You are on page 1of 4

Proceedings of the 3d World Congress on Intelligent Control and Automation

June 28-July 2,2000, Hefei, P.R China


Stabilizing Control of an Inverted PendulumSystem
Based on H- Loop Shaping Design Procedure
Sek Un Cheang Wei J i Chen
Faculty of Science and Technology
University of Maau
Macao, china
Abstract - Loop Shaping Design Procedure (LSDP) is oneof the
most promising fb. design procedure, which incorporates loop
shaping methods to obtain performance or robust stability
tradeoffs. The advantage is that LSDP does not require y-
iteration for its solution, and explicit formulas for the
corresponding controller are available. Many successful
controllers designed by the methods obtain the satisfactory
results, such as aerospace applications. In the paper, wedesign
a robust controller for stabilizing an inverted pendulumsystem
by LSDP. Simulation results show that the designed controller
works satisfactorily.
Key Words: Loop Shaping Design Procedure (LSDP), h
Control, an inverted pendulum.
I. LSDP
Loop shaping is the classical approach in which the
magnitude of the open-loop transfer function IL(jo)l is
shaped. Here L(jo)=G(jo)*&(jo) where K,,(jo) is the
feedback controller to be designed and G(jo) is the transfer
function of the controlled system model, including the plant,
the actuator and the measurement device, etc. Essentially, to
get the benefits of feedback control we want the loop gain,
IL(jo)l, to be as large as possible within the bandwidth
region. However due to time delays, RHP-zeros,
unmodelled high frequency dynamics and limitations on the
allowed manipulated inputs, the ,loop gain has to drop below
one at higher crossover frequency. Thus, disregarding
stability for the moment, it is desirable that IL(jo)l falls
sharply with frequency.
Although the procedure of classical loop shaping may
be extended to more complicated systems (MIMO), the
effort required by the engineer is considerably greater. In
particular, it may be very difficult to achieve stability.
Fortunately, in 1990, McFarland and Glover proposed the
loop-shaping design procedure which is based on H- robust
stabjlization combined with classical loop shaping. It is
essential to two stage design process. First, the open-loop
plant is augmented by pre and postcompensaton to give a
desired shape to the singular values of the open-loop
frequency response. Then, the resulting shaped plant is
robustly stabilized with respect to coprime factor uncertainty
using H= optimization. An important advantage is that no
problem-dependest uncertainty modeling, or weight
selection, is required in this second step. Furthermore, it
does not require y-iteration for its solution, and explicit
formulas for the corresponding controller are available. The
formulas are relatively simple and so will be presented in
full.
-
LSDP can be summarily depicted three systematic
1) Loop Shaping : Using a pre -compensator W1and/or
a post-compensator W2, the singular values of the
nominal plant are shaped to give a desired open-
loop shape, see Figure 1. The nominal plant G and
the shaping functions Wl, W2 are combined to form
the shaped plant, GI= W~GWI.
The gain of the shaped plant is high at low
frequencies, low at high frequencies and does not
roll off at a high rate near crossover. The shaped
plant G, would be expressed as a normalized left
coprime factorization G,=M,N,.
procedures as follows:
\
n
G
Figure 1 Loop shaping design procedure
2) Robust Stabilization: A controller K guarantees that
for a specified py,,,iw Then, the nominal feedback
system is stable.
0-7803-S99S-X/oo/sIO.OO Q2000 IEEE.
3385
The lowest achievable value y,,,j,, of y and the
corresponding maximumstability margin E am derived
by Glover and McFarlane (1989) as
where h a x denotes the spectral radius(maximum
eigenvalue), and from a state-space realization (A, B, C,
0) of G,, Z and X are the unique positive definite
solutions to the following algebraic Riccati equations,
(3)
( A - BS - ' D~C) Z +z ( A- BS- ' D~C) '
-ZC'R-'CZ+BS-'B' = o
(4)
( A - B S - ~ D ' C ~ x +X ( A - B S - ~ D ~ C )
- XBS-' BI x +C ~ R - I C =o
where
R =I +D D ~ , s =I +D ~ D
( 5)
A controller K={ALBK.CLD~ as state-space ealization
for a specified m~,, is given by
where
F =-S-I(D'C+ B'X) (7)
L =(1 - y2) I +xz
(8)
3) The final feedback controller K, is then constructed
by combining the H- controller K with the shaping
functions Wj and W2 such that
K, =Wl KW2 (9)
The designer follows the above the procedures and the
typical design works are stated as follows: the open-loop
singular values of the nominal plant are inspected and are
shaped by pre- and/or post-compensation until nominal
performance specifications are met. A feedback controller K
with associated stability margin for the shaped plant &E-
is then synthesized. If &- is small, then the specified loop
shape is incompatible with robust stability requirements, and
should be adjusted accordingly, then K isreevaluated.
3386
Figure 2 An inverted pendulum system
It is important to emphasize that since we can compute
"(mi. from(2). weget an explicit solution by solving just two
Riccati equations (3) and (4) and avoid the yiteration needed
to solve the general H- problem.
11. AN INVERTED PENDULUM
An Inverted PendulumSystem(1PS) is one of the most
well-known equipment in the field of control systems theory.
It E inexpensive and can be easily built and installed in
laboratories for control education purposes or for research
applications.
An inverted pendulum system on a moving cart treated
in this paper is schematically sketched in Figure 2 and is
cart-pole type of inverted pendulum. The pole is hinged to
the cart and can be rotated in the vertical plane containing
the line of the rail. The cart moves left or right on the rail by
the applied control force F (N). 63 (rad) is the angles of the
pole from the vertical line and r (m) is the distance of the
cart from the center of the rail. Under some assumptions
(FO and a=O), the mathematical model of the pendulum
can be constructed based on the Lagrange equations. The ,
state-space equations are depicted as follows:
Y =c,x
where
x= [r e, i. el]T
U is the control signal of the system, which is the input of
the servo motor.
system are listed as follows:
System's physical parameters of an inverted pendulum
lO.0037Kg.m'
IGain constant from the input voltage
Go
to the amplifier to the force to the
cart 8.646"
Equivalent viscous friction
coefficient, 18.96N.m~
Viscous friction coefficient of the
le, 0.002N.m~
Hence, .
0 1 0 .
A , = F 0
]
-0.794 -6.814 0.003
0 42.872 27.545 -0.157
B, =[O 0 3.107 -12.561]'
c, =
0 0 1 0
0 0 0 1 =)I
111. SYSTEM DESIGN FOR IPS
system and the input voltage to the motor, namely
maxlrl c 0.35 m,
maxlq I c ~.l rud,und
maxbl e 15Y
Considering the design requirements of the pendulum
and based on the LSDP, the selection of pre-compensator is
most important design's work. For the inverted pendulum
system, the proposed pre-compensator is a low pass filter.
Through the low pass filter, the LSDP controller has the
observation noise reduction characteristic in high frequency.
Moreover, we consider the noise comes from the 50Hz
power supply and raise the open loop gain of system, we
define the pre-compensator Wl as follows:
,le
J J
.vi =s+so
Furthermore, the post-compensator is selected as a
constant diagonal matrix with positive constant real value to
achieve the design requirements. The post-compensator is
weighting matrix. Although there is no general guideline for
the selection of weighting matrix, we know that the IPS is
stabilized by minimizing the angle of the pole. The largest
weight of e) is chosen compared with others. Finally, the
extensive computer simulations have been executed. We
found post-compensator Wz as,
r5 o o 01
0 2 0 0 0
wz=lo 0 1 01
10 0 0 1J
The loop shaping of G,=W,GWt is shown in Figure 3.
Obviously, the singular value of system is large in low
frequency and decrease fast in high frequency to rejecting
the noise in high frequency.
Figure 3 Open-loop Singular Value of Nominal Plant G and
Shaped Plant Cj
The lowest achievable value y,,,j. of G, isobtained as
3.622 by the algorithmderived by Glover and McFarlane
(1989). The special y1.05y,,,in is chosen and the final
feedback controller KS is found. The singular value of open-
loop transfer function L=G*K, is drawn in Figure 4. Due to
the system is sampled at a sampling period of T (3ms) and
zem-order-hold method is used on the inputs and outputs for
each sampling period. The stateapace discrete representation
of feedback controller &J (AM, &d, &, &J ) is shown as
follows,
3387
0.052
0.003
0.003
0.287
-1.126
0.808
-0.030 -0.014
0.003 -0.032
o.Oo0 -0.004
0.984 a009
0.065 a914
0.030 0.015
0.001
-0.003
- 0.007
0.006
- 0.072
0.0
IV. CONCLUSION
The control system design using LSDP for an inverted
pendulum system has been successfully done. The LSDP
controllers for the uncertain dynamic systems are very
effective, which is shown in the paper. The simulation
results give US satisfactory. The LSDP is a systematic
C,=[35 0 0 0 0 OIT;DKd=O
A, =
B, =
a861 -0.124 -0.069
0 0.992 -0.121
0 -0.008 0.709
0 0.004 0.201
0 -0.086 -2707
0 0.013 0.098
0.01 0.038 0.0
-0.008 -0.121 om1
-0.008 -0.291 0.001
0.001 0.186 -0.001
-0.077 -2.760 0.006
ao -0.009 0.0
V. REFERENCE
1) D.C. Mcfarlane and K.Glover: Robust Controller
Design Using Normalized Coprime Factor Plant
Description, Lecture Notes in Control and Information
Sciences 138, Springer-Verlag,l990.
Figure 4 Singular Value of Nominal Plant G, Shaped Plant
G, and Open-Loop Transfer Function L=G*K,
The simulation results are shown in Figure 5.
Cart podlion
lime
Pendulum Angle
0 1 2 3 4 5 8 1 8 9
time
- Figure 5 Simulation Results
2) Fang Lei: Control of Double Link Inverted Pendulum
with Nonlinear Friction, master thesis, Faculty of
Science and Technology, University of Macau, 1999.
3) K. Zhou, J.C. Doyle and K. Glover: Robust and Optimal
Control. Prentice Hall, first edition, 1996.
.
4) S. Skogestad an4 I. Postlethwaite: Multivariable
Feedback Control. J ohn Wiley & Sons, first edition.
1996.
3388

You might also like