multiagent Systems
Mingxuan Sun
Abstract—A specific class of nonlinear systems, admitting
finite-duration attractors (FDAs), is introduced in this paper,
and used for consensus of multiagent systems (MASs). The
advantage is that one can obtain a closed-form solution for the
settling time function of the presented nonlinear systems. Such
a system exhibits a two-phase convergence process, and its set-
tling time function takes values over a finite duration. However,
the duration width changes with respect to the initial condition.
The FDA-based consensus protocol is shown to make the MAS
undertaken achieve finite-duration convergence. With illustrative
examples, the perfect consensus property is verified for MASs
with or without a leader agent.
Index Terms—Finite-time convergence, initial conditions,
multiagent systems (MASs).
I. INTRODUCTION
T
HE CONSENSUS characterizes the cooperative and
collective behavior of multiagent systems (MASs), which
has become an area of intense research activity [1]–[5]. By
applying distributed protocols in leader-following systems,
the agents follow the leader and the leader is independent
of their followers, while the agents are required to achieve
consensus in the leaderless systems. The closely related
topic is about the synchronization phenomena in complex
networks.
Recently, the consensus problem of hybrid MASs was
addressed, in [6] and [7], which are composed of continuous-
time and discrete-time dynamic agents. The adaptive system
approach provides a useful tool to deal with parametric
uncertainties. In [8] and [9], adaptive neural network-based
leader-following consensus approach was proposed, by which
the computation burden is alleviated by only updating a small
number of parameters. This is an obvious contribution because
for obtaining the desired approximation accuracy, the con-
ventional methods often require a large number of adaptive
parameters, and are difficult to implement. In [10], by con-
structing the neural network-based state observer for each
follower, the proposed consensus control method solves the
Manuscript received August 22, 2017; revised November 1, 2017; accepted
December 7, 2017. This work was supported by the National Natural Science
Foundation of China under Grant 61174034 and Grant 61573320. This paper
was recommended by Associate Editor Y.-J. Liu.
The author is with the College of Information Engineering,
Zhejiang University of Technology, Hangzhou 310023, China (e-mail:
mxsun@zjut.edu.cn).
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TSMC.2017.2785314
unmeasurable state problem of high-order nonlinear MASs.
In order to cope with the full state constraints, opportune bar-
rier Lyapunov functions are used in the designed adaptive
systems [11], [12]. Also aiming to decrease the computa-
tion burden, the reinforcement learning technique has huge
potential in developing the adaptive optimal strategies, and
less learning parameter-based policies have been reported
in [13] and [14].
Techniques originating in the conventional Lyapunov syn-
thesis, among others, are applicable to solve the asymptotical
consensus problem. The rate of convergence reflects the con-
sensus performance of protocols. It is therefore desirable for
MASs to possess the finite-time consensus attribute. A finite-
time system exhibits attraction with finite settling time of
theorigin[15], [16]. Typically, in [17], one such system
was formed by a fractional-power function with an expo-
nent less than one, while in [18], it was constructed with the
exponent 1/3. In [19], the reported initial rectified attractors
are initial-condition dependent, similar to those in [16]. One
modification of the finite-time stability, called the fixed-time
stability, was introduced in [20]. It is found that the finite-
ness of the settling time of the two-term attractors holds true
for all initial conditions, which has attracted interest and spe-
cial attention in the past years. It makes sense for protocol
designs so that the finite-time attraction holds. In [21], dis-
continuous finite-time protocols were proposed for first-order
agent networks. Through expanding the terminal sliding mode
technique, the finite-time consensus of multirobot systems was
addressed in [22]. The use of finite-time control ideology in
consensus problems was found in [23] and [24]. Fixed-time
consensus assures finiteness of settling time, regardless of
values what the initial condition takes, which was recently
investigated in [25] and [26]. More fixed-time consensus
results reported are the leader-following consensus [27], [28],
distributed consensus [29], and cluster synchronization [30].
In this paper, we propose a novel class of finite-time non-
linear systems admitting finite-duration attractors (FDAs), and
we concern ourselves with the finite-time consensus problem
of MASs for the further performance improvement. To this
end, the concept of FDAs is introduced in this paper, the fea-
tures of which are twofold: F1) the settling time function to
such FDA attractors can be given, in a similar manner to the
conventional finite-time attractors and F2) the width of the
duration of the settling time varies with respect to the initial
condition, but it is finite, for whatever values the initial condi-
tion takes. According to F1, the settling-time function depends
on the initial condition. The settling time can be computed, as
2168-2216
c
2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.