Professional Documents
Culture Documents
b
Department of Mechanical and Industrial Engineering, Concordia University, Montreal H3G 2W1, Canada
Abstract
A fault tolerant control (FTC) design technique against actuator stuck faults is investigated using integral-type sliding mode
control (ISMC) with application to spacecraft attitude maneuvering control system. The principle of the proposed FTC scheme is
to design an integral-type sliding mode attitude controller using on-line parameter adaptive updating law to compensate for the
effects of stuck actuators. This adaptive law also provides both the estimates of the system parameters and external disturbances
such that a prior knowledge of the spacecraft inertia or boundedness of disturbances is not required. Moreover, by including the
integral feedback term, the designed controller can not only tolerate actuator stuck faults, but also compensate the disturbances
with constant components. For the synthesis of controller, the fault time, patterns and values are unknown in advance, as moti-
vated from a practical spacecraft control application. Complete stability and performance analysis are presented and illustrative
simulation results of application to a spacecraft show that high precise attitude control with zero steady-error is successfully
achieved using various scenarios of stuck failures in actuators.
Keywords: integral sliding mode control; attitude maneuvering; stuck failure; adaptive control; spacecraft
dynamic attitude control laws for spacecraft have been posed for satellite’s attitude control system when the
designed in Refs.[4]-[5]. Sliding mode control (SMC) thruster failures occurred[26]. In Ref.[27], a robust
to certain types of disturbances and uncertainties also FDI method based on neural state-space models was
makes it attractive for spacecraft control problems and applied to a satellite attitude control subsystem, and
many relative works have been attempted in the robustness, sensitivity and stability properties of
Refs.[6]-[8] and the references therein. However, these this method were investigated. Chen and Saif[28] pre-
design methods require the information on the bounds sented a fault diagnosis approach in satellite system
of the uncertainties/disturbances for the computation for identifying thruster faults by using an iterative
of the controller gains. Unlike these methods, nonlin- learning observer. In Ref.[29], the authors used the
ear adaptive control methods do not require these dynamic inversion and time-delay theory to design a
bounds, instead, by including an adaptation mecha- passive fault-tolerant controller for a rigid satellite
nism for tuning the time-varying controller gains. A with four reaction wheels to achieve the attitude
variety of adaptive spacecraft controllers have been tracking control. To take into account the redundant
developed[9-10]. Researches have also been focused on thrusters, an indirect adaptive FTC for attitude
the combination of SMC and adaptive control to de- tracking of rigid spacecraft is proposed in the pres-
velop simple and adaptive robust spacecraft controllers ence of unknown uncertainties, disturbances and ac-
that work for a wide range of practical systems[11-14]. tuator failures, in which a bounded parameter of the
However, the methods mentioned above did not ex- lumped perturbations is introduced to be updated
plicitly investigate the effects of constant disturbance on-line[30]. In Ref.[31], an adaptive sliding mode-
torques on the attitude regulation. Elimination of off- based FTC with L2-gain performance was developed
sets arising out of such disturbances using the linear for the flexible spacecraft attitude control system
and/or nonlinear controllers without integral term re- wherein the persistently bounded disturbances and
quires very high proportional gains that are undesir- unknown inertia parameter uncertainties were explic-
able. A nonlinear proportional-derivative (PD) by in- itly taken into account.
geniously incorporating a modified integral variable Although FTC for spacecraft system has been ex-
was developed by Subbarao[15] through a special tensively studied, to the best knowledge of the authors,
Lyapunov function construction involving quadratic so far few articles have been devoted to an especially
cross state weighting such that globally asymptotical serious fault scenario, stuck actuators. Once stuck, the
attitude convergence to zero was achieved. In Ref.[16], actuators can no longer respond to control signals, and
a backstepping-based sliding mode control scheme it is difficult to deal with stuck actuator failures be-
with integral term was discussed for the flexible cause the remaining actuators must compensate for the
spacecraft attitude system design with external distur- effects of the failed actuators in the overall system.
bances including constant component. However, most Ref.[32] suggested a control reconfiguration method
(if not all) of the previous research work can hardly be using an iterative learning observer to accommodate
extended to the spacecraft control system when the stuck actuators. A reconfigurable control law with ad-
actuator faults are taken into account explicitly, espe- justable parameters is designed using the observed
cially with the kind of stuck failures as additional con- system state information. In Ref.[33], actuator stuck
stant disturbances imposed on the system. faults including the outage of partial actuators are con-
For accommodating actuator faults automatically sidered. Through designing an output feedback con-
to achieve high reliability and availability for space- troller with an additional weighting matrix, the
craft control systems, a control law with fault tolerant closed-loop system is stabilized for both fault-free and
capability, called fault tolerant control (FTC), could faulty cases, and a steady-state-based fault detection
be a desirable one. FTC is an area of research that approach is proposed such that arbitrary small stuck
emerges to increase availability by specifically de- actuator faults can be detected effectively. Stuck ac-
signing control algorithms capable of maintaining tuator faults have also been treated as constant distur-
stability and performance despite the occurrence of bances in Ref.[34] where a proportional-integral (PI)
faults, and has received considerable attention from controller is used to reject their effects. An H con-
the control research community and aeronautical troller is designed and an iterative linear matrix ine-
engineering in the past couple of decades[17-18]. The quality (LMI)-based algorithm is used in Ref.[35]. The
available design techniques include the linear quad- design is such that the nominal performance is opti-
ratic control method[19], adaptive control[20-21], ei- mized and the closed-loop system performance under
genstructure assignment [22] and SMC [23-24] , to different fault modes is acceptable.
name a few. As for the application of FTC to In this article, an attempt is made to provide a FTC
spacecraft attitude control system design, in strategy for the spacecraft with redundant actuators,
Ref.[25], Boškoviü, et al. used multiple-model such as four or more thrusters which are commonly
method to detect and isolate actuator faults for used for attitude control, to address the aforementioned
spacecraft attitude control system. Based on dynami- issues. The proposed control strategy is based on an
cally driven recurrent neural network architecture, a adaptive integral sliding mode control theory and it is
fault detection and isolation (FDI) strategy was pro- applied to the spacecraft suffering from unknown
· 34 · HU Qinglei et al. / Chinese Journal of Aeronautics 24(2011) 32-45 No.1
q3]TęR3, ZęR3 is the angular velocity of a Ff ,i being a zero or nonzero constant which denotes
body-fixed reference frame of a spacecraft with re- the stuck value of the ith actuator, Ff,i and Ff ,i are
spect to an inertial reference frame expressed in the
known scalars, and the vector u denotes the designed
body-fixed reference frame, I the identity matrix with
propulsion force vector. The diagonal matrix E satis-
proper dimensions, and q× a skew-symmetric matrix
fies
which satisfies
E diag(e1 , e2 ,", el ) (5)
ª 0 q3 q2 º
qu «« q3 0 q1 »» (2) with ei=0 or ei=1 (i=1,2,…,l). Obviously, ei=0 for any
1il, the fault model corresponds to the case, when
«¬ q2 q1 0 »¼
the ith thruster gets faulty, particularly, if the stuck
Note that the unit quaternion is subject to the con- value Ff,i=0, the model denotes an important case:
straint qTq+q0=1. It can be easily verified from the outage. If ei=1, the ith actuator is fault-free.
preceding definition that In view of the stuck failure given by Eq.(4), the sys-
ªq º tem Eq.(3) can be rewritten as
) T) I3 , )T « 0» 0 J Z Z u ( J Z DEu D( I E ) Ff Td (6)
q
¬ ¼
These properties will be used in the following de- Note that the effect of stuck actuators can be con-
velopment. Note that since the unit quaternion pa- sidered as additional constant disturbances imposed on
rameter set [q0 qT]T given in Eq.(1) is redundant, a the system, which may drive the attitude away from
given physical attitude for a rigid body will have two the desired position. The closed-loop system stability
mathematical representations, where one of these in- may also be affected due to the loss of some control
cludes a rotation of ±2ʌ about an axis relative to the channels. In order to return to its original position, the
other. Therefore, if [1 0T]T represents the desired remaining actuators must be adjusted accordingly, to
equilibrium point, then [í1 0T]T represents the same counteract the effect of the stuck actuators and the
attitude after a rotation of ±2ʌ about an arbitrary axis. change in the dynamics as the result of such a failure.
In the following, only [1 0T]T is considered as the In order to develop the controller, the transformation
desired attitude equilibrium point for the controller technique addressed in Ref.[37] is used here for the
design synthesis. system Eq.(6), and then the following dynamic equa-
tions are given as follows:
2.2. Dynamic equation with actuator stuck failure
J * (q, q0 )q C * (q , q, q0 )q
The dynamic model of a spacecraft is governed by 1 1
PDEu P[Td D ( I E ) Ff ] (7a)
the following differential equation[36]: 2 2
No.1 HU Qinglei et al. / Chinese Journal of Aeronautics 24(2011) 32-45 · 35 ·
the general nonlinear equations of motion for space- where KP and KI are determined such that the sliding
craft with the possible actuator stuck failures and ex- mode on S=0 is stable, i.e., the convergence of S to
ternal disturbances, which will be used in the zero in turn guarantees that q and q also converge to
following controller synthesis. zero. Note that any positive definite KP and KI will
The model in Eq.(7a) has the following properties: satisfy this condition. If KI=0 is selected, this kind of
Property 1 The matrix J* is symmetric positive sliding surface becomes the conventional linear sliding
definite and the matrix J * (q, q0 ) 2C * (q , q, q0 ) is surface, S q K P q , as stated in the literature. More-
skew-symmetric, that is over, the additional integral provides one more degree
x T ( J * (q, q0 ) 2C * (q , q, q0 )) x 0, x R 3 (8) of freedom in design than the conventional linear slid-
ing surface. In addition, the introduced integral term
Property 2 The inertia matrix is bounded, will help to reduce the effect of constant disturbances,
i.e., ýJýa0, where a0 is unknown constant, and which will be discussed in later section.
there also exists an unknown scalar a1˚0 such that the For convenience of control law design, a new vari-
following inequality able is introduced as[38]
qr q S (12)
|| C * ||d a1 || q || (9)
Then from the definition of S, we have
is satisfied, where ý•ýdenotes the Euclidean norm. t
Property 3 With the disturbances considered, it is qr K P q K I ³ qdW (13)
0
reasonable to assume that Td(t) is bounded and satis-
By rearranging Eq.7(a), the following can be ob-
fies[36]
tained
|| Td ||d a2 a3 || q ||2 (10) 1
J * S C * S PDEu
with a2>0 and a3>0 unknown but constant. 2
1
P[Td D( I E ) Ff ] ( J *qr C *qr ) (14)
3. Fault Tolerant Attitude Controller Design with 2
Stuck Failures This nonlinear equation will be used to prove stabil-
ity of the closed-loop system.
In this section, we present the design procedure to
implement adaptive integral sliding mode-based FTC 3.2. Adaptive FTC law design
for spacecraft attitude control system. Adaptive control
technique deals with situations in which some of the The basic idea is to alter the system dynamics along
parameters are unknown or slowly time varying, and the sliding surface, such that the trajectory of the sys-
the basic idea in this method is to estimate these un- tem is steered onto the sliding manifold described by
known parameters online and then use the estimated S=0. As stated in Property 2 and Property 3, the system
ones in place of the unknown ones in the feedback parameters, disturbances and the stuck failures are
control law. The overall design can be divided into two assumed to be bounded and, therefore, for the conven-
main steps. Step 1 involves the construction of a slid- ience of the controller development, the following
ing surface, containing integral term to ensure that, variables can be introduced:
once the system is restricted to the sliding surface, the
spacecraft can be expected to be in the desired position. Y1 >|| (qr O S ) ||
|| qr |||| q ||@ ½°
T
(15a)
¾
Step 2 entails the derivation of parameter adaptation 41 [a0 a1 ]T °¿
laws and feedback control gains that can drive the
|| P || ½
spacecraft attitude to the sliding surface and maintain Y2 [1 || D || || q ||2 ]T °
it in the manifold. 2 ¾ (15b)
4 2 [a2 || Ff || a3 ]T °
¿
3.1. Integral-type sliding surface design
where O˚0 is given constant to specify the speed of
The SMC design starts with building a sliding sur- convergence of the system.
face in the system state space. The motion of the sys- To achieve the sliding motion, the following FTC
tem along the sliding mode is expected to meet the law is proposed:
· 36 · HU Qinglei et al. / Chinese Journal of Aeronautics 24(2011) 32-45 No.1
°1 X0
2 2
1
¯ ¦ Di || Yi || || S || ¦ 4GJ D i2 Ei2 d
i 1 i 1 i
We then have the following statement. 1 1
Theorem 1 Consider the spacecraft attitude con- O S T J * S S T PDEu || S || || P || || Td ||
trol system in Eq.(7a) with the integral-type sliding 2 2
surface given in Eq.(11). If the control laws in 1
|| S || || P || || D || || I E || || Ff || || S || || J * || || (q O S ) ||
Eqs.(16)-(17) are implemented with the proper pa- 2
rameters, then the closed-loop system is globally as- 2
1 2 2 2 1
ymptotically stable, and the attitude and velocity respec- || S || || C * || || qr || ¦ D i E i ¦ D i E i2D i
i 1 GJ i i 1 GJ i
tively converge to zero, i.e., lim q o 0 , lim q0 o 1 and 2 2
tof t of 1
lim Z o 0 . ¦ Di || Yi || || S || ¦ 4GJ D i2 Ei2 (22)
tof i 1 i 1 i
Proof Define a Lyapunov function candidate
In view of the Property 2 and Property 3 and the
2 2 K Ei1 definition in Eq.(15), further simplification of Eq.(22)
1 T * 1
V S J S ¦ Di2 ¦ D i2 Ei2 (19) leads to
2 i 1 2GJ i i 1 8GJ i
1
where G >0 is some unknown constant but less than the V d O S T J * S S T PDEu || S || || Y1 || || 41 ||
2
minimum eigenvalue of matrix DEDT and Di denotes 2
2 2 K 1
1 1 Substituting the control law given by Eq.(16) into
S T J * S S T J * S ¦ DiDi ¦ i Di2 Ei Ei
E
V
2 i 1 GJ i i 1 4GJ i
the inequality Eq.(23), we can further obtain
1 1 ½ 2
S T ® PDEu P[Td D( I E ) Ff ] ( J *qr C *qr ) ¾ V d O S T J * S G S T KS ¦ GDˆ i || Yi || || S ||
¯ 2 2 ¿ i 1
2
1 T * 1
2 2
S ( J 2C * ) S ¦ Di Ei2Dˆi J i || Yi || || S || ¦D || Yi || || S || ¦ D i || Yi || || S ||
i 1 GJ i
i
2 i 1 i 1
2 K Ei1 1 1
2
Ei2 §¨ D i Di ·¸ d
2
V(f) exists for some finite V(f)ęR+. Also from Remark 5 In this control law, an integral feedback
Eq.(23) and boundedness of all signals within subse- is involved, which will achieve zero steady-error in the
quent time derivative of V(t), it is easy to establish that presence of constant disturbance torques or stuck fail-
V(t ) Lf , or, in other words, uniform continuity ures. A detailed discussion of the ability of the integral
feedback to reject constant input disturbances specifi-
for V (t ) . This result, in conjunction with the conver- cally for the rigid body attitude control problem can be
gence of V(t) to V(), permits application of Barbalat’s found in Ref.[40].
lemma (using the alternative statement of this lemma Remark 6 In order to avoid the chattering phe-
from Ref.[39]) to provide V (t ) o 0 as tĺ. This al- nomenon due to the imperfect implementation of the
lows us to go further and conclude that lim S 0 and sign function in the control law of Eq.(16), the fol-
t of lowing saturation function
then sliding condition can be guaranteed. From the 1 X !H
definition of the sliding surface, the stability of the °
sliding surface guarantees that the variables q and q sat X ®X |X | d H (26)
° 1 X H
will also converge to zero. Consequently from the ¯
unit-norm constraint on the unit quaternion, we can is a simple choice to replace the discontinuous func-
obtain that lim q0 z 0 (more precisely lim q0 1 is tion, where H >0 is a small constant. Note that when
t of t of
the saturation function is introduced, the uniformly
considered here). Then using the identity stated in ultimately bounded stability will be achieved for the
Eq.(2b), it follows that lim Z 0 . Thus we show that closed-loop system. In the next section, numerical
t of
simulation and comparison are given to verify the
lim [q Z S] 0 (25) success of the integral-type sliding mode control
t of
(ISMC)-based FTC law in conjunction with the adap-
thereby completing the proof of achieving the stated tive control technique.
control objective.
Remark 1 From the proceeding analysis, the con- 4. Simulation and Comparison Results
trol law does not need the knowledge of inertia matrix,
disturbances and/or their upper bounds for implemen- To study the effectiveness and performance of the
tation. This fact shows that it is robust to the inertia proposed control strategy the detailed response is nu-
matrix, which may itself be subject to uncertainties, merically simulated using the set of governing equa-
the external disturbances. The parameters Dˆi (i=1, 2) tions of motion Eqs.(1)-(6) in conjunction with the
proposed control laws Eqs.(16)-(17) and Eq.(26). The
are estimated on-line using the adaptive algorithm in
spacecraft parameter and the external disturbances
Eq.(17) starting from any initial value.
used in the numerical simulations are shown in Table 1.
Remark 2 From the designed adaptive laws in
Note that for all numerical examples considered in this
Eq.(17), a low-pass-filter form of the parameter update
section, the net disturbance torque acting on the sys-
law, which makes suitable corrections when the pa-
tem is viewed to be time varying plus constant parts
rameters, disturbances and faults are overestimated,
due to gravitation, solar radiation, magnetic force and
ensures that the parameters Dˆi (i=1, 2) are bounded. aerodynamic drags.
Remark 3 From the designed control law in In the simulation, four thrusters are assumed to be
Eqs.(16)-(17), the health condition matrix E and the distributed on the side face of the spacecraft in each
stuck failure Ff are not involved in the control scheme, corner of the square[41]. The body-fixed y-z plane is
implying that the proposed control law is able to orthogonal to the body x axis and at a distance d along
achieve the control objective regardless of the thruster the –x axis from the center of mass, and the side length
health condition as long as the remaining active of the thruster assembly is 2c. Thus, the moment arms
thrusters are capable of producing the combined forces in the spacecraft body axes are
sufficient enough to allow the spacecraft to perform a ª d º ª d º ª d º ª d º
given maneuvering.
r1 « c » , r2 « c » , r3 « c » , r4 «« c »»
« » « » « »
Remark 4 From the proceeding proof, although
the parameter G, less than the minimum eigenvalue of ¬« c ¼» ¬« c ¼» ¬« c ¼» ¬« c ¼»
matrix DEDT, is involved in stability analysis, an ana- (27)
lytical estimate of this parameter is not needed because To achieve attitude controllability, the thruster direc-
the proposed control algorithms do not involve such a tions are canted from the –x axis by - =5° to produce
parameter. But the full-rank requirement for the actua- control moment along the x axis. The direction of force
tor distribution matrix D should be satisfied such that generated by each thruster in the body-fixed y-z plane
the stability is ensured only if the matrix DEDT is posi- is also different from the principle axes by 45°. As a
tive definite. This requirement can be easily achieved result, control torques along different axes can be gen-
by properly placing the thrusters on the spacecraft[30]. erated and the distribution matrix is indeed:
· 38 · HU Qinglei et al. / Chinese Journal of Aeronautics 24(2011) 32-45 No.1
method since no integral control term is involved in stant when the thruster is subject to abrupt outage
the feedback loop. From the comparison, the perform- fault.
ance of the proposed design is better than the other
two controllers even if these designs will adapt to the
system parameters under the external disturbances.
Fig.5 Time response of thruster control forces under thrust outage case.
No.1 HU Qinglei et al. / Chinese Journal of Aeronautics 24(2011) 32-45 · 41 ·
Fig.8 Time response of thruster control forces under stuck failure case I.
· 42 · HU Qinglei et al. / Chinese Journal of Aeronautics 24(2011) 32-45 No.1
is seen that fairly good control performance is simulation is conducted corresponding to the stuck at
achieved under this severe scenario even though more í1.0 in value. Similar responses compared with
settling time (about 100 s) is required due to the con- Figs.7-9 are obtained, so they are omitted here.
stant-like disturbance for the proposed controller; for As can be seen from the simulation results in the
the ASMC, severe oscillations can be observed and above two cases under stuck failures, stuck failures are
also with large steady-state error in the range of 0.01 even more difficulty to be handled compared with the
in value; while for PID case, it exhibits poor perform- outage failure case since their adverse effects add con-
ance in maneuvering even if the integral control term stant disturbances to the system. For more severe fail-
is introduced in the feedback loop, and no fault toler- ures, better performances of the proposed AISMC
ance design is considered explicitly in the system de- scheme over ASMC and PID have been achieved.
sign. Fig.9 also presents the adaptive parameter esti- Summarizing all the cases (normal case and fault
mates when the abrupt stuck occurs at t=5 s. In addi- cases), it is noted that the proposed controller design
tion, to examine the sensitivity of the controller to the method can improve the normal performance more
stuck at a certain position in inverse direction, simula- significantly than the PID or ASMC method in both
tion is done corresponding to the stuck at í0.5 in value. theory and simulations. Also, in fault cases simulated
Similar responses compared with Figs.7-9 are obtained, in this article, the proposed method has better results
so they are omitted here for space limitation. than those of existing controllers. It can also be ob-
served that as more severe fault cases are considered in
the design, the proposed controller can still guarantee
system to be stable. In addition, extensive simulations
are also conducted using different control parameters,
disturbance inputs and parameter uncertainties. These
results show that in the closed-loop system, attitude
control is accomplished in spite of these undesired
effects in the system. Moreover, the flexibility in the
choice of control parameters can be utilized to obtain
desirable performance while meeting the constraints
on the control magnitude. These control schemes pro-
vide a theoretical basis for the practical application of
the advanced control theory to future spacecraft atti-
tude control system design.
Fig.11 Time response of thruster control forces under stuck failure case II.
restriction explicitly on the magnitude of the desired Journal of Aerospace Engineering 2009; 22(2): 122-
control, and the design with explicitly considering the 135.
actuator saturation limit is also under investigation. [17] Zhang Y M, Jiang J. Bibliographical review on recon-
figurable fault-tolerant control systems. Annual Re-
Future work is planned to study the digital implemen-
views in Control 2008; 32(2): 229-252.
tation of such control scheme on hardware platforms [18] Benosman M. A survey of some recent results on
for attitude control experimentation. nonlinear fault tolerant control. Mathematical Pro-
blems in Engineering 2010; 2010: 1-25. DOI: 10.1155/
References 2010/586169.
[19] Hsieh C S. Performance gain margins of the two-stage
[1] Singh S N. Rotational maneuvers of nonlinear uncer- LQ reliable control. Automatica 2002; 38(11): 1985-
tain spacecraft. IEEE Transactions on Aerospace Elec- 1990.
tronic and Systems 1988; 24(2): 114-123. [20] Ye D, Yang G H. Adaptive fault-tolerant tracking con-
[2] Nagata T, Modi V J, Matsuo H. Dynamics and control trol against actuator faults with application to flight
of flexible multi-body systems Part II: simulation code control. IEEE Transactions on Control Systems Tech-
and parametric studies with nonlinear control. Acta As- nology 2006; 14(6): 1088-1096.
tronautica 2001; 49(11): 595-610. [21] Tang X D, Tao G, Joshi S M. Adaptive actuator failure
[3] Karray F, Grewal A, Glaum M, et al. Stiffening control compensation for nonlinear MIMO systems with an
of a class of nonlinear affine systems. IEEE Transac- aircraft control application. Automatica 2007; 43(11):
tions on Aerospace Electronic and Systems 1997; 33(2): 1869-1883.
473-484. [22] Zhang Y M, Jiang J. An active fault-tolerant control
[4] Tsiotras P. Further passivity results for the attitude system against partial actuator failures. IEE Proceed-
control problem. IEEE Transactions on Automatic ings Control Theory and Applications 2002; 149(1):
Control 1998; 43(11): 1597-1600. 95-104.
[5] Hokayem P F, Schilling K. Input-to-state stable attitude [23] Alwi H, Edwards C. Fault detection and fault-tolerant
control. Journal of Guidance, Control, and Dynamics control of a civil aircraft using a sliding-mode-based
2008; 31(6): 1826-1829. scheme. IEEE Transactions on Control Systems Tech-
[6] Hu Q L, Ma G F. Variable structure control and active nology 2008; 16(3): 499-510.
vibration suppression of flexible spacecraft during at- [24] Kim D, Kim Y. Robust variable structure controller
titude maneuver. Aerospace Science and Technology design for fault tolerant flight control. Journal of
2005; 9(4): 307-317. Guidance, Control, and Dynamics 2000; 23(3): 430-
[7] Iyer A, Singh S N. Variable structure slewing control 437.
and vibration damping of flexible spacecraft. Acta As- [25] Boškoviü J D, Li S M, Mehra R K. Intelligent control
tronautica 1991; 25(1): 1-9. of spacecraft in the presence of actuator failures. Pro-
[8] Hu Q L. Sliding mode maneuvering control and active ceedings of the 38th IEEE Conference on Decision and
vibration damping of three-axis stabilized flexible Control. 1999; 4472-4477.
spacecraft with actuator dynamics. Nonlinear Dynam- [26] Li L, Ma L Y, Khorasani K. A dynamic recurrent neural
ics 2008; 52(3): 227-248. network fault diagnosis and isolation architecture for
[9] Singh S N, Araujo A D. Adaptive control and stabiliza- satellite’s actuator/thruster failures. Lecture Notes in
tion of elastic spacecraft. IEEE Transactions on Aero- Computer Science 2005; 3498(3): 574-583.
space and Electronic Systems 1999; 35(1): 115-122. [27] Wu Q, Saif M. Robust fault diagnosis for a satellite
[10] Maganti G B, Singh S N. Simplified adaptive control large angle attitude system using an iterative neuron
of an orbiting flexible spacecraft. Acta Astronautica PID observer. Proceedings of the American Control
2007; 61(7-8): 575-589. Conference. 2006; 5710-5715.
[11] Singh S N, Zhang R. Adaptive output feedback control [28] Chen W, Saif M. Observer-based fault diagnosis of
of spacecraft with flexible appendages by modeling satellite systems subject to time-varying thruster faults.
error compensation. Acta Astronautica 2004; 54(4): Journal of Dynamic Systems, Measurement, and Con-
229-243. trol 2007; 129(3): 352-356.
[12] Hu Q L, Shi P, Gao H J. Adaptive variable structure [29] Jin J H, Ko S H, Ryoo C K. Fault tolerant control for
and commanding shaped vibration control of flexible satellites with four reaction wheels. Control Engineer-
spacecraft. Journal of Guidance, Control, and Dynam- ing Practice 2008; 16(10): 1250-1258.
ics 2007; 30(3): 804-815. [30] Cai W C, Xiao X L, Song Y D. Indirect robust adaptive
[13] Zeng Y, Araujo A D, Singh S N. Output feedback vari- fault-tolerant control for attitude tracking of spacecraft.
able structure adaptive control of a flexible spacecraft. Journal of Guidance, Control, and Dynamics 2008;
Acta Astronautica 1999; 44(1): 11-22. 31(5): 1456-1463.
[14] Hu Q L, Ma G F. Adaptive variable structure maneu- [31] Hu Q L. Robust adaptive sliding-mode fault-tolerant
vering control and vibration reduction of three-axis sta- control with L2-gain performance for flexible space-
bilized flexible spacecraft. European Journal of Control craft using redundant reaction wheels. IET Control
2006; 12(6): 654-668. Theory & Applications 2010; 4(6): 1055-1070.
[15] Subbarao K. Nonlinear PID-like controllers for [32] Chen W, Jiang J. Fault-tolerant control against stuck
rigid-body attitude stabilization. The Journal of the As- actuator faults. IEE Proceedings Control Theory &
tronautical Sciences 2004; 52(1-2): 61-74. Applications 2005; 152(2): 138-146.
[16] Hu Q L, Cao J, Zhang Y Z. Robust backstepping slid- [33] Yang G H, Wang H, Xie L H. Fault detection for output
ing mode attitude tracking and vibration damping of feedback control systems with actuator stuck faults: a
flexible spacecraft with actuator dynamics. ASCE steady-state-based approach. International Journal of
No.1 HU Qinglei et al. / Chinese Journal of Aeronautics 24(2011) 32-45 · 45 ·