Академический Документы
Профессиональный Документы
Культура Документы
Abstract—In this paper, a stability theorem of nonlinear frac- II. FRACTIONAL DERIVATIVE AND ITS APPROXIMATION
tional-order differential equations is proven theoretically by using
the Gronwall–Bellman lemma. According to this theorem, the
In what follows, we introduce the definition of fractional
linear state feedback controller is introduced for stabilizing a derivative in brief. Fractional calculus is a generalization of
class of nonlinear fractional-order systems. And, a new criterion integration and differentiation to a noninteger-order integro-dif-
is derived for designing the controller gains for stabilization, in ferential operator [2] defined by
which control parameters can be selected via the pole placement
technique of the linear fractional-order control theory. Finally, the
theoretical results are further substantiated by simulation results (1)
of the fractional-order chaotic Lorenz System with desired design
requirements.
Index Terms—Chaos, fractional-order system, linear state feed- where is a complex number, is the real part of , and
back, nonlinear, stability. the subscripts and denote lower limit and upper limit of the
integral operator, respectively.
Formulations of noninteger-order derivatives fall into
I. INTRODUCTION two main classes: Riemann–Liouville derivative and Grün-
ward–Letnikov derivative, on one hand, defined as [1]
RACTIONAL calculus is a 300-year-old mathematical
F topic. Despite of having a long history, it was not used
in physics and engineering for a long period. However, during
(2)
the last ten years or so, fractional calculus has started to attract
or the Caputo derivative on the other, defined as [16]
increasing attention of physicists and engineers from an ap-
plication point of view [1]–[3]. It is found that many systems
in interdisciplinary fields can be elegantly described with the (3)
help of fractional derivatives. Many systems are known to
display fractional-order dynamics, such as viscoelastic systems with , i.e., is the first integer that is not less than ,
[4], dielectric polarization [5], electrode–electrolyte polar- where is the Gamma function. The geometric and phys-
ization [6], electromagnetic waves [7], quantitative finance ical interpretation of the fractional derivatives was given in [16].
[8], and quantum evolution of complex systems [9]. More Here and throughout, only the Caputo definition is used since its
recently, a lot of chaotic behaviors have been shown in the frac- initial conditions take on the same form as for integer-order dif-
tional-order systems [10]–[15]. At the same time, controlling ferential equations. The Laplace transform of the Caputo frac-
fractional-order systems has become one of the most active tional derivative (3) is
fields, especially controlling nonlinear fractional-order sys-
tems. As far as we know, most of the existing control methods
for fractional-order systems only focus on the numerical anal-
ysis but not on the theoretic analysis. In this paper, a stability (4)
theorem of nonlinear fractional-order differential equations is
with , where denotes the Laplace
proven theoretically with Gronwall–Bellman lemma. Then,
operator. Upon considering all the initial conditions to be zero,
based on this stability theorem, a simple criterion is derived for
(4) can be reduced to
designing the gains of the state feedback controller for stabi-
lizing a class of nonlinear fractional-order systems, in which
(5)
control parameters can be selected via the pole placement
technique of the linear control theory.
Therefore, the fractional integral operator of order “ ” can be
represented by the transfer function in the fre-
Manuscript received December 01, 2007; revised April 12, 2008 and June
06, 2008. Current version published December 10, 2008. This paper was rec-
quency domain. The standard definitions of fractional differin-
ommended by Associate Editor O. De Feo. tegral do not allow direct implementation of fractional opera-
X.-J. Wen is with NNPSB, Guangxi Power Grid Corporation, Nanning tors in time-domain simulations. An efficient method to circum-
530031, China (e-mail: wxjsjtu@126.com). vent this problem is to approximate fractional operators by using
Z.-M. Wu and J.-G. Lu are with the Department of Automation, Shanghai
Jiaotong University, Shanghai 200030, China. standard integer-order operators. An effective algorithm was de-
Digital Object Identifier 10.1109/TCSII.2008.2002571 veloped in [17] to approximate fractional-order transfer func-
1549-7747/$25.00 © 2008 IEEE
WEN et al.: STABILITY ANALYSIS OF A CLASS OF NONLINEAR FRACTIONAL-ORDER SYSTEMS 1179
Then, we have
(7)
(9)
(16)
where denotes the eigenvalues of the matrix and
denotes the -norm. where .
Proof: Considering the -dimension matrix , if it has Lemma 2 [19]: If and
different eigenvalues, a proper nonsingular coordinate transfor- is the spectral radius of , then for any
mation can be found. It makes matrix norm
(10) (17)
1180 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—II: EXPRESS BRIEFS, VOL. 55, NO. 11, NOVEMBER 2008
Lemma 3 [20] (Gronwall–Bellman lemma): Assume that: 1) From (23) to (25) and the corollary 1 (9), (22) gives
and are real-valued piecewise-continuous functions
defined on the real interval ; 2) is also real-valued
and ; and 3) and are nonnegative on
this interval. If for all
(18)
(20) (27)
(21)
(29)
where is an identity matrix.
Then, taking Laplace inverse transform for (21) by using the Similarly,
inverse Laplace transform formula of the Mittag-Leffler func-
tion in two parameters and the integral convolution, it yields
(22)
(24)
(32)
(33)
Fig. 1. Attractor of the Lorenz system with fractional order = 0:8.
where , is the state vector,
is a constant linear matrix, is a nonlinear func-
tion, and is a control input to be designed. When where , , , and are four parameters and is the
, there exists unstable behavior, even chaos. Now, we use fractional order. And it can be rewritten as
the linear state feedback to construct the controller as follows:
(37)
(34)
where and
where and .
Theorem 2: For the controlled fractional-order system (33),
if the nonlinear term satisfies
(35)