Vous êtes sur la page 1sur 13

1338

IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 23, NO. 4, JULY 2015

Finite-Time Attitude Tracking of Spacecraft


With Fault-Tolerant Capability
Bing Xiao, Qinglei Hu, Member, IEEE, and Youmin Zhang, Senior Member, IEEE

Abstract A finite-time attitude tracking control scheme is


presented for rigid spacecraft subject to constant but unknown
inertia and external disturbances. The controller, developed
using sliding mode control technique, has great fault-tolerant
capability to accommodate four types of actuator faults. Different
from most of the existing works on attitude fault-tolerant
control (FTC), the developed controller guarantees the desired
attitude to be followed in finite time, which is critical for FTC
systems. Moreover, the convergence time is an explicit parameter
for designers choice. Thus, the controller design simply meets
the finite-time requirement. The attitude-tracking performance
is evaluated through a numerical example.

oi 3
d 3
e 3
bi 3
(t) 3
v (t) 3
Q 4
Qd  4

Index Terms Attitude tracking, fault-tolerant control (FTC),


finite-time, sliding mode control (SMC), spacecraft.

|| ||
min ()
In
Fi
Fo
Fb
Fd
ac
e
0
bi 3
bo 3

N OMENCLATURE
Euclidean norm of vector or its induced
norm of matrix.
Minimum singular value of a matrix.
n-by-n identity matrix.
Inertial fixed reference frame.
Orbital reference frame.
Body-fixed reference frame.
Desired reference frame.
Distance from the center of the Earth to
the spacecrafts center of mass.
Gravitational parameter
of the Earth.

Orbital rate, (e /ac3 ) 1/2.


General angular velocity in Fb .
Angular velocity with respect to Fo .

Manuscript received March 8, 2014; revised August 13, 2014; accepted


October 13, 2014. Date of publication November 5, 2014; date of current
version June 12, 2015. Manuscript received in final form October 15, 2014.
This work was supported in part by the Heilongjiang Province Science
Foundation for Youths under Grant QC2012C024, in part by the Program for
New Century Excellent Talents in University under Grant NCET-11-0801, in
part by the Natural Sciences and Engineering Research Council of Canada, in
part by the Research Fund for Doctoral Program of Higher Education of China
under Grant 20132302110028, and in part by the National Natural Science
Foundation of China under Project 61273175. Recommended by Associate
Editor X. Zhang.
B. Xiao is with the College of Engineering, Bohai University, Jinzhou
121000, China (e-mail: bxiaobing@gmail.com).
Q. Hu is with the School of Automation Science and Electrical Engineering,
Beihang University, Beijing 100191, China (e-mail: huql_buaa@buaa.edu.cn).
Y. Zhang is with the Department of Mechanical and Industrial Engineering, Concordia University, Montral, QC H3G 1M8, Canada (e-mail:
ymzhang@encs.concordia.ca).
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/TCST.2014.2364124

Qe  4
R(Q) 33 .
R(Qd ) 33
x

Angular velocity of Fo with respect to


Fi , [00 0]T .
Desired angular velocity.
Angular velocity tracking error.
Measured angular velocity in Fb .
Drift-rate bias vector of gyros.
Drift-rate noise vector of gyros.
Unit-quaternion mapping from Fb to
Fo ; Q = [ q0 q T ]T , q 3 ; q02 + qT q = 1.
Desired attitude trajectory representing
by unit-quaternion, denoting the attitude
between Fd and Fo ; Qd = [ qd0 qdT ],
2 + qT q = 1.
qd 3 ; qd0
d d
Error quaternion represents the relative
orientation between Fb and Fd ;
Qe = [e0 eT ]T , e 3 .
Directional cosine matrix associated with
Q; R(Q) = (q02 qT q)I 3 + 2qqT 2q0 q .
Directional cosine matrix associated with
2 qT q )I + 2q qT 2q q.
Qd ; (qd0
d0 d
d d
d d 3
Cross-product matrix associated with
vector x = [ x 1 x 2 x 3 ]T

0
x = x 3
x 2
J 33
u 3
d 3
N
D 3N
ua  N

N
 N
E(t)  NN

x 3
0
x1

x2
x 1 .
0

Symmetric positive-definite inertia matrix.


Total control torque.
External disturbance torque
Number of reaction wheels mounted.
Torque allocation matrix with full-row
rank.
Applied torque by N actuators,
[ u a1 u a2 , . . . , u a N ]T , u ai is the actual
control torque generated by the
i th reaction wheel, i = 1, 2, . . . , N.
Commanded control, [ 1 2 , . . . , N ]T ,
i is the nominal torque commanded by
i th reaction wheel, i = 1, 2, . . . , N.
Uncertain failure, [ 1 2 , . . . , N ]T , i is
the uncertain failure of the i th reaction
wheel, i = 1, 2, . . . , N.
Reaction wheel effectiveness matrix,
diag(h 1 (t), h 2 (t), . . . , h N (t)), 0 h i 1

1063-6536 2014 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.

XIAO et al.: FINITE-TIME ATTITUDE TRACKING OF SPACECRAFT WITH FTC

 3

is the health indicator of the i th reaction


wheel, i = 1, 2, . . . , N.
Attitude Euler angles of Fb with
respect to Fo , [ ]T .
I. I NTRODUCTION

O ACCOMPLISH orbital missions such as Earth


imaging, surveillance, and telecommunication, spacecraft
often needs to perform attitude-tracking maneuvers. This may
require the spacecraft to rotate along a relatively large angular
amplitude trajectory. There have been several important developments in the design of feedback control laws for attitude
tracking [1], [2].
Sliding mode control (SMC) [3] is an effective approach
for uncertain systems with highly coupled nonlinear dynamics. Some of its advantages are rapid response, as well as
insensitivity to uncertain parameters and disturbances. Choi [4]
had investigated a sliding-mode output feedback control for
a class of uncertain multivariable systems. A multiple-model
control-based SMC approach was proposed in [5] for a
two-degree-of-freedom robot. An SMC-based exponential
reaching control law was presented in [6]. The first attempt
of using SMC to achieve large-angle attitude maneuver of
spacecraft was made in [7], and further pursued in [8]. By considering disturbance and uncertain inertia, a sliding-mode
adaptive attitude-tracking controller was developed in [9] for a
rigid spacecraft with thrusters. The attitude-tracking problem
with uncertain inertia was addressed in [10] by presenting a
higher-order SMC law. An adaptive SMC law was synthesized
in [11] to accomplish attitude tracking in the presence of
uncertain inertia. An SMC scheme was reported in [12] to
address attitude stabilization. Actuator saturation and external
disturbance were considered, but the knowledge of inertia
parameters was not needed to implement the controller.
Although many schemes as mentioned previously have been
presented for the problem of attitude-tracking control, there
still remains certain open problems. In particular, reliability
and fault-tolerant capability are the key issues that need to
be addressed. A spacecrafts challenging operating conditions
increase the possibility of faults in actuators/sensors. Once a
spacecraft is launched, it is highly unlikely that its hardware
can be repaired easily. Thus, any component failure cannot
be fixed with replacement parts. Those issues can potentially
cause a host of economic, environmental, and safety problems.
Therefore, the necessity of fault-tolerant capability should be
considered in the spacecraft and its controller design stage.
An accident occurred with GPS BII-07, a spacecraft in the
NAVSTAR GPS constellation developed by the U.S. Department of Defense. It suffered a reaction wheel failure that led
to three-axis stabilization failure and a total loss of the spacecraft [13]. This incident strongly motivates the development of
attitude control systems with reliability and the required performance to be guaranteed even when failures occur. Currently,
such a fault-tolerant control (FTC) system design issue is a
widely developed and rapidly growing research, development,
and application area [14][16].
Extensive studies have been carried out on spacecraft attitude control to possess fault-tolerant capability [17][19].

1339

Automated attitude recovery for rigid and flexible spacecraft


was discussed in [20]. The approach was designed on the
basis of feedback linearization control. Another attitude FTC
design to tolerate loss-of-effectiveness failures of reaction
wheels was reported in [21]. Although attitude tracking was
achieved, disturbance and uncertain inertia parameters were
not considered. Of particular interest of [21] is the view of
actuator faults as certain types of uncertainties, by which
SMC can then be applied. Passive and active SMC laws
were developed in [22] to stabilize attitude with actuator
outage fault accommodated. A control approach was proposed
in [23] for a rigid spacecraft subject to thruster failures.
An adaptive SMC approach was derived in [24] to follow
the desired attitude in the presence of partial loss of actuator
effectiveness.
The problem of attitude tracking with fault-tolerant capability is solved by stability analysis in the aforementioned
SMC-based studies. Those schemes guarantee that the attitude
trajectories converge to the stable equilibrium with infinitesettling time. Obviously, the infinite-settling time criterion is
not an option during critical phases of some high demanding
real-time missions. Consider a military satellite tasked with
providing coverage of a specific high priority area. When
actuator faults occur, if it cannot perform attitude maneuver
and handle faults in finite time, ground objects could be totally
lost. Being inspired by finite-time convergence of dynamic
system, terminal SMC becomes a widely used approach.
A terminal SMC controller was developed in [25] for fault-free
case. The control gain takes infinite values in the vicinity of the
steady state. The problem of rapid reorienting of a spacecraft
with external disturbance and uncertain nature of the dynamics
was investigated in [26]. The attitude was governed to be
ultimately uniformly bounded for cases where there was no
control available on either roll or yaw axis. A control method,
similar to terminal SMC, was proposed in [27] to perform restto-rest attitude maneuver. The design requires exact knowledge
on inertia parameters. Although terminal SMC [28] was used
to achieve satellites formation flying, actuator faults were only
considered in simulation. Stability analysis in the presence of
degradation and stuck failures were not provided. A finitetime attitude-tracking control scheme was proposed in [29] for
spacecraft by using terminal SMC. Chebyshev neural network
was used to approximate unknown nonlinear function and
disturbances. However, the terminal SMC design encountered
singularity problem [30].
Some but not many SMC schemes are available to handle
actuator faults with the desired attitude trajectory followed in
finite time. However, these schemes have four drawbacks.
1) Most of the controllers are only able to tolerate single,
or at most two types of actuator faults.
2) The attitude-tracking error cannot be asymptotically
governed to zero when actuator undergoes faults.
3) There exists a reaching phase for the sliding motion.
Thus, the robustness of the SMC system to disturbance
and uncertainty is significantly reduced.
4) It is not explicit to determine control gains for the finite
time. Analytic relationship between gains and the finitetime is not explicitly presented.

1340

IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 23, NO. 4, JULY 2015

With an effort to tackle the aforementioned four issues,


this paper investigates the feasibility of finite-time tracking
of the desired attitude trajectory in the presence of actuator
faults, external disturbances, and constant but unknown inertia
parameters. The main result to be achieved in this paper is
to extend previous work [31] on globally finite-time control
while approaching the challenging case of taking all three
uncertainties into consideration. An adaptive nonlinear control
technique-based SMC control approach is proposed for such a
solution. The main theoretical contributions of this paper can
be briefly outlined as follows.
1) In comparison with SMC-based finite-time attitude
control scheme such as [25], the proposed approach
has great capability to tolerate actuator faults.
Moreover, when concerning the controller design,
the finite-time explicitly appears in the formulation
of the proposed approach, designers can preliminarily
determine a desired convergent time; that is, no
parameter needs to be tuned to achieve the specified
finite time. However, the method presented in [25]
cannot guarantee this. To the best of our knowledge,
there have been few similar results for spacecraft
reported in the literature.
2) Comparing with conventional SMC-based fault-tolerant
attitude-tracking control such as [23], the designed
controller ensures that attitude-tracking maneuver is
accomplished in a finite time. Moreover, the attitude and
the velocity tracking errors are governed to be zero.
3) In contrast to the previous cited works on
fault-tolerant attitude control with finite-time convergence such as [27], the designed controller can
tolerate all types of reaction wheel (a mostly used
actuator for spacecraft attitude control) faults.
4) As stated in [31], in the conventional SMC, there exists a
reaching phase before system states arrives at the sliding
manifold, and this existence of reaching phase will
reduce the performance robustness of the SMC system.
For the proposed control, although there still exists a
reaching phase, system uncertainties and disturbances
are estimated and compensated using adaptive control
technique. Therefore, unlike the conventional SMC, the
performance robustness of the SMC system will not be
reduced by the proposed scheme.
This paper is organized as follows. Section II presents
the mathematical model used to investigate rigid spacecraft
attitude-tracking control problem and fault modeling for
reaction wheel actuator faults. A finite-time control solution
is presented in Section III. Section IV demonstrates the
application of the proposed FTC scheme to a rigid spacecraft.
The conclusion and future work are given in Section IV.
II. N ONLINEAR M ODEL AND
P ROBLEM F ORMULATION
A. Dynamic Model of Rigid Spacecraft
The coordinate systems used for the rigid spacecraft attitude
control are the inertial frame Fi (X I , Y I , Z I ), the orbit
reference frame Fo (X O , Y O , Z O ), and the body-fixed frame

Fb (X B , Y B , Z B ). The frame Fi is with its origin at the center


of the Earth, and it is used to determine the orbital position of
the spacecraft. The frame Fo , rotating about the Y O axis with
respect to Fi , has its origin located in the mass center of the
spacecraft. The axes of Fo are chosen such that the roll axis
X O is in the flight direction, the pitch axis Y O is perpendicular
to the orbital plane, and the yaw axis Z O points toward the
Earth. The frame Fb has the same origin as Fo , and its axes
coincide with the principal axis of inertia.
With the defined coordinate systems, the dynamic equation
of motion for a rigid spacecraft is described by [32]
J bi =
bi Jbi + u + d.

(1)

From bo = bi R(Q)oi and R(Q)


=
bo R(Q), one

has bo = bi + bo R(Q)oi . Using (1), the nonlinear model


of spacecraft attitude system is given by [29]
1
q0 = qT bo
2
1
q = (q + q0 I 3 )bo
2
J bo =
bo Jbo + H 1 + u + d
where H 1 =
R(Q)oi ).

(J
bo

bo J)R(Q)oi

(2)
(3)
(4)

(R(Q)oi ) J(bo +

B. Attitude-Tracking Error Dynamics


Because Qe = [ e0 eT ]T = (Qd )1 Q denotes the relative
attitude between Fb and Fd , where (Qd )1 = [ qd0 qdT ] and
is quaternion multiplication [33]; Qe can be computed as
follows:
e0 = q0 qd0 + qdT q
e = qd0 q q0 qd + q qd .

(5)
(6)

2 + qT q = 1
It is obtained from (5), (6), q02 + qT q = 1, and qd0
d d
such that

e02 + eT e = 1.

(7)

The directional cosine matrix R(Qe ) that brings Fd onto Fb ,


e ) =
is R(Qe ) = R(Q)R(Qd )T . Applying R(Q
e R(Qe ),
e = bo R(Qe )d , and (4)(6), one can obtain
the equations for the open-loop tracking error dynamics
as follows:
1
e0 = eT e
(8)
2
e = P(Qe )e
(9)

J e = e Je e JR(Qe )d (R(Qe )d ) J e
+ H 1 + u + d (R(Qe )d ) JR(Qe )d
+ J[
d]
e R(Qe )d R(Qe )

(10)

where P(Qe ) = 1/2(e + e0 I 3 ).


Assumption 1: The inertia J is unknown but constant.
There exists a positive scalar (unknown) Jmax such that
||J|| Jmax .
Assumption 2: Disturbances d are unknown but bounded
by an unknown scalar dmax > 0, that is, || d|| dmax .

XIAO et al.: FINITE-TIME ATTITUDE TRACKING OF SPACECRAFT WITH FTC

Assumption 3: The desired angular velocity d and its


first-order derivative are bounded, that is, ||d || c0 ,
|| d || c1 , where c0 and c1 are positive scalars.
Manipulation of (9) and (10) results in
M (Qe )e + C(Qe , e)e + F T (H 1 + H 2 ) = F T (u + d)

(11)

where
F = P1
C(Qe , e) = F T J F F T (JF e ) F
M (Qe ) = F T JF

1341

Fault-Free Case: u ai = i , h i = 1, and i = 0.


F1: For this type of faults, 0 < h i < 1, i = 0.
F2: In the case of this fault, h i goes to 1 and i goes to a
nonzero value of the bias torque, that is, u ai = i + i .
F3: Failure to respond to control signal can be accounted
for h i = 0 and i = 0.
F4: This
fault
can
be
denoted
with
h i = 0 and i = 0.
For a spacecraft with redundant reaction wheels (N > 3)
mounted to guarantee reliability, the total control torque u can
be expressed as the sum of u ai in Fb
u = Dua = D[E(t) + ].

and

(16)

Assumption 4: There exists a scalar g0 > 0 such that


H 2 = (F e) JR(Qe )d + (R(Qe )d ) JR(Qe )d
||

|| g0 .

+ (R(Qe )d ) JF e J[(F e ) R(Qe )d R(Qe ) d ]. (12)


It is worth mentioning that, the uncertain failure is
Remark 1: Obtaining (11) requires that P is invertible. bounded due to physical limitation of reaction wheel. Actually,
det(P) = 0.5e0 (t) = 0 should be guaranteed to avoid the is at least less than the maximum torque generated by
singularity of P for t 0. It thus requires the initial conditions actuator. Hence, Assumption 4 is reasonable.
such that e0 (0) = 0, and the subsequent designed controller
ensuring e0 (t) = 0 for t > 0. Regarding the restriction on D. Problem Statement
the initial conditions, it is known from (7) that the desired
Given any angular velocity and the initial attitude-tracking
trajectory can always be initialized to guarantee e0 (0) = 0. error restricted by e (0) = 0, the control objective can be
0
Therefore, the initial condition restriction is actually a very stated as: Consider the attitude-tracking error system described
mild restriction on the desired trajectory.
by (8)(10) in the presence of actuator fault (15), to design
Remark 2: Note that there are two equilibrium points for a commanded control input to guarantee that e(t) 0,
the closed-loop attitude tracking system, that is, Qe = [ 1 0 ]T (t) 0, and e (t) 1 hold for t t , where t is a
e
0
f
f
and Qe = [ 1 0 ]T . These two equilibrium points represent positive scalar. That is, the desired attitude Q and the desired
d
the same equilibrium point in the physical space and they yield velocity d can be followed in finite time t f .
the same attitude rotation matrix R(Qe ). However, only
Qe = [ 1 0 ]T is an attractive equilibrium point, Qe = [1 0]T
III. F INITE -T IME ATTITUDE -T RACKING
is not an attractor, but a repeller equilibrium [33]. Hence,
C ONTROL D ESIGN
Qe = [ 1 0 ]T is chosen as the equilibrium point to be
To achieve attitude tracking in a finite time even when
stabilized.
actuator faults occur, a control scheme is proposed using
The transformed system (11) has the following properties. SMC technique. Adaptive technique is applied to handle
P1: The matrix M (Qe ) is symmetric positive definite.
constant but unknown parameters and disturbances.
(Qe ) 2C(Qe , e) is skew
P2: The matrix M
For the attitude-tracking error system (11), a novel sliding
(Qe ) 2C(Qe , e))x = 0 manifold is designed as [31]
symmetric [34], that is, xT ( M
for x 3.
(17)
S = [ s1 (t) s2 (t) s3 (t) ]T = e + ke f(t)
P3: The first two terms on the left-hand side of (11) can be
linearly parameterized as [1]
where f (t) = [ f 1 (t) f 2 (t) f 3 (t) ]T is the forcing function in
sliding
dynamics and k is the positive scalar.
(13)
M (Qe )e + C(Qe , e)e = F T W(Qe , e)
An auxiliary attitude-tracking error er is first introduced
where W(Qe , e ) 36 is the known regression matrix
(18)
er = e S.
and 6 is a constant vector given by
Using (13), a desired linear parameterization can be defined as

T
J22
J33
J12
J13
J23
= [ J11
]
(14)
M (Qe )er + C(Qe , e)er = F T W r (Qe , e , er , er ). (19)
with Jij being the i j th element of J.
Then, one can rewrite the open-loop dynamics of S as
T

M (Qe ) S+C(Q
e , e )S = F (u+dH 1 H 2 W r ). (20)

C. Reaction Wheel Faults


As discussed in [35], reaction wheels are vulnerable to four
main faults: F1) Decreased reaction torque; F2) Increased
bias torque; F3) Failure to respond to control signals and
F4) Continuous generation of reaction torque. For each reaction wheel, F1F4 can be mathematically modeled by
u ai = h i (t)i + i (t)

(15)

Because ||R(Q)|| = 1 and ||oi || = 0 , it follows that:


||(R(Q)oi ) JR(Q)oi || Jmax 02
||J
bo R(Q)oi || 0 Jmax ||bo ||

(21)
(22)

||
bo JR(Q)oi || 0 Jmax ||bo ||

(23)

||(R(Q)oi ) J bo || 0 Jmax ||bo ||.

(24)

1342

IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 23, NO. 4, JULY 2015

From (21) to (24), ||H 1 || Jmax 02 + 30 Jmax ||bo || is


obtained.
Equation ||R(Qe )|| = 1 and Assumption 3 ensure that
||e || = ||bo R(Qe )d || ||bo || + c0 . Moreover, the
following inequalities can be established:

||(F e) JR(Qe )d || c0 Jmax (||bo || + c0 )


||(R(Qe )d ) JF e|| c0 Jmax (||bo || + c0 )

||(R(Qe )d ) JR(Qe )d || c02 Jmax


||J[(F e) R(Qe )d R(Qe ) d ]||

(25)
(26)
(27)

f(t) is defined as
fi (t)

ei0 + kei0 ,



t ti


 ,
e

+
ke

sin
i0
i0
i
f
m
 2 ti ti

=

t tim
1

2 (ei0 + kei0 i ) 1 + cos f m ,

ti ti

0,

(29)

Additionally, using Assumptions 24 and |h i (t)| 1,


i = 1, 2, . . . , N, one has

||H 2 +H 1 +D +d|| (3c0 Jmax + 30 Jmax )||bo || + dmax




+ c1 + 02 Jmax + g0 ||D||
(30)

where Y 1 = [ Jmax g0 ]T , Y 2 = [(4c02 + c1 )Jmax + dmax (c0 +


0 )Jmax ]T , 1 = [20 ||D||]T , and 2 = [1 3||bo ||]T. Obviously, 1 and 2 can be exactly obtained, whereas Y1 and Y2
are time-invariant but unknown vectors.
For the mission tasked for the considered satellite, the
attitude-tracking maneuver is required to be accomplished in
f
a finite time ti > 0. Y i and are the estimates of Yi and
= ||||, respectively. The following theorem is presented to
achieve attitude control goals.
Theorem 1: Consider the attitude-tracking system described
by (8)(10) with reaction wheel fault defined in (15).
Design as
2
K c DT FS
||Wr || DT FS
DT FS  T

.
=
Y i i
||FS||
||FS||
||FS||2

ti t tim
tim < t ti
t > ti

where ei0 = ei ( ti ), ei0 = ei ( ti ), i = 1, 2, 3, tim is a scalar


f
f
such that ti < tim < ti , and i is given by

where e = F e is used. H 2 is thus bounded by

+ 4c02 Jmax = Y 1T 1 + Y 2T 2

(34)

Jmax [(||bo ||+c0 )c0 +c1 ] (28)

||H 2 || 3c0 Jmax ||bo || + 4c02 Jmax + c1 Jmax .

t < ti

(31)

i=1

Let Y i and be updated by


Y i = i2 Y i + i ||FS||i , i = K i i , Y i (0) > 0,
i = 1, 2
(32)
= 2 + ||W ||||FS||, = K , (0) > 0 (33)
3
r
3
3 3
3
where i , K i , i = 1, 2, 3 are positive scalars and K c is the positive control gain. Suppose that there exists a positive scalar
such that 0 < < min (DE(t)DT ) at any time. Choose the
f
control gain K c satisfying K c V (0)/ ti , where 0 < < 1
is a scalar specified by the designer, and V (0) is the initial
value of V (t) defined in (A1). Then, all the system states will
reach the sliding manifold in a finite time tT ; moreover, it
f
f
follows that tT ti < ti . Provided that the forcing function



  f
  f
f 
T1i (ei0 +kei0 ) 1+exp k ti tim + 1k ei0 exp k ti ti
f

T1i + T2i exp(k(ti tim )) + T3i exp(k(ti ti ))


(35)
f

with T0i = 4(tim ti )2 /(4k 2 (tim ti )2 + 2 ), T1i =


(1/2k) + (k/2)T4i , T2i = (1/2k) kToi + (k/2)T4i , T3i =
f
f
( Toi /2tim ), and T4i = 4(ti tim )2 /(4k 2 (ti tim )2 + 2 ).
Then, the closed-loop Attitude-tracking system is asymptotically stabilized with e(t) 0 and |e0 (t)| 1 guaranteed for
f
all t > ti . Hence, the spacecraft attitude Q follows the desired
f
trajectory Qd in finite time ti .
Proof: Please refer to the Appendix.
In Theorem 1, a sufficient condition is imposed for the
control design, that is, = min (DE(t)DT ) > 0. Because
D is full-row rank, > 0 means that there are at most
N 3 reaction wheels undergoing F3 or F4 at any time.
Otherwise, it leads to = 0, the controller will not be able
to tolerate the corresponding faults. Furthermore, if N 2 or
more reaction wheels experience F3 or F4 at a same time,
then the remaining active reaction wheels may not be able to
produce a combined torque sufficient enough to compensate
for those faults. Consequently, the spacecraft attitude may not
be stabilized. This results in a three-axis attitude maneuver
failure owing to the lack of necessary hardware redundancy.
For such case, the attitude control system becomes to be
underactuated, which is not the main issue investigated in
this paper. Therefore, much more reaction wheels should be
mounted to fully accomplish three-axis attitude tracking. Also,
this is the reason that N should satisfy > 0 to achieve
three-axis attitude control here.
Now, a practical problem must be considered for
SMC-based control approach, namely, the chattering effect.
Because it is impossible to switch the control at infinite rate,
the trajectory of an SMC system chatters with respect to the
sliding manifold. This chattering phenomenon is practically
undesirable as it may excite the neglected high-frequency
dynamics. One practical approach to reduce the chattering is
to replace the discontinuous function FS/||FS|| in (31) by a
continuous approximation such as FS/(||FS|| + ), where is
a small positive scalar [36]. Therefore, the controller (31) can

XIAO et al.: FINITE-TIME ATTITUDE TRACKING OF SPACECRAFT WITH FTC

be modified as following finite-time FTC (FTFTC):


=

2
K c DT FS
||Wr || DT FS
DT FS  T

. (36)
Yi i
||FS|| +
||FS|| +
||FS||2 +
i=1

It is seen in Theorem 1 that when concerning the determif


nation of controller parameters, the terminal time ti can be
explicitly and preliminarily specified in the proposed design,
and the controller that achieves the desired terminal time can
be easily obtained using the proposed scheme. The choice of
f
f
ti is designer friendly in that the smaller ti is, the attitudetracking maneuver can be accomplished within shorter time.
Summarizing the analyses in the proof of Theorem 1, all the
gains in the controller (36) can be chosen according to the
following procedure.
Step 1: Arbitrarily choose a positive k for the sliding
manifold (17).
f
Step 2: Choose 0 < < 1 and ti , i = 1, 2, 3,
according to the time requirement of the planned
f
aerospace mission. If a smaller ti is selected,
then the mission will be accomplished in a shorter
time.
f
f
Step 3: Choose the value of tim such that ti < tim < ti ,
i = 1, 2, 3. However, there is no unique choice
and physical meaning for the time instant tim ,
and decreasing tim will increase the required control effort during an initial period. Additionally,
decreasing tim will increase the convergence rate
during an initial period. Therefore, a compromise
between control effort and converging rate should
be made by selecting tim.
Step 4: Select the value of . Additionally is also needed
to be less than the value = min (DE(t)DT ).
Because is unknown due to the unknown
fault E(t). Hence, should be chosen as small
as possible, such as 0.01.
Step 5: Choose a positive value of K c such that K c
f
V (0)/( ti ). It should be pointed out that
larger K c will result in larger control effort.
Therefore, K c should be chosen within the maximum torque generated by each actuator.
Step 6: Choose i , K i , i = 1, 2, 3, such that i and K i are
positive. Larger i and K i will lead to faster
convergence of Y i and .
Step 7: Choose a positive according to the set of requirements (such as attitude pointing accuracy and
stability) imposed by the mission. If smaller is
selected, then higher accuracy and more chattering
for the control are obtained.
Remark 3: Although the proof of Theorem 1 finished in
the framework of Lyapunov stability analysis is standard, the
main challenges arising from the issues of finite-time tracking
control, fault-tolerance, external disturbances, and constant
but unknown inertia in spacecraft attitude control are all
addressed simultaneously in this paper. This is the main contribution of the paper in terms of practical engineering application or theoretical contribution. Additionally, the designed
controller is implemented with digital computer in modern

1343

aerospace engineering. Hence, the controller (36) can be


implemented and applied in practice.
Remark 4: Although the controller (31) is developed
according to the sliding manifold presented in [31], the controller has great fault-tolerant capabilities, but the controller
in [31] does not. This is another contribution of this paper
when comparing with the result in [31].
Numerical Example To test the proposed controller, a
rigid spacecraft tasked with attitude tracking maneuver is
numerically simulated. The orbit of the spacecraft is circular,
with an altitude of 750 km and an inclination of 95.4.
= 45 kgm2 ,
The principal moments of inertia are J11

2
2
J22 = 42 kgm , and J33 = 37.5 kgm . The products of
inertia are smaller than 0.5 kgm2. Although it is proved
in Theorem 1 that the proposed controller can only handle constant but unknown inertia parameter, uncertain (timevarying) inertia as given in [23] is simulated to further
demonstrate that its robustness against time-varying system uncertainties. Disturbances d are calculated as in [32].
To accomplish a particular aerospace mission, it initiates a
tracking task commanding the spacecraft to reorient toward
some specific high-priority areas and take a series of
high-resolution images. To take images as much as possible, the attitude-tracking maneuver is required to be
f
accomplished in 10 s, that is, ti = 10, i = 1, 2, 3.
Additionally, the mission further imposes a number of
requirements on the attitude control system. It must provide 0.01 attitude pointing accuracy with 0.0055 deg/s
stability.
The spacecraft is developed as a fully redundant three-axis
stabilized system with four reaction wheels. Three reaction
wheels are fixed orthogonally and aligned with the axis
of Fb , and the fourth redundant wheels are mounted at equal
angle (54.7) to each of the body axes. As a common sensor,
rate-integrating gyros are equipped to measure angular rates.
Nongyroscopic attitude sensors are equipped to measure attitude angles. In practice, it is difficult for the attitude dynamics
to provide a high-precision attitude rate reference. For the
equipped gyros, a widely used model is used [37]
bi = bi + (t) + v (t)

(t) = u (t)

(37)
(38)

where v (t) and u (t) are independent zero-mean Gaussian


white-noise processes with E{v (t) vT (m)} = I3 v2 (t m),
E{u (t)uT (m)} = I3 u2 (t m), (t m) is the
Dirac delta function. Attitude sensors are modeled by
a zero-mean Gaussian white-noise process with standard
deviation ST .
To obtain high-accuracy attitude and rate measurement, an
extended Kalman filter is used in attitude determination. The
control gains for the controller are chosen as: k = 0.25,
i = 7.5, K i = 2, K c = 150, = 0.01, = 0.01, = 0.4,
and tim = 7.5, i = 1, 2, 3. The initial values of the
adaptive updating laws (32) and (33) are selected as:
Y i = [0.05 0.05]T and = 0.15, i = 1, 2. After attitude capture, damping, and stabilization maneuvers, the attitude is within 0.25. Thus, the initial attitude angles in the

1344

Fig. 1.

IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 23, NO. 4, JULY 2015

Desired attitude and distributed target.

Fig. 2. Response of the sliding manifold S in fault-free case. (a) Initial


response. (b) Steady-state behavior.

simulation are chosen as [ ]T = [0.2 0.2 0.2]T


with initial velocity of bo (0) = [0.02 0.02 0.02]T deg/s.
With the physical parameters of the gyros and attitude
sensors equipped on the spacecraft, the gyros measurements are simulated with u = 2.68 104 rad/s3/2,
v = 0.34 rad/s1/2 , and an initial bias 0.08 deg/h
on each axis. The attitude measurement is simulated with
ST = 35 arcsecond.
A. Desired Attitude Trajectory
As stated in [32], the attitude of Fb with respect to Fo can
be obtained by a yaw-roll-pitch ( ) sequence of
rotations, where , , and is the yaw, roll, and pitch angle,
respectively. To ensure that the CCD camera mounted on
the spacecraft can successfully take images of specific high-priority areas, the desired attitude trajectory is
planned as a smooth rotation about each axis by d =
5.4 sin(0 t), d = 6.5 sin(0 t), and d = 12 sin(0 t),
corresponding to the desired unit-quaternion value Qd .
Fig. 1 shows the initial response of the first 20 s,
the circular points denote the desired target in intervals
of 2 s.
B. Control Performance in Reaction Wheel
Fault-Free Case
All reaction wheels are fault-free in this case, and only
external disturbances are acting on the spacecraft. With
the choice of gains, one has tT 4. With the application
of the FTFTC controller (36), the time response of the
sliding manifold S is shown in Fig. 2. It is seen that the
states of the attitude-tracking error system will reach the
sliding manifold within 4 s. This verifies the conclusion
in Theorem 1 that, all the system states will reach the
sliding manifold in a finite time. Figs. 3 and 4 show the
control performance. The desired attitude trajectory is
followed after a short period, roughly 5 s. More specifically,
high-attitude pointing accuracy of 8.0 104 degree
is obtained by the steady-state behavior as shown
in Fig. 3. Fig. 4(b) indicates that the controller achieves the
attitude stability with a value of 1.5 103 deg/s even in

Fig. 3. Attitude-angle tracking errors in fault-free case. (a) Initial response.


(b) Steady-state behavior.

the presence of disturbances and unknown inertia parameters.


Consequently, the attitude system by the presented control
meets the set of stringent pointing requirements to provide
precise geometric accuracy for globally observed highresolution images. The considered images taking mission is
thus successfully accomplished.

XIAO et al.: FINITE-TIME ATTITUDE TRACKING OF SPACECRAFT WITH FTC

1345

Fig. 6.

Fig. 4.

Angular velocity tracking error e in fault-free case.

Fig. 5.

Fault scenarios of reaction wheels.

C. Control Performance in Case of Reaction


Wheel Faults
In this case, fault scenarios shown in Fig. 5 are introduced
and simulated.
1) The reaction wheel mounted in line with the roll axis
of Fb (No. 1 wheel) loses 20% of its normal power
when the attitude maneuver is started.
2) The actuator mounted in line with the yaw axis
of Fb (No. 2 wheel) loses its 45% power in the
time-interval between the fourth and the tenth seconds.
Moreover, it will experience F4 after 10 s, with
i = 0.075 Nm.

Response of the sliding manifold S in case of actuator fault.

3) The reaction wheel fixed in line with the pitch axis


of Fb (No. 3 wheel) undergoes 30% loss of effectiveness in the first 15 s, and then experiences F2 with
i = 0.24 Nm.
4) The redundant wheel (No. 4 wheel) decreases 90% of
its reaction torque after 8 s.
When the controller is implemented to the spacecraft, the
time response of the sliding manifold S is shown in Fig. 6.
It is seen that although actuators are faulty, the proposed
controller still guarantees that the states of the attitudetracking error system will reach the sliding manifold. The
resultant attitude-angle tracking errors and angular-velocity
tracking errors are shown in Figs. 7 and 8, respectively. The
corresponding commanded control torque is shown in Fig. 11.
To compensate for the faults, especially F4 in No. 2 reaction
wheel, extra torques are required. That is why the commanded
signals of all actuators are nonzero even when the desired
attitude is followed by the steady-state behavior of as shown
in Fig. 9(b).
Although inferior attitude-pointing accuracy and stability,
that is, 2.5 103 deg/s and 5.0 103 deg/s, respectively,
were obtained in comparison with the fault-free case
(8.0 104 deg/s and 1.5 103 deg/s, respectively), the
attitude stability and pointing accuracy still satisfy the stringent pointing requirements to provide high-resolution images
with the CCD camera even subject to reaction wheel faults.
Hence, FTFTC still managed to compensate for actuator faults.
Particularly, as can be seen from the initial response of the
velocity tracking error e in Fig. 8(a), the desired attitude
is followed in a finite time of 6 s F4 occurring in the No. 2
reaction wheel after 10 s was also successfully accommodated.
That is because the proposed scheme has a terminal time of
f
an explicit parameter ti = 10. The control objectives are thus
fulfilled in 10 s after the actuator fault occurs. Such a feature
with finite-time convergence is critical for designing practical
FTC systems with respect to the real-time and hard deadline
considerations as outlined in [38].
D. Quantitative Analysis
The control performance obtained from FTFTC is further compared with the indirect adaptive fault-tolerant

1346

IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 23, NO. 4, JULY 2015

Fig. 7. Attitude-angle tracking errors with actuator fault. (a) Initial response.
(b) Steady-state behavior.

control (IAFTC) [23], the finite reaching time-based faulttolerant control (FRFTC) [27], and the conventional finite-time
attitude controller (CFTAC) [25]. In assessing the effectiveness
of these four control schemes, two important criterions should
be considered: the average of square of the commanded control
torque (ASCCT), and the percentage of the successfully taken
images (POSTI) of the high-priority events with respect to the
total high-priority events that need to be imaged
 T0

(t)
2 dt
Nok
, POSTI =
.
ASCCT = 0
T0
Nall
These two indexes state quantitatively how efficient the controller is. Here, T0 is the one orbital period, Nok is the
number of high-priority events that are imaged successfully,
and Nall is the total number of the high-priority events that
need to be imaged. It can be obtained from Section III-A that
Nall = 2995.
The performance index POSTI is compared as shown
in Fig. 10. It is found as follows.
1) When all reaction wheels run normally, the resultant
POSTI is shown in Fig. 10(a). Although FRFTC [27]
and CFTAC [25] can achieve attitude control with finitetime convergence, that finite-time is not represented
by an explicit parameter for users choice with

Fig. 8. Velocity tracking error e with actuator fault. (a) Initial response.
(b) Steady-state behavior.

both controllers, it largely depends on the choice


of the controller gains. If the controller gains are
not appropriately selected, then FRFTC and CFTAC
may lead to a quite large finite-time, within which
attitude maneuver is accomplished. However, the
value of corresponding finite-time for FTFTC can
be preliminarily set by the designer. That is why,
as shown in Fig. 10(a), the POSTI resulted by
FTFTC is much larger than FRFTC and CFTAC.
Additionally, because FRFTC can ensure finite reaching
time for the sliding motion, while CFTAC may
have an infinite-time reaching phase, the POSTI
obtained from CFTAC is inferior to FRFTC.
Additionally, IAFTC [23] can only achieve attitude
control, while finite-time convergence may not be
ensured. Hence, the POSTI resulted by IAFTC is
inferior to other three schemes, although a relatively
high value, that is, 92.75% is obtained, as can be seen
in Fig. 10(a).
2) In the presence of actuator faults, Fig. 10(b) shows the
POSTI resulted from the above four control schemes.
Because CFTAC does not have fault-tolerant capability,
it fails to take even one image of interest owing to the
occurrence of F3 in No. 3 wheel when attitude maneuver

XIAO et al.: FINITE-TIME ATTITUDE TRACKING OF SPACECRAFT WITH FTC

1347

Fig. 9. Commanded control inputs with actuator fault. (a) Initial response.
(b) Steady-state behavior.

starts, that is, POSTI is zero. Although FRFTC can


tolerate F1, it is unable to accommodate F4 in No. 2
wheel. Thus, the mission of image taking cannot be
accomplished after 10 s, and POSTI equaling to 2.00%
is obtained. Not only IAFTC cannot achieve finite-time
control, but it also cannot compensate for F4 and thus
leads POSTI to be 0.07%. For the proposed FTFTC, it
can accomplish attitude-tracking maneuver within finitetime even in the presence of faults. Consequently, it can
accomplish 94.12% of the mission. The resultant POSTI
decreases only 5.78% with respect to the fault-free case,
as shown in Fig. 10(b).
The resultant ASCCT by FTFTC, IAFTC, FRFTC, and
CFTAC is compared (as shown in Fig. 11) and described as
follows.
1) When reaction wheels are fault-free, FTFTC leads to a
smaller ASCCT than other three controllers, as shown
in Fig. 11(a). That is because FTFTC can accomplish
attitude tracking in a short finite-time period. Once the
desired attitude is followed, reaction wheels are only
needed to generate torque to compensate for disturbances.
2) Whether faults occur or not, ASCCT caused by FRFTC
is larger than other three schemes, as shown in Fig. 11(a)
and (b). That is because FRFTC is developed using

Fig. 10. Performance index POSTI in the fault-free and actuator fault cases.
(a) In the absence of reaction wheel faults. (b) In the presence of reaction
wheel faults.

terminal SMC while terminal SMC controller is characterized by larger control power.
3) Because IAFTC, FRFTC, and CFTAC are unable to
tolerate F4 in No. 2 reaction wheel after 10 s, the attitude
control system would continue issuing its maneuver
in spite of F4. The required control effort will thus
quickly saturate the actuator while striving to maintain
the healthy attitude maneuvering performance.
Subsequently, larger control power is resumed.
Thus, larger ASCCT produced by IAFTC, FRFTC,
and CFTAC compared with that of FTFTC is shown
in Fig. 11(b).
From POSTI and ASCCT shown in Figs. 10 and 11, it is
seen that the proposed control strategy provides good performance compared with the controllers presented in [23], [25],
and [27], regardless of reaction wheel faults. The proposed
solution provides a faster response and higher pointing accuracy to guarantee the spacecraft to take much more images,
and thus accomplish the planned mission as much as possible.

1348

IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 23, NO. 4, JULY 2015

A PPENDIX
Proof of Theorem 1: The proof uses Lyapunov stability
theory and is organized as follows: we first prove that all the
system states reach the sliding manifold S = 0 in a finite time.
Then, finite-time convergence of the attitude-tracking error e
is demonstrated.
Step 1 (Finite-Time Convergence of S): Consider a candidate Lyapunov function as
 (Yi Y i )T (Yi Y i )
1 T
S M (Qe )S +
2
2 i
2

V (t) =

2


i=1
1 2 T
K i i Y i Y i

8 i

i=1

K 1 2 2
( )2
+ 3 3 . (A1)
2 3
8 3

With (16), it leaves the error dynamics (20) as


M (Qe )S + C(Qe , e )S = FT D[E(t) + ] + FT d
FT H1 FT H2 FT Wr .

(A2)

Applying the property P2, (30) and (A1) yield




V = ST FT D[E(t) + ] + FT d FT H1 FT H2
+


K 31 3 3 2
(Yi Y i )T Y i
ST FT Wr
4 3
i
i=1

2
( )  K i1 i i YiT Yi

+
3
4 i
i=1

S F DE(t) + ||FS||
T

Fig. 11. Performance index ASCCT in the fault-free and actuator fault cases.
(a) In the absence of reaction wheel faults. (b) In the presence of reaction
wheel faults.

IV. C ONCLUSION AND F UTURE W ORKS


An SMC-based fault-tolerant attitude-tracking control
approach has been proposed for a rigid spacecraft with
the capability of finite-time convergence with zero tracking
errors. Disturbances and unknown inertia parameters have
been addressed, and more types of actuator faults have been
accommodated simultaneously. Moreover, the scheme has the
range of the finite-time period as an explicit and preliminarily
specified parameter with an ease in obtaining the controller
that fulfills the finite-time requirement. It should also be
stressed that, the estimate of the upper bound on spacecraft
inertia, disturbance, and the desired attitude are used to
synthesize the controller, as described in Assumptions 13
and (21)(28). Hence, the controller has certain conservativeness. In fact, the designed controller was inherently a passive
FTC without the need of a fault detection and diagnosis
scheme, and it was characterized by more conservativeness.
As one of future works, one can try to design an active
FTC to reduce such conservativeness. This can be carried out
by proposing a fault detection and diagnosis scheme to identify
fault in real-time, and reconfigure controller on-line by using
the identified model under fault conditions. Experimental study
including designing testbed and testing the proposed algorithm
should also be carried out.

2


YiT i + ||Wr ||||FS||

i=1

K 31 3 3 2
4 3

2

i=1

(Yi Y i )T Y i
i

2
( )  K i1 i i YiT Yi

+
.
3
4 i

(A3)

i=1

Because Y i (0) and (0) are positive, it is obtained from (32)


and (33) that (t) 0 and Y i (t) 0 for t 0, i = 1, 2.
Substituting the control law (31) into (A3) yields
V ST FTDE(t)

2
DT FS  T
||Wr || DT FS K c DT FS

Yi i +
+
||FS||
||FS||
||FS||2
i=1

+ ||FS||

2


YiT i + ||Wr ||||FS||

i=1

2

(Yi Y i )T Y i
i
i=1

2
K 1 3 3 2
( )  K i1 i i YiT Yi

+
+ 3
3
4 i
4 3
i=1

K c + ||FS||

2


(Yi Y i )T i + ( )||Wr ||||FS||

i=1


2
32 2 ( )  (Yi Y i )T Y i K i1 i i YiT Yi

.
4 3
3
i
4 i
i=1
(A4)

XIAO et al.: FINITE-TIME ATTITUDE TRACKING OF SPACECRAFT WITH FTC

Imposing the updating laws (32) and (33) results in


2

i2
2 ( 2 )2
||YiT 2 Yi ||2 3
4 i
4 3
i=1
K c < 0.
(A5)

V K c

Integrating (A5) from 0 to t yields


 t
 t
V d = V (t) V (0)
K c d = K c t.
0

(A6)

It leaves (A6) as V (t) 0 for t (V (0)/K c ) = tT . It is


thus solved from the definition of V (t) in (A1) that, S(t) 0
holds for all t tT . That is to say, all the system states reach
the sliding manifold S = 0 by the finite time t = tT .
Furthermore, from the choice of the control gain K c ,
f
it leaves tT as tT = (V (0)/K c ) ti . This ensures S(t) 0
f
for all t ti . To that end, the finite-time convergence of S
can be concluded.
Step 2 (Finite-Time Convergence of e): As S proved
f
to be zero within finite time ti in Step 1, it follows
from (17) and (34) that
ei + kei

si + ei0 + kei0 ,

(t ti )

,
ei0 + kei0 i sin
f
m
 2(ti ti )

=
t tim

2 (ei0 + kei0 i ) 1+cos f m ,

ti ti

0,

t < ti

ti t tim
tim < t ti

t > ti .
(A7)

To show the finite-time convergence of e, we only need


f
to verify its response in the time interval ti t < ,
because all states absolutely reach the sliding manifold S = 0
f
after time ti . To do this, the following three procedures are
involved.
f
Procedure 1: It starts with the first interval ti t tim.
With the given initial values ei0 and ei0 , solving (A7) over
f
ti t tim yields


 
 
f
f
t ti
t ti
+ C3 cos 
ei (t) = C1 + C2 sin 
f
f
2 tim ti
2 tim ti


f 
(A8)
+ C4 exp k t ti
where
ei0 + kei0
k
C2 = ki T0i
i T0i
C3 =
f
2(tim ti )
ei0 + kC3
.
(A9)
C4 =
k
It is known from (9) that ei always exists for t 0. This
ensures the continuity of ei (t). Then, it is obtained from (A8)
that
 
f 
ei (tim ) = limm ei (t) = C1+C2+C4 exp k tim ti . (A10)
C1 =

t (ti )+

1349

Procedure 2: Owing to the continuity of ei (t), solving (A7)


f
for tim t ti with the initial value (A10) results in




t tim
t tim
+ C7 cos f
ei (t) = C5 + C6 sin f
ti tim
ti tim



f
+ C8 exp k t ti
(A11)
where C5 = ei0 + kei0 i /(2k), C6 = T4i (ei0 + kei0
f
i /2(ti tim )), C7 = k 2 T4i C5 , and C8 = C4 + (C1 C5 +
f
C2 C7 ) exp(k(tim ti )).
With the continuity of ei (t), using (35) and rearranging (A11) lead to
f

ei (ti ) =

lim ei (t) = 0.
f

t (ti )+
f

Procedure 3: For t ti , (A7) can be rewritten as ei +


f
kei = 0. Using ei (ti ) = 0, it shows that ei (t) = 0 for
f
t ti . Further, solving for the unity constraint for Qe yields
f
|e0 (t)| = 1 for t ti . To this end, it is obtained that
f
Qe = [1 0]T or Qe = [1 0]T for t ti . As stated in
Remark 2, only Qe = [1 0]T is chosen as the equilibrium
point for the attitude-tracking error system to be stabilized.
Then, one has Qd = Qd Qe = Qd (Qd )1 Q = Q by using
quaternion multiplication, that is, Qd and Q coincide for
f
all t ti . Hence, the attitude Q follows the desired attitude
f

Qd in finite time ti .
ACKNOWLEDGMENT
The authors would like to thank the reviewers and the editors for their valuable comments and constructive suggestions
that helped to improve the paper significantly.
R EFERENCES
[1] B. T. Costic, D. M. Dawson, M. S. de Queiroz, and V. Kapila,
Quaternion-based adaptive attitude tracking controller without velocity
measurements, J. Guid., Control, Dyn., vol. 24, no. 6, pp. 12141222,
Nov./Dec. 2001.
[2] R. Sharma and A. Tewari, Optimal nonlinear tracking of spacecraft
attitude maneuvers, IEEE Trans. Control Syst. Technol., vol. 12, no. 5,
pp. 677682, Sep. 2004.
[3] J. Zhang and Y. Xia, Design of static output feedback sliding mode
control for uncertain linear systems, IEEE Trans. Ind. Electron., vol. 57,
no. 6, pp. 21612170, Jun. 2010.
[4] H. H. Choi, Sliding-mode output feedback control design, IEEE Trans.
Ind. Electron., vol. 55, no. 11, pp. 40474054, Nov. 2008.
[5] S. Islam and P. X. Liu, Robust sliding mode control for robot manipulators, IEEE Trans. Ind. Electron., vol. 58, no. 6, pp. 24442453,
Jun. 2011.
[6] C. J. Fallaha, M. Saad, H. Y. Kanaan, and K. Al-Haddad, Sliding-mode
robot control with exponential reaching law, IEEE Trans. Ind. Electron.,
vol. 58, no. 2, pp. 600610, Feb. 2011.
[7] T. A. W. Dwyer, III, and H. Sira-Ramirez, Variable-structure control of
spacecraft attitude maneuvers, J. Guid., Control, Dyn., vol. 11, no. 3,
pp. 262270, May/Jun. 1988.
[8] Y.-P. Chen and S.-C. Lo, Sliding-mode controller design for spacecraft attitude tracking maneuvers, IEEE Trans. Aerosp. Electron. Syst.,
vol. 29, no. 4, pp. 13281333, Oct. 1993.
[9] F.-K. Yeh, Sliding-mode adaptive attitude controller design for spacecrafts with thrusters, IET Control Theory Appl., vol. 4, no. 7,
pp. 12541264, Jul. 2010.
[10] C. Pukdeboon, A. S. I. Zinober, and M.-W. L. Thein, Quasi-continuous
higher order sliding-mode controllers for spacecraft-attitude-tracking
maneuvers, IEEE Trans. Ind. Electron., vol. 57, no. 4, pp. 14361444,
Apr. 2010.
[11] Y. Xia, Z. Zhu, M. Fu, and S. Wang, Attitude tracking of rigid spacecraft
with bounded disturbances, IEEE Trans. Ind. Electron., vol. 58, no. 2,
pp. 647659, Feb. 2011.

1350

IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 23, NO. 4, JULY 2015

[12] Z. Zhu, Y. Xia, and M. Fu, Adaptive sliding mode control for attitude stabilization with actuator saturation, IEEE Trans. Ind. Electron.,
vol. 58, no. 10, pp. 48984907, Oct. 2011.
[13] B. Robertson and E. Stoneking, Satellite GN&C anomaly trends,
presented at the 26th Annu. AAS Rocky Mountain Guid. Control Conf.,
San Diego, CA, USA, 2003, pp. 115.
[14] Y.-W. Liang, S.-D. Xu, and L.-W. Ting, TS model-based SMC reliable
design for a class of nonlinear control systems, IEEE Trans. Ind.
Electron., vol. 56, no. 9, pp. 32863295, Sep. 2009.
[15] J. W. Bennett, G. J. Atkinson, B. C. Mecrow, and D. J. Atkinson,
Fault-tolerant design considerations and control strategies for aerospace
drives, IEEE Trans. Ind. Electron., vol. 59, no. 5, pp. 20492058,
May 2011.
[16] H. Kim and H. Lee, Fault-tolerant control algorithm for a four-corner
closed-loop air suspension system, IEEE Trans. Ind. Electron., vol. 58,
no. 10, pp. 48664879, Oct. 2011.
[17] A. de Ruiter, A fault-tolerant magnetic spin stabilizing controller for the
JC2Sat-FF mission, Acta Astronautica, vol. 68, nos. 12, pp. 160171,
2011.
[18] B. Xiao, Q. Hu, and Y. Zhang, Fault-tolerant attitude control for flexible
spacecraft without angular velocity magnitude measurement, J. Guid.,
Control, Dyn., vol. 34, no. 5, pp. 15561561, Sep./Oct. 2011.
[19] P. Kabore and H. Wang, Design of fault diagnosis filters and faulttolerant control for a class of nonlinear systems, IEEE Trans. Autom.
Control, vol. 46, no. 11, pp. 18051810, Nov. 2001.
[20] S. Tafazoli and K. Khorasani, Nonlinear control and stability analysis
of spacecraft attitude recovery, IEEE Trans. Aerosp. Electron. Syst.,
vol. 42, no. 3, pp. 825845, Jul. 2006.
[21] J. Jin, S. Ko, and C.-K. Ryoo, Fault tolerant control for satellites
with four reaction wheels, Control Eng. Pract., vol. 16, no. 10,
pp. 12501258, Oct. 2008.
[22] Y.-W. Liang, S.-D. Xu, and C.-L. Tsai, Study of VSC reliable designs
with application to spacecraft attitude stabilization, IEEE Trans. Control
Syst. Technol., vol. 15, no. 2, pp. 332338, Mar. 2007.
[23] W. Cai, X. H. Liao, and Y. D. Song, Indirect robust adaptive faulttolerant control for attitude tracking of spacecraft, J. Guid., Control,
Dyn., vol. 31, no. 5, pp. 14561463, Sep./Oct. 2008.
[24] B. Xiao, Q. Hu, and Y. Zhang, Adaptive sliding mode fault tolerant
attitude tracking control for flexible spacecraft under actuator saturation, IEEE Trans. Control Syst. Technol., vol. 20, no. 6, pp. 16051612,
Nov. 2012.
[25] Z. Zhu, Y. Xia, and M. Fu, Attitude stabilization of rigid spacecraft
with finite-time convergence, Int. J. Robust Nonlinear Control, vol. 21,
no. 6, pp. 686702, Apr. 2011.
[26] Godard and K. D. Kumar, Robust attitude stabilization of spacecraft
subject to actuator failures, Acta Astron., vol. 68, no. 7, pp. 12421259,
2011.
[27] H. Lee and Y. Kim, Fault-tolerant control scheme for satellite attitude
control system, IET Control Theory Appl., vol. 4, no. 8, pp. 14361450,
Aug. 2010.
[28] G. Godard and K. D. Kumar, Fault tolerant reconfigurable satellite formations using adaptive variable structure techniques, J. Guid., Control,
Dyn., vol. 33, no. 3, pp. 969984, May/Jun. 2010.
[29] A.-M. Zou, K. D. Kumar, Z.-G. Hou, and X. Liu, Finite-time attitude
tracking control for spacecraft using terminal sliding mode and Chebyshev neural network, IEEE Trans. Syst., Man, Cybern. B, Cybern.,
vol. 41, no. 4, pp. 950963, Aug. 2011.
[30] K.-B. Park and J.-J. Lee, Comments on A robust MIMO terminal
sliding mode control scheme for rigid robotic manipulators, IEEE
Trans. Autom. Control, vol. 41, no. 5, pp. 761762, May 1996.
[31] Y.-S. Lu, C.-W. Chiu, and J.-S. Chen, Time-varying sliding-mode
control for finite-time convergence, Elect. Eng., vol. 92, nos. 78,
pp. 257268, 2010.
[32] M. J. Sidi, Spacecraft Dynamics and Control: A Practical Engineering
Approach. Cambridge, U.K.: Cambridge Univ. Press, 1997.
[33] A. Tayebi, Unit quaternion-based output feedback for the attitude
tracking problem, IEEE Trans. Autom. Control, vol. 53, no. 6,
pp. 15161520, Jul. 2008.
[34] S.-C. Lo and Y.-P. Chen, Smooth sliding-mode control for spacecraft
attitude tracking maneuvers, J. Guid., Control, Dyn., vol. 18, no. 6,
pp. 13451349, Nov./Dec. 1995.
[35] S. Murugesan and P. S. Goel, Fault-tolerant spacecraft attitude control
system, Sadhana, vol. 11, nos. 12, pp. 233261, Oct. 1987.
[36] J. Y. Hung, W. Gao, and J. C. Hung, Variable structure control:
A survey, IEEE Trans. Ind. Electron., vol. 40, no. 1, pp. 222,
Feb. 1993.

[37] R. L. Farrenkopf, Analytic steady-state accuracy solutions for two


common spacecraft attitude estimators, J. Guid., Control, Dyn., vol. 1,
no. 4, pp. 282284, 1978.
[38] Y. Zhang and J. Jiang, Bibliographical review on reconfigurable
fault-tolerant control systems, Annu. Rev. Control, vol. 32, no. 2,
pp. 229252, Dec. 2008.
Bing Xiao received the B.S. degree in mathematics
from Tianjin Polytechnic University, Tianjin, China,
in 2007, and the M.S. and Ph.D. degrees in engineering from the Department of Control Science and
Engineering, Harbin Institute of Technology, Harbin,
China, in 2010 and 2014, respectively.
He is currently a Professor with Bohai University,
Jinzhou, China. He has authored or co-authored over
50 papers in journals and conferences. His current
research interests include spacecraft attitude control,
fault diagnosis, fault-tolerant control, robot control,
and mechatronics.
Qinglei Hu received the B.Eng. degree from the
Department of Electrical and Electronic Engineering, Zhengzhou University, Zhengzhou, China, in
2001, and the M.Eng. and Ph.D. degrees with a
specialization in controls from the Department of
Control Science and Engineering, Harbin Institute
of Technology, Harbin, China, in 2003 and 2006,
respectively.
He was a Post-Doctoral Research Fellow with
the School of Electrical and Electronic Engineering,
Nanyang Technological University, Singapore, from
2006 to 2007. From 2008 to 2009, he was with the University of Bristol,
Bristol, U.K., as a Senior Research Fellow. He is currently a Professor with
Beihang University, Beijing, China. He has authored or co-authored over
80 papers in journals and conferences. His current research interests include
variable structure control and applications, spacecraft fault-tolerant control
and applications, and spacecraft formation flying.
Dr. Hu was a recipient of the Royal Society Fellowship Award from 2008 to
2009. He was an Associate Editor of the Aerospace Science and Technology.
Youmin Zhang (SM07) received the B.S., M.S.,
and Ph.D. degrees with a specialization in automatic
controls from Northwestern Polytechnical University, Xian, China, in 1983, 1986, and 1995, respectively.
He is currently a Professor with the Department of Mechanical and Industrial Engineering
and the Concordia Institute of Aerospace Design
and Innovation, Faculty of Engineering and Computer Science, Concordia University, Montreal, QC,
Canada. His current research interests include condition monitoring, health management, fault diagnosis, and fault-tolerant
(flight) control systems, cooperative guidance, navigation, and control of
unmanned aerial/space/ground/surface vehicles, dynamic systems modeling,
estimation, identification, and advanced control techniques, and advanced
signal processing techniques for diagnosis, prognosis, and health management of safety-critical systems, renewable energy systems and smart
grids, and manufacturing processes. He has authored four books, over
330 journal and conference papers, and book chapters.
Prof. Zhang is a Senior Member of the American Institute of Aeronautics and Astronautics (AIAA) and a member of the Technical Committee
(TC) for several scientific societies, including the International Federation
of Automatic Control TC on Fault Detection, Supervision and Safety for
Technical Processes, the AIAA Infotech@Aerospace Program Committee
on Unmanned Systems, the IEEE Robotics and Automation Society TC on
Aerial Robotics and Unmanned Aerial Vehicles, the ASME/IEEE TC on
Mechatronics and Embedded Systems and Applications, and the International
Conference on Unmanned Aircraft Systems (ICUAS) Association Executive
Committee. He has been invited to deliver plenary talks at international
conferences/workshops and research seminars worldwide for over 45 times.
He is the Editor-in-Chief of the Journal of Instrumentation, Automation
and Systems, an Editorial Board Member, and/or Editor-at-Large, Senior or
Associate Editor of six other international journals (including three newly
launched journals on Unmanned Systems). He has served as the General
Chair, the Program Chair, the Program Vice Chair, and a IPC Member of many
international conferences, including the General Chair of the 10th International
Conference on Intelligent Unmanned Systems in 2014, Montreal, Canada, the
Program Chair of the ICUAS in 2014, Orlando, FL, USA, and one of the
General Chairs of the ICUAS in 2015, Denver, CO, USA.

All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately.

Vous aimerez peut-être aussi