0% ont trouvé ce document utile (0 vote)
32 vues110 pagesChapitre 12
design of control system
Transféré par
maniCopyright
© © All Rights Reserved
Nous prenons très au sérieux les droits relatifs au contenu. Si vous pensez qu’il s’agit de votre contenu, signalez une atteinte au droit d’auteur ici.
Formats disponibles
Téléchargez aux formats PDF ou lisez en ligne sur Scribd
0% ont trouvé ce document utile (0 vote)
32 vues110 pagesChapitre 12
design of control system
Transféré par
maniCopyright
© © All Rights Reserved
Nous prenons très au sérieux les droits relatifs au contenu. Si vous pensez qu’il s’agit de votre contenu, signalez une atteinte au droit d’auteur ici.
Formats disponibles
Téléchargez aux formats PDF ou lisez en ligne sur Scribd
12
Design of Control
Systems in State Space
12-1 INTRODUCTION
786
‘Thic chapter first diseusses the pole-placement problem. We discuss the design of co
trol systems by pole placement. We design an inverted-pendulum system in which the
pendulum is subjected to initial conditions and we want to bring the pendulum back to
the vertical position within a specified time. Then we discuss the design of state observers
Finally, we treat servo systems design where (1) the plant has an integrator and (2) the
plant has no integrator. We include a discussion of designing an inverted-pendulum sys-
tem such that when the cart,on which the pendulum is mounted, is given a step i
system is stable (that is, the pendulum will not fall down).
Outline of the chapter. Section 12-1 has presented introductory material. Sec-
tion 12-2 diecuseor the pole placement approach to the desiga of conteal systems, We
begin with the derivation of the necessary and sufficient conditions for arbitrary pole
placement. Then we derive equations for the state feedback gain matrix K for pole
placement. Section 12-3 presents the solution of the pole-placement problem with
MATLAB In this section we give sample MATLAB programs for pole-placement de-
sign. Section 12-4 gives an example of the design of regulator type systems by the pole-
placement approach, We use the inverted-pendulum system as an example, Both the
‘analytical solution and the MATLAB solution are presented.
Section 12-5 presents state observers, We discuss both full-order and minimum-
order state observers. Three different approaches for the determination of the ob-
setver gain matrix Ke are presented. The concept of the controtier observer is introduced, Section 12-6 discusses the design of state observers with MATLAB. Sec-
tion 12-7 treats the design of servo systems. In particular. we discuss the design of type
1 servo systems when (1) the plant has an integrator and (2) the plant has no integra-
tor. Section 12-8 presents an example of control system design with MATLAB. We
specifically treat the design of an inverted-pendulum control system with MATLAB.
Unit-step response curves of the designed system are obtained by using MATLAB.
12-2 POLE PLACEMENT
In this section we shall present a design method commonly called the pole-placement or
pole-assignment technique. We assume that all state variables are measurable and are
available for feedback. It will be shown that if the system considered is completely state
controllable, then poles of the closed-loop system may be placed at any desired loca-
ons by means of state feedback through an appropriate state feedback gain matrix.
‘The present design technique begins with a determination of the desired closed-loop
poles based on the transient-response and/or frequency-response requirements, such as
speed, damping ratio, or bandwidth, as well as steady-state requirements,
Let us assume that we decide that the desired closed-loop poles are to be at s = ju.
5 = ya... 8 = A By choosing an appropriate gain matrix for state feedback, itis pos-
sible to force the system to have closed-loop poles at the desired locations, provided that
the original system is completely state controllable.
In what follows, we shall treat the case where the control signal isa scalar and prove
that a necessary and sufficient condition that the closed-loop poles can be placed at any
arbitrary locations in the s plane is that the system be completely state controllable.
‘Then we shall discuss three methods for determining the required state feedback gain
mateix.
It is noted that when the control signal isa vector quantity, the mathematical aspects
of the pote-placement scheme become complicated. Therefore, we shall not discuss such
aa case in this book. (For the interested reader, refer to Ref. O-3 for the mathematical
derivation in such a case.) Its also noted that when the control signal is a vector quan-
tity the state feedback gain matrix is not unique. Itis possible to choose freely more than
‘ parameters; that is,in addition to being able to place 1 closed-loop poles properly, we
have the freedom to satisfy some or all of the other requirements, if any. of the closed:
loop system.
Design by pole placement. In the conventional approach to the design ofa single-
input-single-output control system, we design a controller (compensator) such that the
dominant closed-loop poles have a desired damping ratio Zand undamped natural fre-
quency wp, In this approach, the order of the system may be raised by I or 2 unless
pole-zero cancellation takes place. Note that in this approach we assume the effects on
the responses of nondominant closed-loop poles to be negligible.
Different from specifying only dominant closed-loop poles (conventional design ap-
proach), the present pole-placement approach specifies all closed-loop poles. (There is
a cost associated with placing all closed-loop poles, however, because placing al closed-
oop poles requires successful measurements of all state variables or else requires the
Inclusion of a state observer in the system.) There is also a requirement on the part of
Section 12-2 / Pole Placement 187 the system for the closed-loop poles to be placed at arbitrarily chosen locations. The re-
quirement js that the system be completely state controllable, We shall prove this fact,
in this section.
Consider a control system
Ax + Bu (2-4)
where x
tate vector (n-vector)
u = control signal (scalar)
A= nn constant matrix
B =n 1 constant matrix
We shall choose the control signal to be
w= -Kx (12-2)
This means that the control signal is determined by an instantaneous state. Such a
scheme is called state feedback. The 1 % n matrix K is called the state feedback gain ma-
trix. In the following analysis we assume that u is unconstrained.
Figure 12-1(a) shows the system defined by Equation (12-1). Itis an open-loop con-
trol system, because the state x is not fed back to the control signal u. Figure 12-1(b)
shows the system with state feedback. This is a closed-loop control system, because the
state x is fed back to the control signal 1.
‘Substituting Equation (12-2) into Equation (12-1) gives
3) = (A - BK)x()
‘The solution of this equation is given by
x(1) =e ™™K(0) (123)
where x(0) is the initial state caused by external disturbances. The stability and tran-
sient-response characteristics are determined by the eigenvalues of matrix A BK. If
‘matrix K is chosen properly, the matrix A ~ BK can be made an asymptotically stable
matrix, and for all x(0) # 0 itis possible to make x() approach 0 as ¢ approaches in
ity. The eigenvalues of matrix A ~ BK are called the regulator poles If these regulator
poies are placed in the left-half s plane, then x(1) approaches 0 as t approaches infinity
The problem of placing the closed-loop poles at the desired location is called a pole-
placement problem.
f x | B L ~
L: | :
x
ee
Chapter 12. / Design of Contiol Systems In State Space In what follows, we shall prove that arbitrary pole placement for a given system is
possible if and only if the system is completely state controllable.
Necessary and sufficient condition for arbitrary pole placement. Consider
the control system defined by Equation (12-1). We assume that the magnitude of the
control signal u is unbounded. If the control signal z is chosen as
w= —Kx
where K is the state feedback gain matrix (1X n matrix), then the system becomes a
closed-loop control system as shown in Figure 12-1(b) and the solution to Equation
(12-1) becomes as given by Equation (12-3), or
x(0) = &4-BKIA(Q)
Note that the eigenvalues of matrix A — BK (which we denote 1,
desired closed-loop poles.
We shall now prove that a necessary and sufficient condition for arbitrary pole
placement is that the system be completely state controllable. We shall first derive the
necessary condition. We begin by proving that ifthe system is not completely state con-
trollable then there ate eigenvalues of matrix A — BK that cannot be controlled by
state feedback.
‘Suppose that the system of Equation (12-1) ts not completely state controllable.
‘Then the rank of the controllability matrix is less than m, or
rank [B | AB
‘This means that there are q linearly independent column vectors in the controllability
sla) ae the
AU IB]=q | An 7B) (12-5)
and
y=) Onan a 1
G2 Ans oe 1 0
w oe (12-6)
a 1 . o 0
1 0 . o 0
where the a;'s are coefficients of the characteristic polynomial
sl — A] =o" bast ss tages + ay
Define a new state vector Sby
xo Te
Ifthe rank of the controllability matrix M isn (meaning that the system is completely state
controllable), then the inverse of matrix T exists, and Equation (12-1) can be modified to
k= ‘ATS + T "Bu (12-7)
where
o 1 0 0
oo 1 0
TUAT=| 7 : (12-8)
o 0 0
Chapter 12. / Design of Control Systems in State Space T'B=|- (12-9)
0
1
{See Problems A-12-2 and A-12-3 for the derivation of Equations (12-8) and (12-9).]
Equation (12-7) is in the controllable canonical form. Thus, given a state equation,
Equation (12-1), it can be transformed into the controllable canonical form if the
system Is completely state controllable and if we transform the state vector x into
state vector & by use of the transformation matrix T given by Equation (12-4)
Let us choose a set of the desired eigenvalues as j1, 12,.-.. fle Then the desired
characteristic equation becomes
he ayst te es
(6 = pad(s = pea) = (= pe) (12-10)
Let us write
R= KTH (5, bey Bi czy
When u = —K& = —KT¢ is used to control the system given by Equation (12-7), the
system equation becomes,
R= T'ATR — T'BKTS
‘The characteristic equation is
[st - TAT + T'BKT| = 0
‘This characteristic equation is the same as the characteristic equation for the system,
defined by Equation (12-1), when u = — Kx is used as the control signal. This ean be
seen as follows: Since
Ax +B
the characteristic equation for this system is
|st — A + BK| = |T-'1~ A + BKIT| = [st - TAT + T'BKT| = 0
Now let us simplify the characteristic equation of the system in the controllable canon-
ical form. Referring to Equations (12-8). (12-9). and (1211), we have
ju Tare rm
0 0
= |at- de |
1] fo
Section 12-2 / Pole Placement 791
(A BKyx 792
(ay + By Ans + Beat
stats
SoM (ay FB) + py + Beals + (My + By) =O (12-12)
is isthe characteristic equation for the system with state feedback, Therefore, it must
bbe equal to Equation (12-10), the desired characteristic equation, By equating the co-
efficients of like powers of s,we get
ay + 8, = oy
ay + 8: = a2
A, + By = Oy
Solving the preceding equations for the 6/s and substitu
(12-11), we obtain
K=RT'= (8, Ga 7 6]
1% them imo Equation
= [on ~ an | Oar ann | fag — a | ay a)T! (12-13)
‘Thus, if the system is completely state controllable, all eigenvalues can be arbitrarily
placed by choosing matrix K according to Equation (12-13) (sufficient condition).
We have thus proved that the necessary and sufficient condition for arbi
placement is that the system be completely state controllable.
Design steps for pole placement. Suppose that the system is defined by
X= Ax + Bu
and the control signal is given by
w= Kx
“The feedback gain matrix K that forces the eigenvalues of A ~ BK to be jt 2, - fn
(desired values) can be determined by the following steps. (If 4 is a complex eigenvalue,
then its conjugate must also be an eigenvalue of A — BK.)
Step I: Check the controllability condition for the system. If the system is completely
state controllable, then use the following steps.
Step 2:From the characteristic polynomial for matrix A,
jr Al=
determine the values of a1, a2,...
Step 3: Determine the transform: ix T that transforms the system state equa-
tion into the controllable canonical form. (If he given system equation is already in the
"Fast + + yas + ay
Chapter 12. / Design of Control Systems in State Space controllable canonical form, then T
in the controllable canonical form, All we need here is to find the matri
formation matrix T is given by Equation (12-4), or
T-MW
where M is given by Equation (12-5) and W is given by Equation (12-6).
Step 4: Using the desired eigenvalues (desired closed-loop poles), write the desired
characteristic polynomial
.) Ibis not necessary to write the state equation
T. The trans-
+ ays" +
(5 — mls — pa)» + (8 = te) = Faas + ay
and determine the values of a, 425...
‘Step 5: The required state feedback gain matrix K can be determined from Equation
(12-13), rewritten thus:
fp dy | Oy dey | + | a | ay as)T!
Comments. Note that if the system is of low order (n = 3), then direct substitu-
tion of matrix K into the desired characteristic polynomial may he simpler. For exam-
ple, ifm = 3, then write the state feedback gain matrix K as
Kalk b&b)
Substitute this K matrix into the desired characteri
equate it to (s = ju)(s— 4)(5 — 4s), oF
[st = A + BK] = (6 — wee ~ pats - 4s)
Since both sides of this characteristic equation are polynomials ins, by equating the co-
efficients of the like powers of « on both sides, it is possible to determine the values of
ej, ko, and ks. This approach is convenient if n = 2 or 3, (For n= 4,5, 6,-.. this ap-
proach may become very tedious.)
There are other approaches for the determination of the state teedback gain matrix
K. In what follows, we shall present a well-known formula, known as Ackermann’s for-
mula, for the determination of the state feedback gain matrix K.
polynomial [s1- A + BK| and
Ackermann’s formula. Consider the system given by Equation (12-1), rewritten
thus:
i= Ax + Bu
We assume that the system is completely state controllable. We also assume that the de-
sited closed-loop poles are at $= 11,8 = fl... 48 = flm
Use of the state feedback control
w= Kx
modifies the system equation to
(A BK)x (12-14)
Let uy define
A=A-BK
“The desired characteristic equation is
Section 12-2 / Pole Placement 793 794
= A BK] = [ot = Al= (= is pd 0 a)
= stay bs tags tay = 0
Since the Cayley-Hamilton theorem states that A satisfies its own characteristic equa-
tion, we have
a
ofA) +al=0 (12-15)
We shall utilize Equation (12-15) to derive Ackermann’s formula. To simplify the de-
rivation, we consider the case where n = 3. (For any other positive integer m. the fol-
lowing derivation can be easily extended.)
Consider the following identities:
1=1
A=A- BK
B® =(A- BK? ~ ABK - BKA
A’ = (A ~ BK)’ = A’ — A°BK ~ ABKA — BKA’
Multiplying the preceding equations in order by a, a2, a1, ao (were ap = 1), respec~
tively, and adding the results, we obtain
aul + aA + aA? + AP
= ag + a(A — BK) + @,(A? — ABK ~ BKA) + A - A°BK
— ABKA - BKA?
pl + 2A + aA? + A®— aBK ~ a ABK ~ a:BKA — A°BK
—ABKA — BKA? (12-16)
Referring to Equation (12-15), we have
asl + A+ ay + AP = 4A)
Also, we have
axl + aA + aA? + A= (A) #0
Substituting the last two equations into Equation (12-16), we have
(A) = o(A) — BK - «BKA — BKA? — a, ABK - ABKA ~ A°BK
Since (A) = 0), we obtain
(A) = BogK + «KA 4 KA2) + AB(K + KA) + ORK
aK + aKA + KA?
ak + KA 217)
K
Since the system is completely state controllable, the
matrix
werse of the controllability
[B | AB! A’B]
Chapter 12. J. Design of Control Systems in State Space EXAMPLE 12-1
exits. Premultiplying the inverse of the controllability matrix to both sides of Equation
(12-17), we obtain
agK + @KA + KA?
[B | AB { A’B]'¢(A) = @K+KA
K
Premultiplying hoth sides of this last equation by [0 0.1], we obtain
aK + KA + KM?
fo 0 10: ABE ABE AY=[0 0 1] K+ KA =K
K
which can be rewritten as
K=[0 0 1fB } AB: A°BI'6(A)
This last equation gives the required state feedback gain matrix K.
For an arbitrary positive integer n, we have
K=[0 0 ++ 0 TMB: ABE +++ i Am BY TOA) (12-18)
Equation (12-18) is known as Ackcrmann’s formula for the determination of the state
feedback gain matrix K.
‘Consider the system defined by
where
By using the state feedback control u = ~Kx, it is desired to have the closed-loop poles at
$= 22 Hands = ~ 10. Determine the state feedback gain matrix K.
Fimst, we need ta check the contrallahility matrix af the system. Since the controllability ma-
trix M is given by
oo 4
PAB; A%B)=]0 1 -6
1-6 31
we find that IMI = ~ 1 and therefore rank M ~ [Link], the system is completely state control-
lable and arbitrary pole placement is possible
Next, we shall solve this problem. We shall demonstrate each ofthe three methods presented
in this chapter.
‘Method 1-The first method is to use Equation (12-13), The characteristic equation for the sys-
temis
s -1 0
Ju-aj=|o os =
15 546
Section 12-2 / Pole Placement 795 796
= S46 +541
= Stas + asta =0
Hence,
‘The desired characteristic equation is
(s+ 2 — jas +24 AANGs + 10) = 9? + 145" + 60s + 200
8+ ays? + ays + a =0
Hence,
a= 1h a= 60, a= 200
Referring to Equation (12-13), we have
K = [a3 ~ a; | a3 — a | a a)T!
where T =I for this problem because the given state equation is in the controllable canonical
form, Then we have
K= (200-1 | 60-5} 14-6)
=[199 55 8)
Method 2: By defining the desired state feedback gain matrix K as
K-[h kk)
and equating isl ~ A + BK| with the desired characteristic equation, we obtain
s 0 o] fo 1 | fo
Ja-a+ex|=|fo 5 of-] 0 0 f+ loft & All
oo sf [- -s -6} |
s “1
=] 0 s
+k Sth s464hy
=H 6+ kale? + (5 + kals +14 hy
= + Las? + 605 + 200
Thus
6+; Stk 60, 14h = 200
from which we obtain
99, k= 55, ky
or
K=[19 55 8]
‘Method 3: The third method is to use Ackermann’s formula. Referring to Equation (12-18), we
have
K-(0 0 1][B } AB | AB} g(A)
since
WIA) = AB + 14A2 + 6OA + 2001
‘Chapter 12 / Design of Control Systems in State Space -1 -5 -6| “1-5 6)
on 1 0 0
+6} 0 0 1/+200/0 1 0
13 oo 1
19 35 8
=|-8 159 7
-7 43117
and
0 0 1
o 1-6
1-6 9
‘we obtain
0 0 i} f199 ss gi
K=f0 0 nlo 1 6} |-8 19 7
1-6 st) [-7 -4 117]
5 6 1]fte ss 8
=f0 0 ale 1 o|/-8 19 7
10 of|-7 43 117
=[199 55 8}
As a matter of course, the feedback gain matrix K obtained by the three methods are the same.
this state feedback, the closed-loop poles are placed als = ~ 2 jd ands = ~ 10, a8 desired,
kis noted that if the order m of the system were 4 or higher, methods 1 and 3 are recom-
‘mended, since all matrix computations can be carried out by a computer. If method 2s used, hand
‘computations become necessary because a computer may not handle the characteristic equation
with unknown parameters hikes «+e
Comments. It is important to note that matrix K is not unique for a given system,
bbut depends on the desired closed-loop pole locations (which determine the speed and
damping of the response) selected. Note that the selection of the desired closed-loop poles
or the desired characteristic equation is a compromise between the rapidity of the re-
sponse of the error vector and the sensitivity to disturbances and measurement noises.
‘That iif we increase the speed of error response, then the adverse effects of disturbances
and measurement noises generally increase. Ifthe system is of second order, then the sys-
tem dynamics (response characteristics) can be precisely correlated to the location of the
desired closed-loop poles and the zer0(s) of the plant, For higher-order systems, the loca-
tion of the closed-loop poles and the system dynamics (response characteristics) are not
easily correlated. Hence, in determining the state feedback gain matrix K for a given sys-
tem, itis desirable to examine by computer simulations the response characteristics of the
system for several different matrices K (based on several different desired characteristic
equations) and to choose the one that gives the best overall system performance.
‘Section 12-2 / Pole Placement 797 12-3 SOLVING POLE-PLACEMENT PROBLEMS WITH MATLAB
798
Pole-placement problems can be solved easily with MATLAB. Let us solve the same
pole-placement problem as discussed in Examples 12-1. The system equation is
%= Ax+ Bu
where
oon 0
A=|0 0 1, B
-1 -5 -6 1
By using state feedback control u ~ ~ Kx, it is desired to have the closed-loop poles at
5 =i (i= 1,2,3), where
m= 24 i p= 2-7 py = -10
Let us find the necessary state feedback gain matrix K.
If we use the transformation matrix T in designing state feedback control matrix K,
‘we need to find the coefficients a1, q2, and ax of the characteristic equation Isl ~ Ai = 0.
This can be done by entering the statement
P = poly(A)
into the computer. Then the coefficients will appear on the screen as follows:
1.0000 6.0000 5.0000 1.0000
‘Then ay = al = PQ), a2 = a2 = P(3),and ay = a3 = Pid),
‘To get the transformation matrix T, we first enter matrices M and W into the com-
puter, where
M-=[B AB A’B]
and
a a 1
we-lq 1 0
1 0 0
‘Then multiply M and W, which can be done quite easily with MATLAB.
T- Mw
d characteristic equation. This can be done by defining ma-
Next, obtain the d
trix J such that
Chapter 12. / Design of Control Systems in State Space wm 0 «(0] [-24+4 0 0
g=|0 m ol-| 0 -2 4 0
0 0 ws 0 0 =10
and then using the poly()) command, as follows:
yei-2+ ari
Q= polyd
a=
0 OD 2-441 O30 0
114 60. 200
Hence, we have
ay = aal = Q(2), a = 202 = Q(3), ay = 223 = Q(4)
(We use at for x.)
‘The required state feedback gain matrix K can be determined from
K=[as~ a3 a2 - a2 ay — aT!
or
K=[aa3- a3 aa2- a2 al — at ]Minw)
‘A MATLAB program for solving this example problem using the transformation
matrix T is given as MATLAB Program 12-1
MATLAB Program 12-1
he
Pole placement
9% **+9" Determination of state feedback gain matrix K by use of
2% transformation matrix T
% "8+" Enter matrices A and B °°"
1 00 0
B= ([Link]
5-6)
.% **** Define the controllability matrix M *****
M=IB AB AN2¢B);
1% 177°" Check the rank of matrix M "7"
rankiM)
Section 12-3 / Solving Pole-Piacement Problems with MATLAB 799 ‘Chapter 12 /
3
9% ***4* Since the rank of M is 3, arbitrary pole placement is
% possible **+**
% ***** Obsain the coefficients ofthe characteristic polynomial
‘hist — Al. This can be done by entering statement poly(A) *****
JA = polyiA)
A=
1.0000 6.0000 5.0000 1.0000
a1 = IAQ); a2 = JAG): a3 = JAG);
% ***** Define matrices W and Tas follows *****
We [a2 al al 1 01 0 OF
T= Mw
% ***** Obtain the desired characteristic polynomial by defining,
°% the following matrix | and entering statement poly) *****
J=[-2+j4 0 0,0 -2-j4 60 0 ~10;
I= poly)
ye
1 14 60 200
al = J)2); aa2 = Yk3¥; aa3 = Jd};
4% ***** State feedback gain matrix K can be given by *#***
(aa3—a3 aa2—az aut at}-(inwn)
ke
199 55 8
% **** Hence, k1, k2, and k3 are given by ***
kt = K(1), 2 = Ki2), 13 = KO)
k=
199,
Design of Controt Systems in State Space k=
55
a=
a
Ackermann's formula. If we use Ackermann's formula for the determination of
the state feedback gain matrix, we need to compute a matrix characteristic polynomial
(A), where for this system
GA) = AP + aA? + aA + and
In MATLAB, polyvalm computes the matrix polynomial (A). For the given J, poly)
computes the coefficients of the characteristic polynomial, as shown earlier. For
0 1 0
A=|0 0.1
-1 -5 -6
the command polyvalm(poly(),A) computes the following 6A)
199 55 8B
(A) = A? + 144? + 60A + 2001=] -8 159 7
7-43 117
In fact,
polyvaim(polyi),A)
199 558
“8 159 7
a7 -a3 7
MATLAB Program 12-2 produces the state feedback gain matrix K by use of Acker-
‘man’s formula.
MATLAB Program 12-2
%
——- Pole placement
[% #4" Determination of
* Ackermann’s formula *
we feedback gain matrix K by use of
% #4" Enter matrices A and B*****
A=10 1 00 0 1-1 -5 -6);
Section 12-3. / Solving Pole-Placement Problems with MATLAB sor Chapter 12 /
B= (0:0;11;
% ****" Define the controllability matrix M *****
M= 1B ASB AAD*B}
94 #244" Check the rank of matiix M “=
rankiM)
ans =
3
04 e#¥4 Since the rank of Mis 3, arbitrary pole placement is
°% possible ===
‘420+ Obraln the desired characteristic polynomial by defining
% the following matrix ] and entering statement poly) *
[-24j*4 0 00 -2-"4 00 0-10};
polyth
1 14 60 200
‘% ***** Compute the characteristic polynomial
% Phi = polyvalmipolyi),A)
Phi = polywalmipolyt,Ai
1-07 State feedback gan mattixK can be given by “==
K= [0 0 NinvN4yrPhi
k=
199 55 8
‘ye rere Hence, kl, K2, and k3 are given by =2++
kt = KD), k2 = KQ), kB = KG)
Design of Controt Systems in State Space 12-4 DESIGN OF REGULATOR-TYPE SYSTEMS BY POLE
PLACEMENT
22
Inverted-pendulum
system,
Consider the inverted-pendulum system shown in Figure 12-2, where an inverted pen-
dulum is mounted on a motor-driven cart. Here we consider only the two-dimensional
problem where the pendulum moves only in the plane of the paper.
Itis desired to keep the pendulum upright in the presence of disturbances (such as
‘a gust of wind acting on the mass m, an unexpected force applied to the cart). The
slanted pendulum can be brought back to the vertical position when appropriate con-
trol force wis applied to the cart. At the end of each control process it is desired to bring
the cart back to x = 0, the reference position.
Design a control system such that, given any initial conditions (caused by distur-
bbances), the pendulum can be brought back to the vertical position and also the cart can
be brought back to the reference position (x = 0) quickly (for example, with settling
time of about 2 sec) with reasonable damping (for example, the dominant closed-loop
poles have {= 0.5). Assume the following numerical values for M, m, and
M=2kg, o m=O1kg, 1=05m
It is assumed that the pendulum mass is concentrated at the top of the rod and the rod
is assumed to be massless.
It is desired to design a control system that would maintain the inverted pendulum
in a vertical position for given initial conditions on angle @ and/or an angular velocity 6.
Section 12-4 / Design of Regulator-Type Systems by Pole Placement 803 804
In addition, the control system is required to return the cart to its reference position
at the end of each control process. The system is to successfully respond to any distur-
bances in the form of initial conditions. (The desired angle 0, is always zero and the de-
sired cart position is always at the reference position. Thus, this system is a regulator
system.)
Here the controller will be designed using the state feedback control method by the
pole-placement technique, As noted earlier, the necessary and sufficient condition for
arbitrary pole placement is that the system be completely state controllable.
‘The first step in the design is to derive a mathematical model for the inverted-
pendulum system,
Mathematical modeling. Referring to Section 3-6, we derived a mathematical
‘model for the inverted-pendulum system shown in Figure 3-16(a). When angle is as-
sumed small, the equations describing the dynamics of the system are given by Equa-
tions (3-55) and (3-36), rewritten next
(M + mit + mid
+ mP?)6 + ml = malo
where Is the moment of inertia of the pendulum rod about its center of gravity. Since
in this system the mass is concentrated at the top of the rod, the center of gravity is the
center of the pendulum ball. In this analysis, we assume that the moment of inertia of
the pendulum about its center of gravity is zero, or / = 0. Then the mathematical model
becomes as follows:
(M+ mie + mld= uv (12-19)
PO + mle = melo (12-20)
Equations (12-19) and (12-20) define a mathematical model of the inverted-pendulum
system shown in Figure 12-2. (These linearized equations are valid as [Link] @issmall.)
Equations (12-19) and (12-20) can be modified to
Mb
(M+ migo ~ (12-21)
Mi =u — mg (12-22)
Equation (12-21) was obtained by eliminating ¢ from Equations (12-19) and
(12-20). Equation (12-22) was obtained by eliminating @ from Equations (12-19) and
(12-20). From Equation (12-21) we obtain the plant transfer function to be
a
MiP — (Mt + mg
By substituting the given numerical values and noting that g = 9.81 mvsec?, we have
as) 1 1
=U(s) ? ~ 20.601 ~ s* ~ (4.5397
‘The inverted-pendulum plant has one pole on the negative real axis (s = ~ 4.539) and
another on the positive real axis (¥ = 4.539), Hence, the plant is open-loop unstable.
‘Chapter 12 / Design of Control Systems n State Space Define state variables x1, 22,23, and a by
nae
nad
a
a
Note that angle 6 indicates the rotation of the pendulum rod about point P, and x is the
location of the cart. We consider @ and.x as the outputs of the system, or
[EE
(Notice that both @ and x are easily measurable quantities.) Then, from the definition
Of the state variables and Equations (12-21) and (12-22), we obtain
y
=X
M+m
as
In terms of vector-matrix equations, we have
% 0 1 0 O]fu) fo
1
is 0 of fm) |- i)
+ u (12-23)
is 0 0 0 1] |x} | o
ul [-B~e 0 0 of ful | t
m]oft 9 0 olla:
(*] ~ [0 o.1 ‘| % (am)
Me
Equations (12-23) and (12-24) give a state-space representation of the inverted pen-
