You are on page 1of 14

Design of Asymptotic Second-order Sliding Mode

Control for Satellite Formation Flying


Junquan Li, Yaodong Pan, Krishna Dev Kumar

I. Introduction
S
atellite formation-ying (SFF) is an enabling technology for future space missions. Several approaches
have been proposed for control of SFF; these include linear control and nonlinear control methods which
may be adaptive or non-adaptive.
1
A PD controller with feedback linearisation and Lyapunov-based control
laws was suggested for optimum power consumption and tracking performance.
1
De Queiroz et al.
2
and
Wong et al.
3
proposed Lyapunov-based nonlinear adaptive controllers for SFF while a neural network-
based adaptive controller was suggested by Gurl et al.
4
for SFF in deep space missions. Lim and Bang
5
developed a controller using adaptive control technique for SFF under a thrust misalignment model and
bounded disturbances using an electric propulsion system. This adaptive controller was shown to be more
ecient than the backstepping controller with respect to fuel consumption during formation maneuvers in
the presence of thrust misalignment.
Adaptive sliding mode control (SMC) was developed by Godard and Kumar
6
as a fault tolerant controller
without fault detection and isolation (FDI) for nanosatellite formation ying, where a nonlinear SFF model
was used. Massey and Shtessel
7
proposed a rst order sliding mode controller with sliding mode disturbance
observer and compared it with the super twisting sliding mode controller and the integrated second order
sliding mode controller for SFF. The simulation results showed that this method is eective in reducing the
tracking errors of the follower satellite, and it results in lower fuel consumption than the other two sliding
mode controllers. In this note, a new second order sliding mode controller is proposed. Second order sliding
mode control (2nd SMC)
8,9,10
has been applied to MIMO nonlinear systems, and an asymptotic 2nd SMC
algorithm without using derivatives was proposed
11
such that a 2nd sliding mode (SM) is reached asymptot-
ically with a rst order SM controller (1st SMC). The asymptotic convergence to the 2nd SM also happens in
a 1st SMC system with fast dynamic actuators.
12
However, this algorithm can only be implemented in those
systems where the coecient of the derivative term in the dynamical equation describing the convergence
to the sliding mode is positive, which is a necessary condition for the 2nd SM to be reached asymptotically.
The stability of an asymptotic 2nd SM cannot be proved directly by SM control theory because the 2nd SM
can not be reached in nite time with a 1st SMC.
In this paper, an asymptotic 2nd SMC without using the derivative of the switching function is proposed
and applied to SFF. A Lyapunov function is used to prove the local asymptotic stability of the proposed
controller.
II. Robust 2nd SMC Architecture for Satellite Formation Flying
A. Problem Formulation
In this study, the satellites are modeled as point masses and therefore the rotational dynamics of the leader
and follower satellites are not taken into account. The orbital equations of motion for the leader satellite
and the full nonlinear translational dynamics of the follower satellite relative to the leader satellite taking

Presented as Paper 2010-8296 at the AIAA Guidance, Navigation, and Control Conference, Toronto, Ontario Canada, 2 - 5
August 2010. J. Li, Y. Pan and K. D. Kumar are with the Department of Aerospace Engineering, Ryerson University, 350 Victo-
ria Street, Toronto, Ontario M5B 2K3, Canada (e-mail: junquan.li@ryerson.ca, yaodong.pan@ryerson.ca, kdkumar@ryerson.ca).
1 of 14
American Institute of Aeronautics and Astronautics
into account the thrust and disturbance forces can be written in the following form:
6
r
L
r
L

2
+

r
2
L
= 0 (1)
r
L

+ 2

r
L
= 0 (2)
x 2

y
_

2
x +

y
_
+
_
r
L
+x
r
3
F

1
r
2
L
_
=
1
m
F
U
x
+
1
m
F
D
x
(3)
y + 2

x +
_

2
y
_
+

r
3
F
y =
1
m
F
U
y
+
1
m
F
D
y
(4)
z +

r
3
F
z =
1
m
F
U
z
+
1
m
F
D
z
(5)
Rewrite the SFF system (1)-(5) in state-space form as

X(t) =
_
X
2
(t)
f(X)
_
+B(U +D) (6)
where f(X) =
_

_
2

y +

2
x

y
rL+x
r
3
F

1
r
2
L
2

x +

2
y

rF
3
y


rF
3
z
_

_, B =
_
O
33
g
_
, g =
1
mF
I
3
, U = [U
x
U
y
U
z
]
T
R
3
,
X(t) = [X
1
(t) X
2
(t)]
T
R
6
, X
1
(t) = [x y z]
T
R
3
, X
2
(t) = [ x y z]
T
R
3
. Here D = (d + ) R
3
denotes the dierential J
2
disturbance
13
(d = [d
x
d
y
d
z
]
T
R
3
) and thruster fault ( = [
x

y

z
]
T

R
3
). In low earth orbit (LEO), J
2
disturbance is by far the most dominant perturbation. is the true
anomaly. m
F
is the mass of the follower satellite. r
L

= [ r
L
0 0 ]
T
denes the position vector of the
leader satellite with respect to the inertial coordinate frame. The relative position of the follower satellite
from the origin of the leader satellite coordinate frame is dened by R
3
, = [x y z]
T
. The follower
satellite position with respect to the inertial coordinate frame is given by r
F
=
_
(r
L
+x)
2
+y
2
+z
2
. The
leader satellite is assumed to be in an unperturbed reference orbit dened by Eqs. (1) and (2). A circular
formation and a projected circular formation
6
are the desired trajectories. The control objective is to
compute the thrust U required to drive the follower satellite to a desired formation trajectory with respect
to the leader satellite in the presence of J
2
disturbance force and thruster fault.
Next, in order to design the proposed 2nd order sliding mode control law, a linear switching function is
dened as
(t) = Ce(t) =
_
M I
3
_
e(t) = Me
1
(t) +e
2
(t) (7)
where (t) R
3
. M R
33
is a diagonal constant matrix with positive diagonal elements
i
(i = 1, 2, 3).
The elements of the matrix M are selected such that when the system is in a sliding mode, the eigenvalues
of the closed-loop system are stable. e(t) R
6
is the tracking error dened as
e(t) = X(t) X
d
(t) (8)
Here e(t) = [e
1
(t) e
2
(t)]
T
, e
1
(t) = e
2
(t), e
1
(t) = X
1
(t) X
d1
(t), and e
2
(t) = X
2
(t) X
d2
(t). X
d
(t) =
[X
d1
X
d2
]
T
R
6
is the desired trajectory.
Taking the derivative of the system given by Eq. (6) yields

X(t) = A(X)

X(t) +B(

U(t) +

D(t)) (9)
where A(X) R
33
is dened as
A(X) =
_
O
33
I
3
f(X)
X1
f(X)
X2
_
=
_
O
nn
I
n
a
1
(X) a
2
(X)
_
Assumption 1 The disturbance and its derivative are bounded by
|D(t)| < d
1
, |

D(t)| < d
2
where d
1
and d
2
are positive constants.
2 of 14
American Institute of Aeronautics and Astronautics
Assumption 2 The unknown symmetric matrix g is bounded as
0 < g
1
I
3
< g < g
2
I
3
where g
1
and g
2
are positive constants.
The desired trajectory X
d
(t) is determined by

X
d
(t) = A
d
X
d
(t) +B
d
U
d
(t) (10)
Using Eqs. (8)-(10), the error dynamics of the SFF system is obtained as
e(t) = A(X) e(t) +B(

U(t) +

D(t)) + (A(X) A
d
)

X
d
(t) B
d

U
d
(t) (11)
The desired trajectory is designed such that the term (A(X) A
d
)

X
d
(t) B
d

U
d
(t) satises the matching
condition, i.e.,
(A(X) A
d
)

X
d
(t) B
d

U
d
(t) = B(a

X
d
(t) b

U
d
(t)) (12)
where a and b are functions or constants. Substituting this matching condition into Eq. (11), the error
dynamics can be expressed as
e(t) = A(X) e(t) +B( v(t) +

D(t)) (13)
where v(t) denotes an alternative control input dened as
v(t) =

U(t) +a

X
d
(t) b

U
d
(t) (14)
If a and b are constants, the following holds.
v(t) = U(t) +aX
d
(t) bU
d
(t) (15)
Next, an extended state vector z(t) is dened as
z(t) =
_
e(t)
e
2
(t)
_
R
3
(16)
Substituting e(t) by z(t) into Eq. (11), the extended system is obtained as
z(t) =

A(X)z(t) +

B( v(t) +

D(t)) (17)
where

A(X) =
_

_
O
33
I
3
O
33
O
33
O
33
I
3
O
33
a
1
(X) a
2
(X)
_

_ (18)

B =
_

_
O
33
O
33
g
_

_ (19)
Referring to Eq. (7), the switching function can be expressed in terms of z(t) as follows.
(t) =
_
M I
3
O
33
_
z(t). (20)
Taking the rst and second derivatives of the switching function given by Eq. (7) and using Eq. (20)
yield
(t) = C e(t) = M e
1
(t) + e
2
(t) =
_
O
33
M I
3
_
z(t) (21)
(t) = CA(X) e(t) +CB( v(t) +

D(t)) = CA(X) e(t) +g( v(t) +

D(t)) (22)
3 of 14
American Institute of Aeronautics and Astronautics
Next, a new state vector is dened as
w(t) =
_

_
(t)
(t)
e
1
(t)
_

_ (23)
The transformation matrix T to transfer z(t) to w(t) with w(t) = Tz(t) is given by
T =
_

_
M I
3
O
33
O
33
M I
3
I
3
O
33
O
33
_

_ R
99
(24)
Substituting z(t) by w(t) into Eq. (17), the dynamical equation for the state vector w(t) is obtained as
w(t) =

A(X)w(t) +

B( v(t) +

D(t)) (25)
where

A(X) = T

A(X)T
1
=
_

_
O
33
I
3
O
33
(X) (X) (X)
I
3
O
33
M
_

_ (26)

B = T

B =
_

_
O
33
g
O
33
_

_, T
1
=
_

_
O
33
O
33
I
3
I
3
O
33
M
M I
3
M
2
_

_ (27)
(X) = a
1
(X) M
2
a
2
(X)M (28)
(X) = M +a
2
(X) (29)
(X) = a
1
(X)M + (M +a
2
(X))M
2
(30)
Using Eqs. (7) and (25) yields
= (X) (t) +(X)(t) +(X)e
1
(t) +g( v(t) +

D(t)) (31)
e
1
(t) = Me
1
(t) +(t) (32)
It is to be noted that Eq. (32) gives the reduced-order system on the sliding mode (t) = 0, which is designed
to be stable with positive diagonal elements of M. Eq. (31) will be used to show convergence to the sliding
mode.
B. Design of Chattering-free SMC System
The proposed second order sliding mode control (2nd SMC) is designed as
v(t) = k
1
(t) k
2
_
t
0
sgn((t))dt (33)
A chattering-free sliding mode control law (Chattering Free SMC) with the hyperbolic function is given
by
v(t) = k
1
tanh() (34)
A sliding mode control law (SMC) can be described as
v(t) = k
1
sign() (35)
Taking the derivative of the proposed control law, Eq. (33) and substituting v(t) into Eq. (31) yields
= (k
1
g (X)) (t) +(X)(t) +(X)e
1
(t) gk
2
sgn((t)) +g

D(t) (36)
4 of 14
American Institute of Aeronautics and Astronautics
where k
1
and k
2
are 3 3 diagonal matrices with positive constant diagonal elements.
Remark: The proposed control law given by (33) is a higher order sliding mode controller and is chat-
tering free. It does not require (t) which is usually required in any higher order sliding mode controllers.
In addition, the ordinary chattering free control laws use saturation or hyperbolic functions whereas the
proposed control law does not require these functions. Finally, the stability inside the boundary layer is
guaranteed by the proposed control law as shown by Lemma 1, Lemma 2, and Theorem 1, whereas the
chattering-free sliding mode control laws can not guarantee the stability inside the boundary layer.
Lemma 1 Consider the system (6) represented by two subsystems (31) and (32). With the switching func-
tion (t) dened by (7), the reduced-order system on the 2nd SM (t) = (t) = 0 is asymptotically stable.
Proof: On the 2nd SM, (t) = (t) = 0 holds. Thus, according to the system Eq. (32), the stability
of the reduced-order system on the 2nd SM (t) = (t) = 0 is determined by
e
1
(t) = Me
1
(t)
where M is a diagonal constant matrix used in Eq. (7) to dene the switching function (t) with positive
diagonal elements
i
(i = 1, 2, 3). Therefore the reduced-order system on the 2nd SM (t) = (t) = 0 is
asymptotically stable.
Lemma 2 Consider the system (6) represented by two subsystems (31) and (32). With the switching func-
tion (t) dened by Eq. (7) and the proposed control law (33), e
1
(t) converges to a vicinity of zero, determined
by ||M
1
|| |(t)|.
Proof: Dene a Lyapunov function candidate as
V
e
(t) =
1
2
e
T
1
(t)M
1
e
1
(t) (37)
Then its derivative is given as

V
e
(t) = e
T
1
(t)M
1
(Me
1
(t) +(t)) = e
T
1
(t)e
1
(t) +e
T
1
(t)M
1
(t)
||e
1
(t)||
2
+||e
1
(t)|| ||M
1
|| |(t)|
= ||e
1
(t)||(||e
1
(t)|| ||M
1
|| |(t)|)
which means that e
1
(t) converges until e
1
(t) enters a vicinity of zero determined by
||e
1
(t)|| ||M
1
|| |(t)|
Denote the diagonal matrix gk
2
by G and G
i
(i = 1, 2, 3) are the diagonal elements of G. Then the
dynamics to describe the convergence to the sliding mode is given by
= (X) (t) +(X)(t) +(X)e
1
(t) Gsgn((t)) +g

D(t) (38)
where (X) is dened as
(X) = k
1
g (X)
Choose a Lyapunov function candidate as
V

(t) =
1
2

T
(t)
_
I
3

1
I
3

1
I
3

2
I
3
_
(t) +
3

i=1
G
i
|
i
(t)|
=
1
2

T
(t) (t) +
1
2

T
(t)(t) +
1

T
(t) (t) +
3

i=1
G
i
|
i
(t)| (39)
which is continuously dierentiable when (t) = 0. Two positive constants
1
and
2
are chosen to satisfy

2
1
> 0. Therefore, the Lyapunov function candidate V

(t) is positive denite. (t) is dened as


(t) =
_
(t)
(t)
_
(40)
5 of 14
American Institute of Aeronautics and Astronautics
The derivative of V

(t) for (t) = 0 (refer to Appendix for the derivation of



V

(t)) along the trajectory


given by Eq. (38) is

(t) =
T
(t) (t) +
2

T
(t) (t) +
1

T
(t) (t) +
1

T
(t) (t) +
T
(t)Gsgn()
<
5
|(t)|
2
+|(t)|g
2
d
2
= |(t)|(
5
|(t)| g
2
d
2
) (41)
which means that V

(t) keeps decreasing until (t) converges to a vicinity determined by


|(t)| <
g
2
d
2

5
(42)
Theorem 1 With the proposed control law (33), the system enters a vicinity of the 2nd SM (t) = (t) = 0
and then to a vicinity of the origin, locally and asymptotically.
Proof: The convergence to the vicinity of the 2nd SM (t) = (t) = 0 has been proved by Lemma 1.
Similarly, with a Lyapunov function candidate as
V = V

(t) +V
e
(t)
=
1
2

T
(t) (t) +
1
2

T
(t)(t) +
1

T
(t) (t) +
3

i=1
G
i
|
i
(t)| +
1
2
e
T
1
(t)M
1
e
1
(t)
it can be proved that (t), (t), and e
1
(t) converge to the vicinity of the origin, locally and asymptotically,
which indicates the local and asymptotic stability of the proposed control law (33).
III. Application to Satellite Formation Flying
With a view to assess the eectiveness of the proposed controller, the detailed response is numerically
simulated using the SFF system (6) in conjunction with the proposed control law (33). The SFF system
parameters and the orbital parameters for the leader satellite used in the numerical simulations are shown
in Table 1.
Table 1: Satellite Orbital Parameters and System Parameters
Parameters Values
m
F
(kg) 1

e
(km
3
s
2
) 398600
r
L
(km) 6878
e 0, 0.7
, , i, M (rad) 0
A Chattering Free SMC (34), SMC (35) and a chattering free adaptive fuzzy terminal sliding controller
(AFTSMC) (43) are used for comparison with the proposed 2nd SMC. The AFTSMC is chosen as:
14
U = g(X |
g
)
1
_

X
d
E

f(X |
f
)

h( |
h
) K
1

_
(43)
where

f(X |
f
) =
T
f

f
(X), g(X |
g
) =
T
g

g
(X) and

h( |
h
) =
T
h

h
(), and E = M e +C
0
p
q
e
p
q
1
e. The
terminal sliding surfaces are: s
1
= M
1
e
x
+C
01
e
p
q
x
+ e
x
, s
2
= M
2
e
y
+C
02
e
p
q
y
+ e
y
and s
3
= M
3
e
z
+C
03
e
p
q
z
+ e
z
.
In the numerical simulations for formation keeping, the formation radius and the initial position osets
are considered to be 1 km. The initial velocity components for all states are calculated by taking the
time derivative of the desired projected circular formation trajectory and substituting t = 0. The phase
angle between the leader and follower is assumed to be zero. The maximum thrust is taken as 3 mN. The
design parameters of the 2nd SMC are considered as:
1
=
2
=
3
= 2 (switching function parameters),
k
1
= 1 10
3
(controller parameter) and k
2
= 1 10
9
(controller parameter).
6 of 14
American Institute of Aeronautics and Astronautics
First, the performance of the proposed controller is demonstrated for formation keeping when the leader
satellite is in a circular orbit. Fig. 1 shows the relative position errors, thrust demand and fuel consumption
using 2nd SMC, AFTSMC, Chattering Free SMC and SMC. All controllers exhibit the similar tracking
performance. With respect to the fuel consumption, the proposed controller is the most economical followed
by Chattering Free SMC, AFTSMC and SMC.
Next, a thruster periodic additive fault scenario is considered. Thruster periodic additive fault is dened
by
j
as

j
=
_
0 t < t
f
a
j
+b
j
sin(c
j
t) t t
f
j = x, y, z (44)
where t
f
is an unknown time. To test the thruster periodic additive faults, the faults are injected after 1
orbit with an initial oset error of 1 km. The scenario is considered as a
j
, b
j
= 1 10
5
and c
j
= 9 10
5
,
j = x, y, z, with the maximum thrust of 3 mN. In order to compare the controllers without bias, all controller
parameters are considered to be the same as selected for the fault free case of formation keeping. Fig. 2
shows the relative position errors and thrust demand using 2nd SMC, AFTSMC, Chattering Free SMC, and
SMC. The values of thruster faults are 20% of the minimum thrust value required to stabilize the formation.
It is found that that 2nd SMC stabilizes the formation with the errors being bounded within 2 m with
minimal formation geometry changes. The tracking errors using AFTSMC, Chattering Free SMC and SMC
are 4 m, 10 m and 1 m, respectively. Fig. 2 also shows the fuel consumption over 3 orbits. The 2nd SMC
eectively stabilizes the formation during faults using less fuel. In other words, the 2nd SMC is the most
suitable controller for handling thruster faults.
Next, the performance of the proposed controller is demonstrated for formation keeping when the leader
is in a high eccentricity orbit (e=0.7). AFTSMC, Chattering Free SMC, and SMC all result in the same
tracking performance with the similar trend in fuel consumption as found for the case of a leader satellite
in a circular orbit. Adaptive sliding mode control (ASMC) and nonsingular terminal sliding mode control
(NTSMC)
6
are also compared with the proposed 2nd SMC. The proposed 2nd SMC parameters are chosen
as k
1
= 610
2
and k
2
= 110
9
. With 3 mN thruster, ASMC or NTSMC can not stabilize the formation.
However, with 10 mN thruster as shown in Figs. 3 and 4, the formation is stabilized but the tracking
performance is not as good as exhibited by the proposed 2nd SMC and the formation consumes higher fuel.
There is also an obvious chattering phenomenon when using NTSMC. Thus, the proposed controller can
stabilize the formation with less thrust and fuel as well as no chattering in the control demand.
Finally, the eectiveness of the proposed controller is tested for the case of multiple formation maneuvers
when the leader satellite is in a circular orbit. The follower satellite is maneuvered from a 1 km to a
2 km radius projected circular formation after 1 orbit and then after 2 orbits to a 4 km radius circular
formation (Fig. 5). It is found that 2nd SMC, Chattering Free SMC, and SMC using 3 mN thrusters all
provide the similar maneuver performance. However, AFTSMC handles only the rst maneuver (the plots
of tracking performance and thruster command for AFTSMC are omitted). It is to be noted that AFTSMC
applies the fuzzy logic system that is used to estimate the system uncertainty and therefore it needs a
larger gain. Furthermore, AFTSMC can not maneuver the follower satellite from a 2 km radius projected
circular formation to a 4 km radius circular formation with 3 mN saturation. SMC works for two formation
maneuvers, but it uses a large amount of fuel. Fig. 5 shows that the tracking performance of the proposed
controller is better than the chattering free SMC controller, but it uses a little more fuel. Overall considering
both the tracking performance and fuel consumption, the proposed 2nd SMC is still the best controller for
reliable formation maneuvers using 3 mN thruster saturation.
IV. Conclusions
In this paper, a new nonlinear controller for spacecraft formation ying was proposed. This controller was
based on 2nd order sliding mode. The proposed controller can provide desired performance in the presence
of nonlinear system dynamics, J
2
perturbations, thruster faults and position osets, and is constructive.
It needs no complex o-line computation, and oers lower fuel consumption after faults than the other
controllers presented in this paper. Local asymptotic stability of the proposed controller was proven by
using Lyapunov function. Numerical simulation results conrmed that this controller provides low fuel
consumption and precise and eective tracking performance for formation keeping and formation maneuvers
in the presence of 1 km initial errors, 3 mN thruster saturation and thruster periodic faults.
7 of 14
American Institute of Aeronautics and Astronautics
Appendix
Derivation of

V

(t): Using Eq. (39), the derivative of V

(t) for (t) = 0 along the trajectory given by


Eq. (38) is derived as

(t) =
T
(t) (t) +
2

T
(t) (t) +
1

T
(t) (t) +
1

T
(t) (t) +
T
(t)Gsgn()
=
T
(t)((X) (t) +(X)(t) +(X)e
1
(t) Gsgn(t)) +g

D(t)) +
2

T
(t) (t) +

1

T
(t) (t) +
1

T
(t)((X) (t) +(X)(t) +(X)e
1
(t) Gsgn(t)) +g

D(t))
+
T
(t)Gsgn()
=
1
2

T
(t)Q
1
(X)(t) + (
T
(t) +
1
(t))((X)e
1
(t) +g

D(t))
1

T
(t)Gsgn(t) (45)
where Q
1
(X) is dened as
Q
1
(X) =
_
2((X)
1
I
3
) (X)
2
I
3
+
1

T
(X)

T
(X)
2
I
3
+
1
(X) 2(
3
I
3

1
(X))
_
It is assumed that the parameters of the controller, k
1
and k
2
and the positive constants
i
(i = 1, 2, 3, 4, 5)
are chosen appropriately such that the following conditions hold for all X R
6
and for all (t) in a vicinity
|(t)| < v
s
(t 0, v
s
> 0):

2
>
2
1
(46)
(X) >
1
I
3
(47)

3
I
3
>
1
(X) (48)
Q
1
(X) > 2
4
I
6
(49)
G > (X)e
1
(t) +g

D(t) (50)
Q
2
(X) > 2
5
I
2
(51)
where Q
2
(X) is dened as
Q
2
(X) =
_
2
4
||(X)|| ||M
1
||
||(X)|| ||M
1
|| 2
4
_
Using Eq. (45) and Eqs. (46)-(51) yields

(t) <
4
|(t)|
2
+
T
(t)((X)e
1
(t) +g

D(t)) +
1

T
(t)((X)e
1
(t) +g

D(t) Gsgn(t))
<
4
|(t)|
2
+| (t)|(||(X)|| |e
1
(t)| +g
2
d
2
)
<
4
|(t)|
2
+| (t)|(||(X)|| ||M
1
|| |(t)| +g
2
d
2
)
=
1
2
_
|(t)| | (t)|
_
Q
2
(X)
_
|(t)|
| (t)|
_
+| (t)|g
2
d
2
<
5
|(t)|
2
+|(t)|g
2
d
2
= |(t)|(
5
|(t)| g
2
d
2
) (52)
References
1
Schaub, H., Vadali, S. R., Junkins, J. L., and Alfriend, K. T., Spacecraft Formation Flying: A Review and New Results
on State Feedback Control, Acta Astronautica, Vol. 65, No. 11-12, 2009, pp. 15371552.
2
de Queiroz, M. S., Kapila, V., and Yan, Q., Adaptive Nonlinear Control of Multiple Spacecraft Formation Flying,
Journal of Guidance, Control, and Dynamics, Vol. 23, No. 3, 2000, pp. 385 390.
3
Wong, H., Kapila, V., and Sparks, A. G., Adaptive output feedback tracking control of spacecraft formation, Interna-
tional Journal of Robust and Nonlinear Control , Vol. 12, No. 2-3, 2002, pp. 117 139.
4
Gurl, P., I. M. and Kasdin, N. J., Adaptive Neural Control of Deep-Space Formation Flying, Journal of Guidance,
Control, and Dynamics, Vol. 26, No. 3, 2003, pp. 491 501, doi: 10.2514/2.5072.
5
Lim, H. C. and Bang, H., Adaptive Control for Satellite Formation Flying Under Thrust Misalignment, Acta Astro-
nautica, Vol. 65, No. 1-2, 2009, pp. 112 122.
8 of 14
American Institute of Aeronautics and Astronautics
6
Godard and Kumar, K. D., Fault Tolerant Recongurable Satellite Formations Using Adaptive Variable Structure
Techniques, Journal of Guidance, Control, and Dynamics, Vol. 33, No. 3, 2010, pp. 969 984, doi: 10.2514/1.38580.
7
Massey, T. and Shtessel, Y., Continuous Traditional and High-Order Sliding Modes for Satellite Formation Control,
Journal of Guidance Control and Dynamics, Vol. 28, 2005, pp. 826831.
8
Elimali, H. and Olgac, N., Robust Output Tracking Control of Nonlinear MIMO Systems via Sliding Mode Technique,
Automatica, Vol. 28, No. 1, 1992, pp. 145151, doi: 10.1016/00051098(92)900147.
9
Defoort, M., Floquet, T., Kokosy, A., and Perruquetti, W., A Novel Higher Order Sliding Mode Control Scheme,
Systems and Control Letters, Vol. 58, 2009, pp. 102108, doi:10.1016/j.sysconle.2008.09.004.
10
Wang, Y. Q., Zhou, D. H., and Gao, F., Robust Fault Tolerant Control of a Class of Non-minimum Phase Nonlinear
Processes, Journal of Process Control , 2007, pp. 523537, doi:10.1016/j.jprocont.2006.12.002.
11
Pan, Y., Liu, G., and Kumar, K. D., Reduced-order Design of High-order Sliding Mode Control System, International
Journal of Robust and Nonlinear Control , 2010, doi: 10.1002/rnc.1678.
12
Fridman, L. M., Stability of Motions in Singularly Perturbed Discontinuous Control Systems, Proceedings of IFAC
World Conference, Sydney, 1993, pp. 367370.
13
Schaub, H., Vadali, S. R., Junkins, J. L., and Alfriend, K. T., Spacecraft Formation Flying Control using Mean Orbit
Elements, Journal of the Astronautical Sciences, Vol. 48, No. 1, 2000, pp. 6987.
14
Li, J. and Kumar, K. D., Fault Tolerant Attitude Synchronization Control during Formation Flying, Journal of
Aerospace Engineering, Vol. 24, 2011, pp. 251263, 10.1061/(ASCE)AS.19435525.0000080.
9 of 14
American Institute of Aeronautics and Astronautics
0 0.2 0.4 0.6 0.8 1
1000
0
1000
2000
x


x
d

[
m
]
0 0.2 0.4 0.6 0.8 1
500
0
500
1000
y


y
d

[
m
]
0 0.2 0.4 0.6 0.8 1
1000
0
1000
2000
Orbit
z


z
d

[
m
]
0 0.2 0.4 0.6 0.8 1
9
5
0
5
9
x 10
3
U
x

[
m
/
s
2
]
0 0.2 0.4 0.6 0.8 1
9
5
0
5
9
x 10
3
U
y

[
m
/
s
2
]
0 0.2 0.4 0.6 0.8 1
9
5
0
5
9
x 10
3
Orbit
U
z

[
m
/
s
2
]
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
50
100
150
200
250
300
Orbit
F
u
e
l

C
o
s
t

[
N
m
/
s
]


0 0.2 0.4 0.6 0.8 1
0
5
10
15
20


2nd SMC
AFTSMC
SMC
Chattering Free SMC
Figure 1: Relative position error, thrust demand and fuel consumption for formation keeping e=0 (2nd
SMC-dash-dot lines; AFTSMC-dash lines; Chattering Free SMC-dot lines; SMC-solid lines)
10 of 14
American Institute of Aeronautics and Astronautics
0 0.5 1 1.5 2 2.5 3
500
0
500
1000
1500
x


x
d

[
m
]
0 0.5 1 1.5 2 2.5 3
500
0
500
1000
y


y
d

[
m
]
0 0.5 1 1.5 2 2.5 3
500
0
500
1000
1500
Orbits
z


z
d

[
m
]
0 0.5 1 1.5 2 2.5 3
9
5
0
5
9
x 10
3
U
x

[
m
/
s
2
]
0 0.5 1 1.5 2 2.5 3
9
5
0
5
9
x 10
3
U
y

[
m
/
s
2
]
0 0.5 1 1.5 2 2.5 3
9
5
0
5
9
x 10
3
Orbits
U
z

[
m
/
s
2
]
2.6 2.8 3
20
10
0
10
20
2.6 2.8 3
20
10
0
10
20
2.6 2.8 3
20
10
0
10
20
0 0.5 1 1.5 2 2.5 3
0
500
1000
1500
2000
2500
3000
Orbits
F
u
e
l

C
o
s
t

[
N
m
/
s
]


0 1 2 3
0
10
20
30


2nd SMC
AFTSMC
SMC
Chattering Free SMC
Figure 2: Relative position error, thrust demand and fuel consumption for formation keeping, thruster
fault after 1 orbit e=0 (2nd SMC-dash-dot lines; AFTSMC-dashed lines; Chattering Free SMC-dot lines;
SMC-solid lines)
11 of 14
American Institute of Aeronautics and Astronautics
0 0.5 1 1.5 2
1000
500
0
500
1000
x


x
d

[
m
]
0 0.5 1 1.5 2
1000
500
0
500
1000
y


y
d

[
m
]
0 0.5 1 1.5 2
1000
500
0
500
1000
Orbits
z


z
d

[
m
]


0 0.5 1 1.5 2
0.01
0.005
0
0.005
0.01
U
x

[
m
/
s
2
]


0 0.5 1 1.5 2
0.01
0.005
0
0.005
0.01
U
y

[
m
/
s
2
]
0 0.5 1 1.5 2
0.01
0.005
0
0.005
0.01
Orbits
U
z

[
m
/
s
2
]


ASMC
NTSMC
Figure 3: Relative position error, thrust demand for formation keeping e=0.7 (ASMC-solid lines; NTSMC-
dash lines)
12 of 14
American Institute of Aeronautics and Astronautics
0 0.5 1 1.5 2
2000
1000
0
1000
x


x
d

[
m
]


0 0.5 1 1.5 2
500
0
500
1000
y


y
d

[
m
]
0 0.5 1 1.5 2
1000
500
0
500
1000
Orbits
z


z
d

[
m
]
0 0.5 1 1.5 2
0.01
0.005
0
0.005
0.01
U
x

[
m
/
s
2
]
0 0.5 1 1.5 2
0.01
0.005
0
0.005
0.01
U
y

[
m
/
s
2
]
0 0.5 1 1.5 2
0.01
0.005
0
0.005
0.01
Orbits
U
z

[
m
/
s
2
]
2nd SMC
Figure 4: Relative position error, thrust demand for formation keeping e=0.7 (2nd SMC-solid lines)
13 of 14
American Institute of Aeronautics and Astronautics
0 1 2 3 4
500
0
500
1000
1500
x


x
d

[
m
]
0 1 2 3 4
3000
2000
1000
0
1000
y


y
d

[
m
]
0 1 2 3 4
1000
500
0
500
1000
Orbits
z


z
d

[
m
]
0 1 2 3 4
9
5
0
5
9
x 10
3
U
x

[
m
/
s
2
]
0 1 2 3 4
9
5
0
5
9
x 10
3
U
y

[
m
/
s
2
]
0 1 2 3 4
9
5
0
5
9
x 10
3
Orbits
U
z

[
m
/
s
2
]
3 3.5 4
50
0
50
3 3.5 4
50
0
50
3 3.5 4
50
0
50
0 0.5 1 1.5 2 2.5 3 3.5 4
0
1000
2000
3000
4000
5000
6000
Orbits
F
u
e
l

C
o
s
t

[
N
m
/
s
]


0 0.5 1 1.5 2 2.5
0
10
20
30
40
50
60
70


2nd SMC
AFTSMC
SMC
Chattering Free SMC
Figure 5: Relative position error, thrust demand and fuel consumption for formation maneuvering (2nd
SMC-dash-dot lines; AFTSMC-dashed lines; Chattering SMC free-dot lines; SMC-solid lines)
14 of 14
American Institute of Aeronautics and Astronautics

You might also like