Vous êtes sur la page 1sur 6

2010 11th International Workshop on Variable Structure Systems

Mexico City, Mexico, June 26 - 28, 2010

Output Feedback Second Order Sliding Mode Control Design for a


3-DOF Helicopter Based on Its Simplified Model
Jovan Merida-Rubio and Luis T. Aguilar
Instituto Politecnico Nacional
Centro de Investigacion y Desarrollo de Tecnologa Digital
Avenida del Parque 1310 Mesa de Otay, Tijuana 22510 Mexico
e-mail: merida@citedi.mx; luis.aguilar@ieee.org

AbstractThis paper presents the solution to the tracking while the stability analysis and simulations will be carried out
control problem for an underactuated scaled autonomous heli- in its full model where most of its parameters are assumed
copter using variable structure control via output measurements. unknown (e.g., drag force, main rotor thrust, among others).
First, it is designed a state-feedback second order sliding mode
controller to stabilize height and rotation positions. Controller is The motivation in using SOSM is to guarantee robustness
designed from a reduced model of the helicopter and the resulting against discrepancies in the model. The stability analysis of
control law is tested in its full model thus showing the robustness the closed-loop system was carried out in the frame of non-
against unmodelled dynamics. Simulation results illustrate the smooth Lyapunov function [7].
performance the effectiveness of the controller.
The paper is organized as follows. In Section II is given
I. I NTRODUCTION the dynamic model and problem formulation. The state feed-
back design is presented in Section III complemented with
Among unmanned Aerial Vehicle configurations available Lyapunov stability prove. The velocity observer and output
today, helicopters are one the most maneuverable and versatile feedback design is provided in Section IV. Performance of the
platforms. Helicopters can perform forward flight, side flight, proposed results are given in Section IV through simulations.
climb, hover and any combination of these maneuvers. These Section V presents some conclusions.
capabilities have brought about the use of autonomous minia-
The following definition will be used throughout the paper.
ture helicopters. For these reasons, there is currently great
The norm kxk2 , with x Rn , denotes the Euclidean norm and
interest in using these platforms in a wide range of applications
kxk1 = |x1 | + |x2 | + + |xn | stands for the sum norm. The
that include exploration of places inaccessible, surveillance,
minimum and maximum eigenvalue of a matrix A Rnn is
security purposes and other. For performing safely many types
of these tasks, high maneuverability and robustness of the
denoted by min P{A} and max {A}, respectively; and kAk =
kAk1 = maxj ni=1 |aij | stands for the 1-norm. The vector
controllers with respect to disturbances and modeling errors
sign(x) is given by sign(x) = [ sign(x1 ), . . . , sign(xn )]T
are required. This has generated considerable interest in the
where the signum function is defined as
robust flight control design.
Many efforts for the analysis and control of helicopter
prototypes have gone in many directions. Avila-Vilchis et al. 1
if xi > 0
[1] developed a nonlinear model and design a nonlinear control sign(xi ) = [1, 1] if xi = 0 xi R.


strategy for a VARIO scale model helicopter. Garca-Sanz et 1 if xi < 0
al. [2] addressed the pitch control using linear controllers for
a 3-DOF helicopter prototype from Quanser. Starkov et al. [3] II. DYNAMIC M ODEL AND P ROBLEM S TATEMENT
designed an output feedback sliding mode controller using a
high-gain observer while Orlov et al. [4] developed a sliding The full mathematical model of the 3-DOF VARIO scale
mode observer for the same Quanser prototype. Isidori et al. model helicopter, depicted in Fig. 1, can be described as
[5] solved the problem of controlling the vertical motion of a follows [1]:
helicopter while stabilizing the lateral and horizontal position
by using a continuous nonlinear controller. An adaptive output c0 z = c8 2 u1 + c9 + c10 c7 + wz (1)
feedback control tested on a laboratory model helicopter was m() = c4 [c12 + c13 ] u1 + c5 c11 2 u2
presented by Kutay et al. [6].
In this paper we will design a second order sliding mode c6 [2c5 + c4 ] sin(2c3 )
(SOSM) controller to solve the trajectory tracking control  
c4 c14 2 + c15 + w (2)
problem for a helicopter assuming that a simplified model of  
2 2
the system is available. The main contribution of the paper m() = c1 + c2 cos (c3 ) [c12 + c13 ]u1 c4 c11 u2
is that controller will be derived from the reduced model !  
+ c6 c1 + c2 cos2 (c3 ) + 2c4 sin(2c3 )
Work partially supported by Instituto Politecnico Nacional under Grant SIP  
2010-0132. + c1 + c2 cos2 (c3 ) [c14 2 + c15 ] (3)

978-1-4244-5831-8/10/$26.00 2010 IEEE 364


2010 11th International Workshop on Variable Structure Systems
Mexico City, Mexico, June 26 - 28, 2010

h3 (q) = c4 c6 sin(2c3 )( d )
+ [c6 sin(2c3 )( d )2 + c14 2 + c15 ]
 
c1 + c2 cos2 (c3 ) (11)

the error equation takes the form:



e1 = e2
e2 = f (q, t) + w + g(q) (12)
= k4 (q)1 + k5 (q)2 + m(q)1 h3 (q). (13)

where w1 = c1
0 wz , w2 = m(q)
1
w , and
   
h1 (q) zd k1 (q) 0
f (q, t) = , g(q) = . (14)
z h2 (q) d k2 (q) k3 (q)

It is assumed that the disturbances are uniformly bounded, i.e.,


Fig. 1. Helicopter mounted on a platform.
sup |wi (t)| Wi , i = 1, 2. (15)
t

Since the helicopter is installed in a laboratory environment,


where m() = c1 c5 c24 + c2 c5 cos2 (c3 ). In the above some aerodynamical effects and inertias do not take effect
equations, z R is the height, R is the yaw angle, R in the system. Then, the dynamic equation (12)(14) can be
is the main rotor azimuth angle and ci (i = 0, . . . , 15) are the reduced to the following form [1]
physical constants, given in Table I, representing the inertial,
e1 = e2
gravitational and aerodynamical effects. The yaw angle is (16)
the rotation of the body around the vertical axis z, u1 , and e2 = f(q, t) + w + g(q)
u2 are voltage applied to the engines main and tail rotor,
respectively; c0 is the mass of the helicopter; c1 to c6 are the where
inertia parameters, c7 is the weight of the body, c8 and c9 are  
the main rotor thrust, c10 is the vertical drag force produced by g zd
f(q, t) = ,
the main rotor; c11 , c12 , c14 , and c15 are constant components k3 (q)kg ( + d ) + h2 (q) d
that comprise drag torque of main rotor; and c13 is constant   (17)
k1 (q) 0
Kmot the engine. g(q) = .
k2 (q) k3 (q)
For simplicity, we rewrite the open-loop system (1)(3) in
terms of the error z = z zd , = d where zd and d Here, the term kg comes from an angular velocity feedback
are continuously differentiable functions denoting the desired of the gyro control system for the tail of helicopter and
trajectory of motion of height and yaw angle, respectively.
),
Then, setting e1 = (z, ), e2 = (z, q = (e1 , e2 , 1 , 2 ), k2 (q) = c4 m1 (c12 + c13 ) (18)
and
k3 (q) = c5 c11 2 m1 (19)
2 ! 
k1 (q) = c1
0 c8 , (4) h2 (q) = c4 m1 c14 2 + c15 , (20)
1
k2 (q) = m(q) c4 (c12 + c13 ), (5)
where m = (c1 c5 c24 ) and g = 9.8 is the gravitational
1 2
k3 (q) = m(q) c5 c11 , (6) constant. It is assumed throughout the paper that the speed
k4 (q) = m(q) 1 2
[c1 + c2 cos (c3 )](c12 + c13 ) (7) of the blades can not escape to infinity, i.e.,

k5 (q) = m(q)1 c4 c11 2 (8) kk . (21)


h1 (q) = c1
0 (c9 + c10 c7 ), (9)
Then, the parameters that no require to be estimated are the
h2 (q) = m(q)1 c6 [2c5 + c4 ( + d )][ + d ] mass of the helicopter c0 ; the inertia parameters c1 , c2 , c3 , and
sin(2c3 ) m(q)1 c4 [c14 2 + c15 ] (10) c6 ; the weight of the body c7 ; the main rotor thrust c8 and c9 ;

365
2010 11th International Workshop on Variable Structure Systems
Mexico City, Mexico, June 26 - 28, 2010
TABLE I
3-DOF HELICOPTER MODEL PARAMETERS . By substituting (22) into (12), the closed-loop system takes
ci Numerical value units the form
c0 7.5 kg
c1 0.4305 kg m2 e1 = e2
c2 3 104 kg m2
c3 4.413 e2 = f (q, t) g(q)g(q)1 f(q, t) + w g(q)g(q)1 u. (25)
c4 0.108 kg m2 | {z }
c5 0.4993 kg m2 h(q,t)
c6 6.214 104 kg m2
c7 73.58 N Now, we need to prove boundedness of h(q, t). Regarding the
c8 3.411 kg
c9 0.6004 kg m/s term g(q)g(q)1 and from (4)(6), (18), (19) we obtain
c10 3.769 N
kg m
 
c11 0.1525 1 0
c12 12.01 kg m/s g(q)g(q)1
g(q)g(q)1
= m
= 1. (26)
1 105 N
0
c13 m(q)
c14 1.206 104 kg m2
2.642 N
Regarding the term f(q, t) and from (19), (20), we get
c15

f(q, t) kf(q, t)k


and the constant components c11 and c12 .
g + kg kk3 (q)kk + d k + kh2 (q)k + kzd k + kd k
Our objective is to find a control law (1 , 2 ) such that
asymptotically stabilize the origin zi , i , i = 0, i = 1, 2 = g + kg kk3 (q)kkd k + kh2 (q)k + kzd k + kd k
while also attenuating the effect of the external disturbances.

+ kg kk3 (q)kkk
It should be pointed out that the controller, derived from (16),
(17) will be analyzed and tested in (12)(14).
ka + kb kk (27)

III. S ECOND O RDER S LIDING M ODE C ONTROL D ESIGN where

In this section we will design a stabilization controller for ka = g + kg c5 c11 2 kd k + c4 m1 (c14 2 + c15 )
the vertical and rotational motion. It should be pointed out + kzd k + kd k (28)
that controller are derived from (16)(20) while the stability
proof, made by means of the Lyapunov function framework, kb = kg c5 c11 2 m1
will be done by considering the full model (12)(14).
are positive constants. Finally, regarding the term f (q, t) first
A. Stabilization of the Vertical and Rotational Motion note that
1
km(q)1 k = 0 (29)
In order to globally asymptotically stabilize (12)(14) the c1 c5 c24
following control law
Taking this into account and from (9), (10), (14), we obtain
= g(q) 1
[f(q, t) + u] (22)
f (q, t) kf (q, t)k
is chosen provided that only model and parameters from (16), kh1 (q)k + kh2 (q)k + kzd k + kd k (30)
(17) are available where u is the second order sliding mode + ke kk
2
kc + kd kk
controller given by
where
u = e1 + e2 + a sign(e1 ) + b sign(e2 ) (23)
kc = c1
0 (c9 + c10 c7 ) + 2c5 c6 0 kd k
where
+ c4 c6 0 kd k2 + kzd k + kd k
   
av 0 bv 0 + c4 0 (c14 2 + c15 ) (31)
a= , b= ,
0 ar 0 br
    (24) kd = 20 c6 (c5 + c4 kd k)
0 0
= v , = v . ke = c4 c6 0 .
0 r 0 r

are positive definite matrices. It is assumed that the main rotor Then
is rotating for all time (i.e., 6= 0) thus avoiding that k1 (q)1
be singular.
kh(q, t)k ka +kc +W +(kb +kd )kk+k 2
e kk = kh . (32)

366
2010 11th International Workshop on Variable Structure Systems
Mexico City, Mexico, June 26 - 28, 2010

We can now prove the following. The solutions of (37) when 2 = 0 are: 1 = 355.03 [rad/s],
Theorem 1: Consider the closed-loop system (25) with the 2= 4 380.99i [rad/s], 3 = 347.02 [rad/s]. Only the
parameters chosen as last of these values have a physical meaning for the system
(see Fig. 1 for the rotation sense of main rotor). To analyze
, 0, min {a} kh > max {b} > kh > 0. (33) the stability of the zero dynamics, define

Then the origin of the closed-loop system (25) is finite-time 2 = 2 3 (38)


stable.
Proof: Consider the Lyapunov function candidate whose time derivative is

1 T 1 2 = a1 22 + a2 + a3 21 + a4 22
V (q) = e [ + ]e1 + eT1 e2 + eT2 e2 + a|e1 |, > 0. (39)
2 1 2 = a1 22 + a2 |a3 |21 |a4 |22 .
(34)
Note that V (q) will be positive definite and radially unbounded
if + > 2 . The time derivative of (34) along the solution Rewriting the above expression in terms of velocity error 2 =
of (25) results in 2 + 3 we have

V (q) = eT1 ( + )e1 + eT1 e2 + eT1 e2 + eT2 e2 2 = a1 (2 + 3 )2 + a2 |a3 |(2 + 3 )1 |a4 |(2 + 3 )2 .
(40)
+ a sign(e1 )T e1 The equilibrium points of the above equation are 1 = 702.05
= eT1 ( + )e2 + eT2 e2 + a sign(e1 )T e2 [rad/s], 2 = 343.02 468.2i [rad/s], 3 = 0 [rad/s]. The
dynamics of (40) linearized around the equilibrium point 3 :
+ [eT1 + eT2 ][h(q, t)


g(q)g(q) 1
(e1 + e2 + a sign(e1 ) + b sign(e2 ))] 2 = f (2 ) f (2 ) 2 ,
2 2 =0
min {}ke1 k22 (min {} )ke2 k22 ! 
= 2a1 3 + |a3 |(3 )2 + 2|a4 |(3 )3 2
(min {a} max {b} kh(q, t)k)ke1 k1
= 0.37012.
(min {b} kh(q, t)k)ke2 k1 .
Then, the origin, i.e., the equilibrium point 2 = 0 is locally
Notice that V (q) is negative definite for any vector (q, q)T 6= 0 asymptotically stable.
if inequalities (33) and min {} > holds. Thus, concluding
that the origin of (25) is globally asymptotically stable. More- IV. O UTPUT F EEDBACK D ESIGN
over, one can concludes that q = 0 is reached in finite-time
(cf. [8, Th. 4.4]). The proof is completed. This section addresses the issue of control for helicopter
given in Section II, where now it is assumed that only vertical
B. Zero Dynamics Analysis and rotational positions are only available for measurement. It
should be pointed out that scaled helicopters are provided with
We focus our effort on proving boundedness of 2 . Under an internal loop which allows us to measure the main rotor
(22) the system (12)(13) has the following zero dynamics [9] velocity 2 . By using a second order sliding mode observer
[10], we will estimate the velocity of the system.
2 = k4 (q)1eq + k5 (q)2eq + m1 (q)h3 (q) (35) To estimate the state vector q R6 , the following observer
is proposed [10]
where
ueq = g(q)1 f(q, t) (36) zb1 = zb2 + z |z1 zb1 |
1/2
sign(z1 zb1 )
is the equivalent control. Substituting the above equation into zb2 = k1 (b
q )u1 + g + z sign(z1 zb1 ) (41)
(35) and if the desired trajectories and initial conditions are 1/2
chosen in such a way that the terms including zd , d , 22 , and b1 =
b2 + |1
b1 | sign(1
b1 )
2 can be neglected we have the following simplification: b2 = b
q )u1 + b
k2 (b k3 (b b2 ) + b
q )(u2 kg h2 (b
q)
a3 a4 + sign(1
b1 ) (42)
2 = a1 22 + a2 + + = a1 24 + a2 22 + a3 2 + a4 (37)
2 22
where qb R6 are the estimated states and z , z , , and
where a1 = c1 5 c14 = 2.415 10
4
, a2 = c1
5 c15 = are the observer gains. The initial conditions qbi (0) = qi (0)
5.2914, a3 = (c5 c8 ) (c0 c12 g) = 518.8359, and a4 =
1
and qbi+1 (0) = 0 where i = 1, 3, ensure the convergence of the
(c5 c8 )1 (c0 c13 g) = 432 104 (see Table I). observer. Choosing the parameters of the observer (41)(42)

367
2010 11th International Workshop on Variable Structure Systems
Mexico City, Mexico, June 26 - 28, 2010

according to [10] then its states converge in finite time to the 0


states of system (16), (17).

z [m]
0.2
The output feedback controller u1 and u2 :
0.4

u1 = b
0.6
k1 (b
q )(av sign(z1 ) bv sign(b
z2 zd )
0.8
0 50 100 150 200 250
v z1 v (b
z2 zd ) g) (43)
2
u2 = b q )1 [ar sign(1 ) br sign(
k3 (b b2 d ) r 1

[rad]
0
kg b
k3 (b
q )( b2 d ) b
b2 + d ) r ( q )u1 b
k2 (b h2 (b
q )]
(44) 2

4
are obtained if the corresponding observer variables qb are 0 50 100 150 200 250

substituted into the state feedback law (22). 3


x 10
1

u1 [m]
V. S IMULATION R ESULTS 0

In this section we demonstrate through numerical simula- 1

2
tions the effectiveness of the control (43) and (44) applied
to the system (1)(3). Defining toff = 50 [s], the desired 3

4
trajectory is 0 10 20 30 40 50 60

0.1
0 t toff

u2 [m]

0.2

0.3[e(ttoff )2 /350 ] 0.2
0.05
toff < t 130
zd = 0
0.1 cos [0.1(t 130)] 0.6 130 < t < 20 + 130


0.05

0.5 t 20 + 130, 0.1


0 10 20 30 40 50 60


0 0 t toff

1 e(ttoff )2 /350 toff < t 120
d = 2 0.01
z [m]



e(t120) /350 120 t < 180 0.005
2
1 + e(t180) /350 t 180 0
0.005
0.01
subject to z1 (0) = zb1 (0) = 0.2, z2 (0) = zb2 (0) = 0, 1 (0) = 0.015
0 50 100 150 200 250
b1 (0) = , 2 (0) =
b2 (0) = 0, 1 (0) = , and 2 (0) =
99.5. 1
[rad]

In the simulations, the controller gains in (43) and (43), 0

1
were set to av = 67, bv = 10, v = 8, v = 20, ar = 24,
2
br = 10, r = 8, and r = 10. The parameters of the observer
3
are z = 4, = 4, z = 15, and = 6.6.
4
Figure 2 shows motion around the desired trajectory for the 0 50 100 150 200 250

altitude and yaw angle according to the full system equations,


the control inputs, the errors, and input control. This figure 70
Tm [N]

demonstrate the performance and robustness of the second


80
order sliding mode controllers against unmodelled dynamics.
90
VI. C ONCLUSIONS
100
In this paper we considered the output feedback control of 0 50 100 150 200 250

a scaled helicopter model mounted on a platform, address-


2 [rad/s]

90
ing the trajectory tracking control problem by using second
100
order sliding mode control. We design a control law which
110
solves the problem of elevation and rotation. This control
law was obtained from a reduced model and implemented on 120

the full helicopter model thus demonstrating the robustness 130


0 50 100 150 200 250
of sliding mode against unmodelled dynamics and unknown Time

disturbances. Finite time stability was concluded for vertical


and angular motion of the helicopter. The stability analysis was Fig. 2. Time responses.

368
2010 11th International Workshop on Variable Structure Systems
Mexico City, Mexico, June 26 - 28, 2010

performed by using the Lyapunov stability theory. Effective-


ness and robustness of controllers were demonstrated through
simulations.
R EFERENCES
[1] J. Avila-Vilchis, B. Brogliato, A. Dzul, and R. Lozano, Nonlinear
modelling and control of helicopters, Automatica, vol. 39, pp. 1583
1596, 2003.
[2] M. Garca-Sanz, J. Elso, and I. Egana, Control de angulo de cabeceo
de un helicoptero como benchmark de diseno de controladores, Revista
Iberoamericana de Automatica e Informatica Industrial, vol. 3, no. 2,
pp. 111116, 2006.
[3] K. Starkov, L. Aguilar, and Y. Orlov, Sliding mode control synthesis
of a 3-DOF helicopter prototype using position feedback, in 10th
International Workshop on Variable Structure System, Antalya, Turkey,
June 2008, pp. 233237.
[4] Y. Orlov, M. Meza, and L. Aguilar, Sliding mode velocity-observer-
based stabilization of a 3-DOF helicopter prototype, in IFAC Sympo-
sium Robust Control Design, Haifa, Israel, 2009, pp. 179184.
[5] A. Isidori, L. Marconi, and A. Serrabi, Robust nonlinear motion control
of a helicopter, IEEE Trans. on Automatic Control, vol. 48, no. 3, pp.
413426, March 2003.
[6] A. Kutay, A. J. Calise, M. Idan, and N. Hovakimyan, Experimental
results on adaptive output feedback control using a laboratory model
helicopter, IEEE Trans. on Control Systems Technology, vol. 13, no. 2,
pp. 196202, March 2005.
[7] A. Baccioti and L. Rosier, Lyapunov functions and stability in control
theory. Berlin: Springer, 2005.
[8] Y. Orlov, Discontinuous systems - Lyapunov analysis and robust synthe-
sis under uncertainty condition. London: Springer-Verlag, 2009.
[9] C. Byrnes and A. Isidori, Limit sets, zero dynamics, and internal
models in the problem of nonlinear output regulation, IEEE Trans. on
Automatic Control, vol. 48, no. 10, pp. 17121723, Oct. 2003.
[10] A. Davila, L. Fridman, and A. Levant, Second-order sliding-mode
observer for mechanical systems, IEEE Trans. on Automatic Control,
vol. 50, no. 11, pp. 17851789, Nov. 2005.

369

Vous aimerez peut-être aussi