You are on page 1of 18

System Types in Feedback Control with

Saturating Actuators
Yongsoon Eun, Pierre T. Kabamba, and Semyon M. Meerkov
Department of Electrical Engineering and Computer Science,
University of Michigan, Ann Arbor, MI 48109-2140, USA
May 16, 2003

Abstract
This Technical Note extends the classical notion of system type to feedback
control with saturating actuators. For step, ramp, and parabolic inputs, it defines
the so-called trackable domains and evaluates steady state errors. It shows that,
unlike the linear case, the role of the poles at the origin of the plant and the
controller are different and, on this basis, extends the notion of system type.
Results obtained are useful for selecting controllers and actuators to ensure desired
trackable domains and steady state errors.

This work has been supported by NSF Grant No. CMS-0073302


Please address correspondence to Professor P. T. Kabamba, Department of Aerospace Engineering, University of Michigan, Ann Arbor, MI 48109-2140, USA, e-mail: kabamba@engin.umich.edu, phone: (734) 763-6728,
fax: (734) 763-0578.

1.

INTRODUCTION

The notion of system type is important in linear control systems analysis and design. This
notion, however, is not applicable to systems with saturating actuators. Indeed, the two
systems shown in Figure 1.1 have identical type if the saturation is ignored, but in the presence
of saturation have qualitatively different tracking capabilities. Namely, the system of Figure
1.1(a) can track ramps with a finite steady state error, but that of Figure 1.1(b) cannot track
any ramp. This indicates that the usual definition of system type has to be modified to be
applicable to systems with saturating actuators. This modification is the purpose of this
Technical Note.
Specifically, we show that the roles of poles at the origin of the plant and the controller are
different, and system type is defined by the plant poles. The controller poles, however, also
play a role but it is limited to affecting the steady state error, while not enlarging the class of
trackable references. This class is characterized through the new notion of Trackable Domain,
which quantifies the size of steps or the slope of ramps, that can be tracked when saturation
is present. The new system types and the Trackable Domains are the main contributions of
this work.
Systems with saturating actuators have been studied in numerous publications (see [1] for
an annotated bibliography and the recent monographs [2][4]). However, no results on system
types and Trackable Domains have been reported.
The outline of this note is as follows: Section 2 investigates the existence of steady states
and evaluates the steady state tracking errors. Based on this investigation, in Section 3,
the definition of system type is introduced and the steady state errors are characterized. In
Section 4, a design example is given. Conclusions are given in Section 5. The proofs are
included in the Appendix.
saturating
controller actuator

r +

1
s(s + 1)

saturating
controller actuator

plant

r +

1
s

1
s+1
(b)

(a)

Figure 1.1: Motivating example

plant

2.

EXISTENCE OF STEADY STATES, TRACKABLE DOMAINS,


AND STEADY STATE ERRORS

2.1

Assumption

Consider the SISO systems shown in Figure 2.1 where P (s) and C(s) are the plant and
controller, respectively, and sat (v) is the saturating actuator defined by

if < v,

sat (v) = v
if v ,

if v < .

(2.1)

In state space form, this system can be represented as:

x p = Ap xp + Bp sat (v),
x c = Ac xc + Bc (r y),

(2.2)

y = Cp xp ,
v = Cc xc + Dc (r y),
where xp and xc are the states of P (s) and C(s), respectively.
Results derived in this note are obtained under the assumption that the closed loop system
(2.1), (2.2) has a sufficiently strong stability property characterized below:
Assumption 2.1. The closed loop system (2.1), (2.2), with r = 0, is globally asymptotically
stable and this fact can be established by means of a Lyapunov function of the form
Z v
V (x) = xT Qx +
sat ( )d > 0, Q 0,

(2.3)

where x = [xTp xTc ]T .


In some cases, this assumption can be verified using the methods of absolute stability
theory [5]. In others, the method of [6] can be used. Although (2.3) seems somewhat exacting,

+ e
-

C(s)

sat (v)

P (s)

Figure 2.1: Feedback control system with saturating actuator


2

we need it to ensure the stability of (2.2) with not only the symmetric saturation (2.1) but
also with every asymmetric saturation defined by sat, (v) = sat (v + ) , with || < .
Systems with asymmetric nonlinearities arise in the analysis of (2.2) when r(t) 6= 0.
2.2

Step Input

Consider the closed loop system (2.1), (2.2) and assume that r(t) = r0 u(t), where u(t) is the
unit step function. For s R, define

C0 = lim C(s),

(2.4)

P0 = lim P (s),

(2.5)

s0+

s0+

where the limits (2.4) and (2.5) are allowed to be infinite. Introduce the steady state error
with respect to step
estep
ss = lim [r0 u(t) y(t)].
t

(2.6)

Theorem 2.1. Under Assumption 2.1,


(i) estep
exists if
ss

(ii) if r0 satisfies (2.7), then

|r0 | <
+ P0 ;
C0
estep
ss =

r0
.
1 + C0 P0

(2.7)

(2.8)

Remark 2.1. Since Assumption 2.1 precludes pole-zero cancellation between P (s) and C(s)
1
at s = 0, the terms
and P0 in (2.7) cannot be simultaneously infinite.
C0
The range of r0 , defined in (2.7) is called the Trackable Domain (TD). As indicated in
(2.8), within TD, estep
is identical to that of systems with linear actuators.
ss
Expression (2.7) shows that the set TD becomes larger as the d.c. gain of the plant increases
and becomes smaller as the d.c. gain of the controller increases. Thus, if P (s) has a pole at
the origin, TD is the whole real line R and estep
is zero. Otherwise, TD is finite, even if, due
ss
is zero.
to a pole at the origin of C(s), estep
ss
Outside the Trackable Domain, i.e., if |r0 | |1/C0 + P0 |, estep
may or may not exist. It
ss
does not exist if the closed loop system (2.1), (2.2) has no stable equilibrium. For instance,

consider the system of Figure 2.1 with parameters


C(s) = 1,

P (s) =

s+1
,
s2 + 1

= 1.

(2.9)

If (2.9) is realized as
x 1 = x2 ,
x 2 = x1 + sat (r0 y),

(2.10)

y = x1 + x2 ,

1
p

Assumption 2.1 holds with Q = I. However, if r0


+ P0 + 2x2 (0)2 + 2(1 x1 (0))2 ,
C0

the solution of (2.10) is given by

x1 (t) = 1 (1 x1 (0))cos(t) + x2 (0)sin(t),

(2.11)

x2 (t) = (1 x1 (0))sin(t) + x2 (0)cos(t),


and estep
does not exist.
ss
There seem to be no general methods for characterizing when estep
exists outside the
ss
Trackable Domain. However, if it does exist, then
estep
ss = r0 sign(r0 C0 )P0 .

(2.12)

In this case, the behavior of estep


as a function of r0 is illustrated in Figure 2.2.
ss
2.3

Ramp and Parabolic Inputs

The situation here is similar to that described above. Indeed, consider (2.1), (2.2) and assume
that r(t) is either r1 tu(t) or 21 r2 t2 u(t), where u(t), as before, is the unit step function. For
s R, introduce
P1 =
P2 =

lim sP (s),

(2.13)

lim s2 P (s),

(2.14)

s0+
s0+

and denote
eramp
=
ss
epar
ss

lim [ r1 tu(t) y(t) ],


i
h r
2 2
t u(t) y(t) .
= lim
t
2
t

(2.15)
(2.16)

estep
ss


+ P0
C0

slope =1

r0
slope =

1
1 + C0 P0
Trackable domain

+
P
0
C0

Figure 2.2: estep


as a function of the magnitude of the step input r0
ss
Theorem 2.2. Under Assumption 2.1,
(i) eramp
and epar
ss
ss exist if
|r1 | < |P1 |,

(2.17)

|r2 | < |P2 |,

(2.18)

respectively;
(ii) if r1 and r2 satisfy (2.17) and (2.18) respectively, then
eramp
=
ss
epar
=
ss

r1
,
C0 P1
r2
.
C0 P2

(2.19)
(2.20)

The ranges of r1 and r2 , defined by (2.17) and (2.18), are again referred to as Trackable
Domains. Thus, in the case of ramp and parabolic inputs, TD is independent of the controller
and is empty if the plant has no pole at the origin, irrespective of the number of such poles
in the controller.
Remark 2.2. If C(s) has a pole at the origin, the system of Figure 2.1 is sometimes augmented
by an anti-windup compensator. The purpose of anti-windup is to improve transient response
and stability while maintaining small signal (|v(t)| < ) behavior identical to that of the
5

closed loop system without anti-windup. Thus, if v(t) in the system with anti-windup does
not saturate at steady state, its asymptotic tracking properties remain identical to that of
the system without anti-windup. The condition on r0 in Theorem 2.1 (r1 , r2 in Theorem
2.2, respectively) is sufficient to ensure | limt v(t)| < . Therefore, Theorems 2.1 and 2.2
remain valid in the presense of anti-windup as well.

3.

SYSTEM TYPES

Theorems 2.1 and 2.2 show that the roles of C(s) and P (s) in systems with saturating actuators are different. Therefore, system type cannot be defined in terms of the loop transfer
function C(s)P (s), as in the linear case.
To motivate a classification appropriate for the saturating case, consider the system of
Figure 1.1(b). With P (s) and C(s) indicated, system (2.1), (2.2) satisfies Assumption 2.1.
According to Theorem 2.1, this system tracks steps from T D = {r0 R : |r0 | < }, with zero
steady state error. However, this system does not track any ramp input with finite steady
state error. Thus, this system is similar to a type 1 linear system but not exactly: it does not
track ramps at all. Therefore, an intermediate system type is necessary. This observation
motivates the following definition.
Definition 3.1. The system of Figure 2.1 is of type kS , where S stands for saturating, if the
plant has k poles at the origin. It is of type kS+ if, in addition, the controller has one or more
poles at the origin.
In terms of these system types, the steady state errors and trackable domains in systems
with saturating actuators can be characterized as shown in Table 3.1. As it follows from
Definition 3.1, the system of Figure 1.1(a) is of type 1S , whereas that of Figure 1.1(b) is of
type 0+
S . This, together with Table 3.1, explains their tracking capabilities alluded to in the
introduction.
Remark 3.1. Note that the definition of system type kS is a proper extension of types
for linear systems. Indeed, when the level of saturation, , tends to infinity, all Trackable
Domains become the real line R and type kS systems become type k.
Remark 3.2. Theorems 2.1 and 2.2 require Assumption 2.1, i.e., the existence of Q. In
some cases, this may be difficult to verify. However, if existence and global stability of an
6

Table 3.1: Steady state errors and Trackable Domain for various system types.
r(t)=r0 u(t)

r(t)= 21 r2 t2 u(t)

r(t)=r1 tu(t)

estep
ss

TD

eramp
ss

TD

epar
ss

r0
1+P0 C0

Not applicable

Not applicable

{r0 R:|r0 |<|P0 |}

Not applicable

Not applicable

1S

{r1 R:|r1 |<|P1 |}

r1
P1 C0

Not applicable

1+
S

{r1 R:|r1 |<|P1 |}

Not applicable

2S

{r2 R:|r2 |<|P2 |}

r2
P2 C0

Type

0S

TD

{r0 R:|r0 |< C1 +P0 }


0

0+
S

equilibrium have been ascertained for given r0 (respectively, r1 and r2 ), then the claims in
Theorem 2.1 (respectively Theorem 2.2) still hold.
Remark 3.3. In Table 3.1, only types up to 2S are included since systems of higher types can
only be stabilized locally by linear controllers with actuator saturation [7]. Trackable Domains
and steady state errors for systems of type higher than 2S can also be characterized if the
assumption on global stability of the closed loop system (2.1) and (2.2) is replaced by local
stability and initial conditions are restricted to the domain of attraction. This, however, would
necessitate estimation of the domain of attraction, which is generally difficult. Moreover, the
Trackable Domain will depend on this estimate. Therefore we limit our analysis to globally
asymptotically stable cases.

4.

EXAMPLE

In this section, we give an illustrative example of the controller design process, using the new
system types. Specifically, for a given plant with saturating actuator, we select a controller
structure to satisfy given steady state performance specifications and then adjust the controller
parameters to improve transient response.
Consider a DC motor modeled by
P (s) =

1
,
s(Js + b)

(4.1)

where J = 0.06 kg m2 and b = 0.01 kg m2 /sec. Assume that the maximum attainable torque
of the motor is 2.5 N m, i.e.,
= 2.5.
7

(4.2)

Using this motor, the problem is to design a servo system satisfying the following steady state
performance specifications:
(i) steps of magnitude |r0 | < 100 rad should be tracked with zero steady state error;
(ii) ramps of slope |r1 | < 20 rad/sec should be tracked with steady state error less than or
equal to 0.01r1 .
A solution to this problem is as follows: According to Table 3.1, specification (i) requires
a system of type at least 0+
S . Since the plant has a pole at the origin, the system is of type at
least 1S , which guarantees estep
ss = 0 and T D = R. Specification (ii) requires
|P1 | 20,

(4.3)

and
eramp
=
ss

r1
0.01r1 .
P1 C0

(4.4)

Inequality (4.3) is satisfied, since P1 = 100 and = 2.5. Inequality (4.4) is met if C0 1.
Thus, a controller of the form
C(s) = C0

n
Y
(1 s/zi )
i=1

(4.5)

(1 s/pi )

guarantees (i) and (ii), if C0 1 and the zi s and the pi s are selected so that Assumption 2.1
is satisfied (i.e., the closed loop system is globally asymptotically stable).
Remark 4.1. Note that, if a Trackable Domain for the ramp were specified by |r1 | < M
rad/sec with M > |P1 | = 250 rad/sec, no linear controller, satisfying this specification,
would exist.
Although the closed loop system (4.1), (4.2), and (4.5) does satisfy the steady state specifications, it is of interest to analyze its transient behavior as well. To accomplish this, consider
the simplest controller of the form (4.5):
C(s) = 1.

(4.6)

To check if Assumption 2.1 is satisfied, consider the following realizations of the plant (4.1)
and the controller (4.6)

0
1
,
Ap =
0 b/J

0
,
Bp =
1/J

Ac = 0, Bc = 0, Cc = 0, Dc = 1.
8

h
i
C0 = 1 0 ,

(4.7)

Using these realizations and the method of [6], it is possible to show that Assumption 2.1 is
satisfied with

0
0
.
Q=
0 0.03

(4.8)

The transients of (4.1), (4.2), (4.6) shown in Figure 4.1, are obviously deficient: both the
overshoot and the settling time are too large. A similar situation takes place for ramp inputs
as well: transients are too long. To improve this situation, consider the controller (4.5) with
C0 = 1, z = 0.15, and p = 50, i.e.,
C(s) =

1 + s/0.15
,
1 + s/50

(4.9)

and its realization


Ac = 50, Bc = 50(50/0.15 1), Cc = 1, Dc = 50/0.15.

(4.10)

For this case, Assumption 2.1 is satisfied with

0.25 1.5 0

Q = 1.5 10 0 .

0
0 0

(4.11)

The transients of (4.1), (4.2), (4.9), illustrated in Figure 4.2, show an improvement in overshoot but still long settling time. This, perhaps, is due to the fact that, as it follows from
150

rad

100
50
0

10

10

20
30
40
(a) output y(t) (solid) and reference r(t) (dashed)

50

60
sec

Nm

2
0
2
20

30
40
(b) control input sat2.5(v(t))

50

Figure 4.1: Transient response with controller C(s) = 1


9

60
sec

150

rad

100
50
0

10

10

20
30
40
(a) output y(t) (solid) and reference r(t) (dashed)

50

60
sec

Nm

2
0
2
20

30
40
(b) control input sat2.5(v(t))

Figure 4.2: Transient response with controller C(s) =

50

60
sec

1+s/0.15
1+s/50

150

rad

100
50
0

10

10

20
30
40
(a) output y(t) (solid) and reference r(t) (dashed)

50

60
sec

Nm

2
0
2
20

30
40
(b) control input sat2.5(v(t))

Figure 4.3: Transient response with controller C(s) =

50

60
sec

1+s/1.5
1+s/50

Figure 4.2 (b), the control effort is underutilized. To correct this situation, consider the
controller (4.5) with C0 = 1, z = 1.5, and p = 50, i.e.,
C(s) =

1 + s/1.5
.
1 + s/50

(4.12)

With this controller, the transients are shown in Figure 4.3. As one can see, the settling
time has improved considerably and is very close to its lower limit (about 2.1 sec), which is
10

imposed by the limitation of the control effort and the magnitude of the step. Also, Figure
4.3 (b) shows that the control effort is fully utilized and behaves almost like the time-optimal
bang-bang controller. Similar behavior is observed for all magnitudes of steps up to 100 rad
and for ramps with slopes less than 20 rad/sec. We remark that, although the matrix Q
for this controller had not been found, the global asymptotic stability of the system (4.1),
(4.2), (4.12) has been shown using the method of [8] and the associated software posted at
http://www.cds.caltech.edu/jmg/personal.html.

5.

CONCLUSIONS

This technical note investigates steady state tracking errors with respect to step, ramp and
parabolic inputs in systems with saturating actuators. It shows that the roles of poles at the
origin of the plant and the controller are different. Based on this, system types are introduced
and the steady state errors and Trackable Domains are characterized in terms of these system
types. As illustrated by an example, the results derived may be useful for selecting controllers
and/or actuators that ensure desired trackable domains and steady state errors in feedback
systems with saturating actuators.

APPENDIX
The proofs of Theorem 2.1 and 2.2 are based on the following lemmas.
Lemma A.1. Define

Ap
A=
Bc Cp


h
i
B
0
, B = p , C = Dc Cp Cc ,
0
Ac

(A.1)

and
1
1
M = (A + BC)T (Q + C T C) + (Q + C T C)(A + BC),
2
2

(A.2)

where Ap , Bp , Cp , Ac , Bc , Cc , Dc are defined in (2.2) and Q is defined in (2.3). Then, under


Assumption 2.1,
(a) A + BC is Hurwitz,
(b) Q + 21 C T C > 0,
(c) M 0,
(d) (A + BC, M ) is observable.
11

Proof: Using (A.1), rewrite (2.2) with r = 0 as


x = Ax + B sat (v),

(A.3)

v = Cx.
Under Assumption 2.1, global asymptotic stability of closed loop system (2.1), (2.2), or equivalently the closed loop system (2.1), (A.3), can be established using the Lyapunov function
Z v
T
sat ( )d.
(A.4)
V (x) = x Qx +
0

If |Cx| , this V (x) reduces to


1
V (x) = xT (Q + C T C)x,
2

(A.5)

the derivative of V (x) along the trajectory of (2.1), (A.3) becomes


V (x) = xT M x,

(A.6)

x = (A + BC)x.

(A.7)

and (A.3) can be written as

Since the closed loop system (2.1), (A.3) is globally asymptotically stable, A + BC in (A.7)
is Hurwitz, which proves (a). Since V (x) > 0 and V (x) 0, matrices Q + 12 C T C and M (see
(A.5) and (A.6)) are positive definite, and negative semidefinite, respectively. This proves (b)
and (c). Finally, (a), (b), (c) imply that the pair (A + BC, M ) is observable [9], which proves
(d).
Lemma A.2. Let Assumption 2.1 hold. Then, for every asymmetric saturation, sat, (v) =
sat (v + ) , with || < , the closed loop system (2.2) with r = 0, i.e.,
x p = Ap xp + Bp sat, (v),
x c = Ac xc + Bc (y),

(A.8)

y = Cp xp ,
v = Cc xc + Dc (y),
is globally asymptotically stable.
Proof: Under Assumption 2.1, global asymptotic stability of closed loop system (2.1), (2.2)
can be established using the Lyapunov function
Z
T
V (x) = x Qx +

12

sat ( )d,

(A.9)

where x = [xTp xTc ]T . It will be shown that, based on this V (x), a Lyapunov function can be
found that establishes global asymptotic stability of (A.8).
Without loss of generality, assume 0 < < . Introducing the state transformation

=
x and the notation =
v, rewrite (A.8) as
+
+
= A + B(),

(A.10)

= C.

Here A, B, and C are defined in (A.1) and () =


sat,
+

,
if < ,

() = ,
if < < ,

, if < ,

+
, i.e.,

(A.11)

( )
< .
+
For the closed loop system (A.10), (A.11), select a Lyapunov function candidate as follows:
Z
T
( )d,
(A.12)
V1 () = Q +

where =

where Q is the same as in (A.4). The derivative of V1 () along the trajectories of (A.10),
(A.11) is
V 1 () = T (AT Q + QA) + (C)(2B T Q + CA) + CB[(C)]2 ,

(A.13)

while the derivative of V (x) along the trajectories of (2.1), (2.2) is


V (x) = xT (AT Q + QA)x + sat (Cx)(2B T Q + CA)x + CB[sat (Cx)]2 .

(A.14)

Now we show by contradiction that V 1 () 0. Assume there exist 6= 0 such that


V 1 ( ) > 0. This must satisfy (C ) 6= sat (C ), i.e., C > , otherwise, it would result

in V (x) > 0 at x = . Define x = . Then, Cx > , (C ) = and sat (Cx ) = .

Substituting x in (A.14) yields


V (x ) = x T (AT Q + QA)x + sat (Cx )(2B T Q + CA)x + CB[sat (Cx )]2
= (/)2 T (AT Q + QA) + (2B T Q + CA) / + CB2
= (/)2 [ T (AT Q + QA) + (2B T Q + CA) + CB 2 ]
= (/)2 [ T (AT Q + QA) + (C )(2B T Q + CA) + CB[(C )]2 ]
= (/)2 V 1 ( ) > 0,
13

(A.15)

which contradicts Assumption 2.1. Therefore, V 1 () 0 for all 6= 0.


Next we show, again by contradiction, that the only solution of (A.10), (A.11) that is
contained in { |V 1 () = 0} is the trivial solution (t) 0.

Let (t),
t 0, be a non-trivial solution of (A.10), (A.11) that satisfies V 1 ((t))
0.
for all t 0. Since sat (C (t))

Assume first C (t)


= (C (t)),
system (2.1), (2.2) and
is a non-trivial solution of (2.1),
system (A.10), (A.11) are identical; therefore, x
(t) = (t)

(2.2) as well. Moreover, as it follows from (A.13) and (A.14), V (


x(t)) = V 1 ((t))
0. This
cannot satisfy C (t)
for all t 0.
contradicts Assumption 2.1. Hence, (t)
for all t 0. Then, from (A.10), (A.11),
Assume now that C (t)
Z t
At

(t) = e (0) +
eA Bd.

(A.16)

Define x
(t) = (t).
Then, x
(t) is a solution of (2.1), (2.2), since C x
(t) > and

= eAt x
(0) +
x
(t) = (t)

eA Bd.

(A.17)

Moreover, using the chain of equalities similar to (A.15), we obtain V (


x(t)) = (/)2 V 1 ((t))

cannot satisfy C (t)


for all t 0
0. This again contradicts Assumption 2.1. Hence, (t)

either. Therefore, there must exist an interval (t1 , t2 ), t1 6= t2 , such that |C (t)|
< for all
t (t1 , t2 ). In this interval,
= e(A+BC)(tt1 ) (t
1 ),
(t)

t (t1 , t2 ),

(A.18)

and

= 0,
V 1 ((t))
= T (t)M (t)

t (t1 , t2 ),

(A.19)

where M is defined in (A.2). Since M is negative semidefinite by Lemma A.1, it follows that
= M e(A+BC)(tt1 ) (t
1 ) = 0,
M (t)

t (t1 , t2 ).

(A.20)

This, however, contradicts the observability of (A + BC, M ), which must take place according
0 is the only solution of (A.10), (A.11) that is contained in
to Lemma A.1. Thus, (t)
{ |V 1 () = 0}.
Finally, it can be shown, again by contradiction, that V1 () > 0 for all 6= 0 and that
V1 () as || . Therefore, the system (A.10) and (A.11), and, hence, (A.8) is globally
asymptotically stable.

14

Proof of Theorem 2.1: Three cases are possible: (a) |C0 | =


6 and |P0 | =
6 ; (b) |C0 | =
and |P0 | =
6 ; (c) |P0 | = .
(a) Assume |C0 | =
6 and |P0 | =
6 . Define xp , xc , y , and v as
v =

C0
r0 ,
1 + C0 P0

y =

C0 P0
r0 ,
1 + C0 P0

xp = A1
p Bp v ,

xc = A1
c Bc (r0 y ). (A.21)

Using
x
p = xp xp ,

x
c = xc xc ,

y = y y ,

v = v v ,

(A.22)

rewrite (2.2) as
x
p = Ap x
p + Bp sat,

v ),
v (

x
c = Ac x
c + Bc (
y ),

(A.23)

y = Cp x
p ,
v = Cc x
c + Dc (
y ).
Note that condition (2.7) implies |v | < . Therefore, Assumption 2.1 and Lemma A.2 ensure
the global asymptotic stablility of (A.23). Hence, the steady state exists and y(t) converges
to y , i.e., estep
exists. This proves (i). Moreover, by definition, estep
is given by
ss
ss

estep
ss = r0 y = r0

C0 P0
r0
r0 =
,
1 + C0 P0
1 + C0 P0

(A.24)

which proves (ii).


(b) |C0 | = implies that Ac has an eigenvalue 0. Choose xc 6= 0 to satisfy
Ac xc = 0,

Cc xc =

r0
,
P0

(A.25)

and let
v =

r0
,
P0

y = r0 ,

xp = A1
p Bp v .

(A.26)

Proceeding similarly to (a), it can be shown that if |r0 | < |P0 |, then estep
= 0.
ss
(c) |P0 | = implies Ap has an eigenvalue 0. Choose xp 6= 0 to satisfy
Ap xp = 0,

Cp xp = r0 ,

(A.27)

and let v = 0, y = r0 , xc = 0. Similarly to (a), it can be shown that estep


= 0. This
ss
completes the proof.

15

r1 us (t)

1
s

+ e(t)
-

C(s)

v(t)

sat (v)

P (s)

y(t)

(a)
r1 us (t) +

e(t)

1
C(s)
s

v(t)

sat (v)

sP (s)

y(t)

(b)
Figure 5.4: Block diagram of a system with ramp input
Proof of Theorem 2.2: Consider the system with ramp input shown in Figure 5.4(a).
Equivalently, it can be represented as shown in Figure 5.4(b) where the input and output are
step and y rather than ramp and y as in Figure 5.4(a). We refer to these systems as system
(a) and (b), respectively. Since the input to system (b) is step, eramp
of system (a) can be
ss
analyzed using Theorem 2.1 applied to system (b). Indeed, Assumption 2.1 holds for system
(b) since it is assumed to hold for system (a). Also, introducing
1
b
C(s)
= C(s),
s

Pb(s) = sP (s),

(A.28)

and noting that


Pb0 = P1 ,

b0 = ,
C

(A.29)

1
+ Pb0 for system (b). Thus,
condition (2.17) for system (a) is equivalent to |r1 | <
b0
C
Theorem 2.1 ensures the existence of steady state in system (b) and
lim e(t)
=

r1
= 0.
b0
1 + Pb0 C

(A.30)

This implies that limt e(t) = const., which proves (i).

It follows from the proof of Theorem 2.1 that | limt v(t)| < . Therefore, eramp
must
ss
be identical to that of the system with linear actuator, i.e.,
eramp
=
ss

r1
.
limt sP (s)C(s)

(A.31)

Since P1 = limt sP (s) 6= 0, it follows that


lim sP (s)C(s) = lim sP (s) lim C(s) = P1 C0 ,

16

(A.32)

and
eramp
=
ss

r1
.
P1 C0

(A.33)

This proves (ii).


The proof for the parabolic input case is similar.

REFERENCES
[1] D. S. Bernstein, and A. N. Michel, Chronological bibliography on saturating actuators,
Int. J. Robust and Nonlinear Contr., vol. 5, no. 5, pp. 375-380, 1995.
[2] A. Saberi, A. A. Stoorvogel, and P. Sannuti, Control of Linear Systems with Regulation
and Input Constraints, London: Springer, 2000.
[3] T. Hu, Z. Lin, Control Systems with Actuator Saturation : Analysis and Design, Boston:
Birkhaser, 2001.
[4] V. Kapila (editor), Actuator Saturation Control, New York: Marcel Dekker, 2002.
[5] K. S. Narendra and J. H. Taylor, Frequency Domain Criteria for Absolute Stability , New
York: Academic Press, 1973.
[6] F. Tyan and D. S. Bernstein, Global stabilization of systems containing a double integrator using a saturated linear controller, Int. J. Robust and Nonlinear Contr., vol. 9,
pp. 11431156, 1999.
[7] H. J. Sussmann and Y. Yang, On the stabilizability of multiple integrators by means of
bounded feedback controls, Proc. Conf. Dec. Contr., pp. 7072, Brighton, U.K., 1991.
[8] J. Goncalves, Quadratic surface Lyapunov functions in global stability analysis of saturation systems, Proc. Amer. Contr. Conf., Arlington, VA, June 2001.
[9] K. Zhou, J. C. Doyle, and K. Glover, Robust and Optimal Control, New Jersey: Prentice
Hall, 1996.

17