Está en la página 1de 9

www.ietdl.

org

Published in IET Control Theory and Applications


Received on 4th September 2007
Revised on 23rd April 2008
doi: 10.1049/iet-cta:20070335

ISSN 1751-8644

Decoupling control of a twin rotor


MIMO system using robust deadbeat
control technique
P. Wen T.-W. Lu
Faculty of Engineering and Surveying, University of Southern Queensland, Toowoomba 4350, QLD, Australia
E-mail: pengwen@usq.edu.au

Abstract: The decoupling control of a twin rotor multi-input multi-output (MIMO) system is studied and
proposed to apply robust deadbeat control technique to this nonlinear system. First, the nonlinear problem is
identied and system model is developed. Then, it is shown that the system is able to be decoupled into two
single-input single-output (SISO) systems, and the cross couplings can be considered as disturbances to each
other. Finally, a robust deadbeat control scheme is applied to the two SISO systems and a controller is
designed for each of them. This design is evaluated in simulations, and the nal result is tested in a twin
rotor MIMO system. Comparing with a traditional system with two proportional, integral and derivative (PID)
controllers, this method is easy to follow, and the results show that the proposed scheme has less overshoot,
shorter settling time and is more robust to cross-coupling disturbances.

1 Introduction identication of linear and nonlinear dynamic systems by


the analysis of input and output signals [9]. Kim and
Helicopters are typically described in the literature as having Calise [10] have employed neural networks to perform dual
unstable, nonlinear, time varying and coupled dynamics. roles of identifying the input output model parameters
Many suites of PID control modules have been employed (ofine learning) using the mathematical model of an
to produce agile, low observable and multi-functional aircraft and an adaptive network that compensates for
helicopters capable of performing complicated missions imperfect inversion and in-ight changes in the actual
[1 7]. One of the problems with such simple PID aircraft dynamics. Talebi et al. [11] have carried out an
approaches is that the tuning of gains is noted to be investigation into the dynamic modelling of a exible-link
tedious. In recent years, more advanced control approaches manipulator using neural networks. An innovative time-
such as gain scheduling [3] and linearisation feedback [4, domain nonlinear mapping-based identication method has
5] have been applied with considerable success. However, been addressed by Lyshevski [12] for the identication of
these techniques require a dynamic model for their design. unsteady ight dynamics. Bruce and Kellet [13] have
The development of such dynamic models for small size investigated B-splines in modelling and identication of
helicopters has commanded considerable research focus nonlinear aerodynamic functions of aircraft. In all these
over recent years [6, 7]. cases, the model structure is known. Shaheed and Tokhi
[14] proposed an approach which yields input output
A considerable effort has been made to devise models with neither a priori dened model order nor
methodologies to identify and control systems with specic parameter settings reecting any physical aspects.
nonlinearity and uncertain dynamics. For instance, Blythe The approach is, thus, useful in modelling a class of air
and Chamitoff [8] have used neural networks to estimate vehicles whose dynamics are not well understood. A similar
the aerodynamic coefcients of unmanned air vehicles. approach has previously been addressed by Ahmad et al.
Chon and Cohen [9] have addressed parametric system [15]. They used radial base function feed forward neural

IET Control Theory Appl., 2008, Vol. 2, No. 11, pp. 999 1007 999
doi: 10.1049/iet-cta:20070335 & The Institution of Engineering and Technology 2008
www.ietdl.org

networks with well-known orthogonal least square algorithm a helicopter. It can be well perceived as a static test rig for
to model a twin rotor multi-input multi-output (MIMO) an air vehicle with formidable control challenges [2].
system (TRMS).
This TRMS consists of a beam pivoted on its base in such
This study addresses the decoupling and robust deadbeat a way that it can rotate freely in both its horizontal and
control of a TRMS based on the available models and vertical planes. There are two rotors (the main and tail
techniques. First, the system model is identied. It is rotors), driven by DC motors, at each end of the beam. If
shown, that the identied system is able to be decoupled necessary, either or both axes of rotation can be locked by
into two single-input single-output (SISO) systems, and means of two locking screws provided for physically
the crossing couplings can be considered as disturbances to restricting the horizontal or vertical plane rotation. Thus,
each of the SISO systems. Then, a PID-based robust the system permits both 1 and 2 degree-of-freedom (DOF)
deadbeat control scheme is applied to the SISO systems, experiments. The two rotors are controlled by variable
and a deadbeat controller is designed for each of them. speed electric motors enabling the helicopter to rotate in a
These robust deadbeat controllers can tolerate system vertical and horizontal plane (pitch and yaw). The tail rotor
parameter changes for up to 50% [16]. This feature is used could be rotated in either direction, allowing the helicopter
to surpass the cross-coupling effects between the main and to yaw right or left. The motion of the helicopter was
tail rotors. This design is investigated in simulations using damped by a pendulum, which hung from a central pivot
Simulink, and compared with a system with two point. In a typical helicopter, the aerodynamic force is
independent PID controllers. The results show that the controlled by changing the angle of attack of the blades.
proposed scheme has less overshoot, shorter settling time The laboratory setup is constructed such that the angle of
and is more robust to cross-coupling disturbances. The attack of the blades is xed. The aerodynamic force is
nal result is tested in a TRMS in the lab. controlled by varying the speed of the motors.

The paper is organised as follows. The TRMS is introduced The mathematical model of the TRMS is developed under
and described in the system and modelling section. The following assumptions.
system model is then presented. This is followed by the
robust deadbeat control method and procedure section, The dynamics of the propeller subsystem can be described
where the details of the robust deadbeat control scheme and by rst-order differential equations.
our design are addressed. The system is then simulated
using simulink in the simulation and experiment section. In The friction in the system is of the viscous type.
this section, we also analysed and compared the simulated
results. Finally, the proposed method is tested in a TRMS The propeller air subsystem could be described in
and the main ndings of this study are summarised in the accordance with the postulates of the ow theory.
conclusion.
The mechanical system of TRMS is simplied using a four
point-mass system shown in Fig. 2, includes main rotor, tail
2 System and modelling rotor, balance-weight and counter-weight. Based on
Lagranges equations, we can classify the mechanical system
Similar to most ight vehicles, the helicopter consists of into two parts, the forces around the horizontal axis and
several elastic parts such as rotor, engine and control the forces around the vertical axis.
surfaces. The nonlinear aerodynamic forces and gravity act
on the vehicle, and exible structures increase complexity
and make a realistic analysis difcult. For control purpose,
it is necessary to nd a representative model that shows the
same dynamic characteristics as the real aircraft. The twin
rotor MIMO system or TRMS is a laboratory setup
designed for ight control experiments. The schematic
diagram of the laboratory setup is shown in Fig. 1. The
behaviour of the TRMS in certain aspects resembles that of

Figure 1 TRMS diagram Figure 2 Simplied four point-mass system

1000 IET Control Theory Appl., 2008, Vol. 2, No. 11, pp. 999 1007
& The Institution of Engineering and Technology 2008 doi: 10.1049/iet-cta:20070335
www.ietdl.org

The parameters in the simplied four point-mass system are The tail rotor model

Mvl is the return torque corresponding to the force of dSh


It Sf Fh (vt ) cos av  Vh kh (4)
gravity, Mmr is the mass of the DC motor within the dt
main rotor, Mm is the mass of the main part of the
beam, Mtr is the mass of the DC motor within tail da h S Jmr vm cos av
Vh h
rotor, Mt is the mass of the tail part of the beam, Mcb is dt Jh
the mass of the counter weight, Mb is the mass of the Sh Jmr vm cos av
counter-weight beam, Mms is the mass of the main (5)
D sin2 av E cos2 av F
shield, Mts is the mass of the tail shield, Im is the length
of the main part of the beam, It is the length of the tail dSh
part of the beam, Ib is the length of the counter-weight Mh (6)
dt
beam, Icb is the distance between the counter-weight and
joint, and g is the gravitational acceleration. where Jmr is the moment of inertia in DC-motor main
propeller subsystem, Jtr the moment of inertia in DC-
Consider the rotation of the beam in the vertical plane motor tail propeller subsystem, Sh the angular momentum
(around the horizontal axis). The driving torques are in the horizontal plane of the beam, Sv the angular
produced by the propellers, and the rotation can be momentum in the vertical plane of the beam and Sf the
described in principle as the motion of a pendulum. We balance scale.
can write the equations describing this motion as follows.
Furthermore, the angular velocities (vm , vt) are nonlinear
The main rotor model functions of the input voltage of the DC motor (uv , ut), and
the model of the motor propeller dynamics is obtained by
dSv substituting the nonlinear system by a serial connection of a
Im Sf Fv (vm )  Vv kv g((A  B) cos av  C sin av ) linear dynamics system. This can be expressed as
dt
1 duvv 1
 V2h (A B C) sin 2av (1) (  uvv uv ) (7)
2 dt Tmr

da v J v vm Pv (uvv ) (8)
Vv Sv tr t (2)
dt Jv
duhh 1
(  uhh uh ) (9)
dt Ttr
dSv
Mv (3) vt Ph (uhh ) (10)
dt

Similarly, we can describe the motion of the beam in the where Tmr is the time constant of the main rotor propeller
horizontal plane (around the vertical axis) as shown in system and Ttr is the time constant of the tail motor
Fig. 3. The driving torques are produced by the rotors and propeller system.
that the moment of inertia depends on the pitch angle of
the beam. The static characteristics of the propellers are measured
using a proper electronic balance with voltage output [7].
Thus, we can identify the following nonlinear functions:
two nonlinear input characteristics determining the
dependence of DC-motor rotational speed on input voltage

vm Pv (uvv ) (11)

vt Ph (uhh ) (12)

Two nonlinear characteristics determining the dependence of


propeller thrust on DC-motor rotational speeds

F h F h (v t ) (13)

F v F v (v m ) (14)

Based on the above equations, a system diagram can be drawn


Figure 3 Torques around the vertical axis as shown in Fig. 4.

IET Control Theory Appl., 2008, Vol. 2, No. 11, pp. 999 1007 1001
doi: 10.1049/iet-cta:20070335 & The Institution of Engineering and Technology 2008
www.ietdl.org

Figure 4 Block diagram of TRMS system

It is obvious that this is a high-order, non-linear and cross- T (s) as an example


coupled system. The details of the TRMS and its model are
provided in [17]. v4n
T (s) (15)
s4 avn s3 bv2n s2 gv3n s v4n

3 Robust deadbeat control First, normalising the system by dividing the numerator and
method and procedure denominator by v4n

Fig. 5 shows the basic structure of the PID-based robust 1


deadbeat control system, and Table 1 is the look-up table T (s)
(s=vn )4 a(s=vn )3 b(s=vn )2 g (s=vn ) 1
for controller designing. This technique is initially
proposed by Dorf et al. in 1994 [16]. However, this (16)
technique works only for lower-order plants. As a result,
there is a need for higher gain when higher-order systems Let S s=vn to obtain
are considered. This design with a proper high gain will
result in systems that are insensitive to plant parameter 1
T (s) (17)
variations of up to 50% [16]. s as bs2 gs 1
4 3

To show the design procedure of a PID-based robust Equation (17) is the normalised, fourth-order and closed-
deadbeat control, we take the following fourth-order system loop transfer function. For a higher-order system, the same
method is used to derive the normalised equation. The
coefcients of the equation a, b and g are selected from
Table 1. Taking the above fourth order system as an
example, with a required settling time of 0.95 s, we can
nd the normalised settling time from Table 1

vn Ts 4:81

Figure 5 Robust deadbeat control structure Therefore we require

Table 1 Robust deadbeat control look-up table [16] 4:81 4:81


vn 5:063
Ts 0:95
Order (np) a b g d TS0
2nd 1.82 4.82 The characteristic equation of the closed-loop transfer
function is
3rd 1.90 2.20 4.04
4th 2.20 3.50 2.80 4.81 s4 avn s3 bv2n s2 gv3n s v4n
5th 2.70 4.90 5.40 3.40 5.43
From Table 1, we have
Note: Ts is the settling time based on
normalised vn . a 2:20; b 3:50; g 2:80

1002 IET Control Theory Appl., 2008, Vol. 2, No. 11, pp. 999 1007
& The Institution of Engineering and Technology 2008 doi: 10.1049/iet-cta:20070335
www.ietdl.org

Hence, the transfer function is where

T (s)
657:1 K [K3 (s2 Xs Y )]
s4 11:1386s3 89:71889s2 363:397s 657:1 G1 (s) Gc (s)
s
(18)
15:02
G2 (s)
To apply the above technique to our TRMS, rst the s3
3:458s2 2:225s
decouple techniques are required to separate the system 15:02
G(s)
into two SISO systems. Without angular momentum and s(s 2:603)(s 0:8547)
reaction turning moment, the TRMS system is modelled
into two 1-DOF systems: vertical part (main rotor) and H1 (s) (1 Kb s)
horizontal part (tail rotor) shown in Figs. 6 and 7.
H2 (s) Ka

The closed-loop transfer function can be written as


3.1 Tail rotor controller design
Now, we apply the above robust deadbeat control scheme to C(s) 15:02K [K3 (s2 Xs Y )]
4
the tail rotor system. The block diagram of the designed R(S) s {3:458 15:02Kb KK3 }s3 {2:225 15:02KK3
system is shown in Fig. 8. 15:02KKb K3 X }s2 {15:02Ka 15:02KK3 X
15:02KKb K3 Y }s {15:02KK3 Y }
Determine the closed-loop transfer function
Next, we can apply Dorfs method to determine these
C(s) G1 (s)G2 (s) parameters. In this method, the characteristic equation of

R(s) 1 G2 (s)H2 (s) G1 (s)G2 (s)H1 (s) the above transfer function is equal to the characteristic
equation of the deadbeat transfer function. To obtain the
characteristic equation of the deadbeat transfer function, we
set the characteristic equation of the closed-loop transfer
function as follows

s4 avn s3 bv2n s2 gv3n s v4n

From looking up Table 1, we have

a 2:20; b 3:50; g 2:80


Figure 6 Vertical robust control system structure
Ts0
vn
80% of the desired settling time Ts

If we choose the desired settling time as 2 s, then we have

Ts0 4:81
vn 3:00625
Ts  80% 1:6

Therefore the characteristic equation of the deadbeat transfer


Figure 7 Horizontal robust control system structure function is

s4 6:6138s3 31:6314s2 76:0735s 81:6771

Comparing the characteristic equation and let K equal to 1,


we have

{7 Kb K3 } 11:1386

{14 K3 Kb K3 X } 89:71889

{8 Ka K3 X Kb K3 Y } 363:397

Figure 8 Robust tail rotor control {K3 Y } 657:1

IET Control Theory Appl., 2008, Vol. 2, No. 11, pp. 999 1007 1003
doi: 10.1049/iet-cta:20070335 & The Institution of Engineering and Technology 2008
www.ietdl.org

Hence we have

Kb 0:243; Ka 45:848 Kb 0:5; Ka 2:5453

K3 17; X 14:21; Y 38:6529 K3 7:723; X 3:131; Y 6:963

Finally, the robust deadbeat controllers designed based on


Adjust K until the system obtains a deadbeat response.
two SISO systems can be put into our TRMS system and
form a robust deadbeat control MIMO system shown in
Fig. 9.
3.2 Main rotor controller design
Similar to the previous section, we can apply this technique to
the main rotor control of TRMS system. First, let us 4 Simulation and experiment
determine its closed-loop transfer function
To thoroughly investigate the system performance, the
proposed control schemes were implemented and tested in
C(s) G1 (s)G2 (s)
simulation using Simulink and Matlab Control Toolbox.
R(s) 1 G2 (s)H2 (s) G1 (s)G2 (s)H1 (s) We start with 1-DOF system by xing another freedom.
The specications we investigated are overshoot, settling
where time and steady-state error.

K [K3 (s2 Xs Y )] For 1-DOF vertical control, the settling time is desired to
G1 (s) Gc (s)
s be 2 s. At K 10, we obtained the desired response. The
system response is shown in Fig. 10.
1:519
G2 (s)
s3
0:748s2
1:533s 1:046 For 1-DOF horizontal control, we also choose a settling
1:519 time of 2 s. We obtain the desired performance at K 7,
2
G(s)
(s 0:6982)(s 0:04983s 1:498) which is shown in Fig. 11.

H1 (s) (1 Kb s); H2 (s) Ka For the 2-DOF system shown in Fig. 12, we choose the
settling time of both tail and main rotors as 4 s. By tuning
The closed-loop transfer function can be written as each K both in horizontal and vertical control, we obtain
the desired system response which is shown in Fig. 13.
K [K3 (s2 Xs Y )]
0 1
s4 (0:748 1:519KK3 Kb )s3 Figs. 10, 11 and 13 illustrate the responses of the TRMS
B (1:533 0:1549KK3 )s2 l C system in three different situations. It can be seen that all the
B C
@ (1:046 1:519K 0:1549KK K Y )s A responses settle within a given time frame. In the 2-DOF
a 3 b
1:519KK3 Y case with the introduction of cross-coupling disturbances,
both tail and main rotors reach the desired positions within
The characteristic equation of the closed-loop transfer given time frames. Although there are overshot, the system
function is equal to responses still meet all the specications. This is evident
that the scheme is robust against the cross-coupling.
s4 avn s3 bv2n s2 gv3n s v4n

From Table 1, we have

a 2:20; b 3:50; g 2:80

We choose the settling time as 2 s again, then we obtain

Ts0 4:81
Ts 2 s; vn 3:00625
Ts  80% 1:6

Therefore

s4 6:6138s3 31:6314s2 76:0735s 81:6771

Comparing the characteristic equation and let K equal to 1, Figure 9 robust deadbeat control MIMO system

1004 IET Control Theory Appl., 2008, Vol. 2, No. 11, pp. 999 1007
& The Institution of Engineering and Technology 2008 doi: 10.1049/iet-cta:20070335
www.ietdl.org

Figure 10 Vertical response of the main rotor (K 10) Figure 13 Two-DOF robust deadbeat control response
Notes: The input and output of the main rotor are shifted down to
avoid the overlaps in plot

Fig. 14 shows the response of a traditional PID control


TRMS. The parameters of trail rotor PID controller are
Kp 1.1002, Ki 2.87 and Kd 2.87. The parameters of
the main rotor PID controller are Kp 0.077, Ki 0.385
and Kd 1.186. Comparing Figs. 13 and 14, it is obvious
that the robust deadbeat control performance is much
better than the traditional PID control.

Finally, this design is experimentally tested on our TRMS


system in the lab. In this system, we implemented the robust
deadbeat control scheme using Matlab, real-time windows
and real-time workshop in a PC computer. The control
and the data acquisition are implemented using the
equipped PC1711 interface cards. The experiments are
carried out in calm air in our lab as the TRMS is very
sensitive to atmospheric disturbances. The input signal
frequency is chosen to enable the system to reach its steady
Figure 11 Horizontal response of the tail rotor (K 7) state. The results we obtained are different from but similar

Figure 12 Two-DOF system block diagram for simulation

IET Control Theory Appl., 2008, Vol. 2, No. 11, pp. 999 1007 1005
doi: 10.1049/iet-cta:20070335 & The Institution of Engineering and Technology 2008
www.ietdl.org

the settling time has been shortened to 20 s and the


overshoot has been reduced about 20% in the tail rotor and
completely eliminated from the main rotor. Comparing
these responses, we can clearly note the following

For the two SISO systems: The settling time has been reduced
to about 6 and 12 s in tail and main rotors, respectively; the
amount of overshoot has been reduced about 20%.

For the 2-DOF system: The settling time has been reduced
to 20 s; the amount of overshoot has also been reduced as well.

This model-based design procedure, however, needs a


accurate system transfer function, especially for the main
rotor. To further improve the system response we need to
improve the accuracy of system model. For the cross-
coupling effects, another solution is to design a decoupler,
Figure 14 Two-DOF PID scheme responses which will cancel the effects of the cross-coupling effects
among the main and tail rotors.
to the simulation results. However, the main rotor takes four
more seconds to settle and the tail rotor has a small overshot.
We also tried different amplitudes of the input signal. We
noticed that the amplitudes of the input signal do not 6 References
affect the response much in the normal operation scale. We
predicted but did not nd the strong effect from the [1] SHAHEED M.H.: Feed forward neural network based
nonlinear elements. We believed that this is because of the nonlinear dynamic modelling of a TRMS using RPROP
robust deadbeat controller which is able to tolerate the large algorithm, Aircr. Eng. Aerosp. Technol., 2005, 77, (1),
range of parameter changes. We also implemented the pp. 13 22
traditional PID control scheme to this TRMS system. In
the beginning, we use the simulation parameters. However, [2] BUSKEY G., ROBERT J., WYETH G.: A helicopter named dolly
the system did not work at all. After, we re-adjusted them behavioural cloning for autonomous helicopter control.
manually, the PID control started to work but both the tail Proc. Australasian Conf. Robotics and Automation,
rotor and main rotor had big overshoots before they reach Brisbane, December 2003
their specied positions.
[3] SPRAGUE K. , GAVRILETS V. , DUGAIL D. , METTLER B. , FERON E. :
Design and applications of an avionics system for a
miniature acrobatic helicopter. Digital Avionics Systems
5 Conclusions Conf., Daytona Beach, Florida, October 2001, p. 3.C.5-
In this study, we have successfully modelled a TRMS whose 1 3.C.5-10
dynamics resemble that of a helicopter. The extracted model
has been decoupled into two SISO systems. We applied a [4] KOO T.J., SASTRY S.: Output tracking control design of a
PID-based robust deadbeat control scheme to these two helicopter model based on approximate linearization,
SISO systems and designed a deadbeat controller for each IEEE Conf. Decis. Control, 1998, 4, pp. 3635 3640
of them. This design can tolerate system parameter changes
to about 50% without degrading system performance. [5] PRASAD J.V.R., CALISE A.J. , PEI Y. , CORBAN J.E.: Adaptive
We have successfully employed this robust feature to nonlinear controller synthesis and ight test evaluation.
suppress the cross-coupling effects between the tail and the Proc. 1999 IEEE Int. Conf. Control Applications, Kohala
main rotors. Coast, Island of Hawaii, August 1999, pp. 137 142

Comparing the system responses obtained using individual [6] BUSKEY G. , ROBERTS J., WYETH G. : Online learning of
PID controllers, the proposed robust deadbeat control autonomous helicopter ontrol. Australasian Conf.
scheme is simpler and the performance is better. In PID Robotics and Automation, Auckland, New Zealand,
control design in a 2-DOF case, it has six parameters to December 2002, pp. 19 24
tune. However, we have reduced them to two in our robust
deadbeat control scheme. In addition, this control scheme [7] GAVRILETS V., METTLER B., FERON E.: Nonlinear model for a
does not include any complicated math and calculation small-size acrobatic helicopter. AIAA Guidance, Navigation
except the normalisation and look-up table. It is easily and Control Conf., Montreal, Canada, August 2001,
accepted by industrial designers. In system performance, pp. 1593 1600

1006 IET Control Theory Appl., 2008, Vol. 2, No. 11, pp. 999 1007
& The Institution of Engineering and Technology 2008 doi: 10.1049/iet-cta:20070335
www.ietdl.org

[8] BLYTHE P.W., CHAMITOFF G.: Estimation of aircrafts [13] BRUCE P.D., KELLET M.G.: Modelling and identication of
aerodynamic coefcients using recurrent neural non-linear aerodynamic functions using b-splines, Proc.
networks. Proc. 2nd Pacic Int. Conf. Aerospace Science Instn. Mech. Eng. G, 2000, 214, pp. 27 40
and Technology, Australia, 1995
[14] SHAHEED M.H., TOKHI M.O.: Dynamic modeling of a single-
[9] CHON K.H., COHEN R.J.: Linear and non-linear ARMA model link exible manipulator: parametric and non-parametric
parameter estimation using an articial neural network, IEEE approaches, Robotics, 2002, 20, pp. 93 109
Trans. Biomed. Eng., 1997, 44, (3), pp. 168 174
[15] AHMAD S.M., SHAHEED M.H., CHIPPERFIELD A.J., TOKHI M.O.: Non-
[10] KIM B.S., CALISE A.J.: Non-linear ight control using neural linear modeling of a one-degree-freedom of twin-rotor
networks, J. Guid. Control Dyn., 1998, 20, (1), pp. 26 33 multi-input multi-output system using radial basis
function networks, Proc. Instn. Mech. Eng. G, 2002, 216,
[11] TALEBI H.A., PATEL R.V., ASMER H.: Dynamic modelling of pp. 197 208
exible-link manipulators using neural networks with
application to the SSRMS. Proc. IEEE Int. Conf. Intelligent [16] DAWES J., NG L., DORF R.C., TAM C.: Design of deadbeat
Robots and Systems, Victoria, Canada, 1998 robust systems. Proc. 28th Annual Asilomar Conf. Signals,
Systems, and Computers, 1994, pp. 1597 1598
[12] LYSHEVSKI S.E. : Identication of non-linear ight
dynamics: theory and practice, IEEE Trans. Aerosp. [17] Twin Rotor MIMO System Manual, Feedback
Electron. Syst., 2000, 36, (2), pp. 383 392 Instruments Ltd., UK, 2002

IET Control Theory Appl., 2008, Vol. 2, No. 11, pp. 999 1007 1007
doi: 10.1049/iet-cta:20070335 & The Institution of Engineering and Technology 2008

También podría gustarte