Академический Документы
Профессиональный Документы
Культура Документы
com
Automatica 40 (2004) 905915
www.elsevier.com/locate/automatica
Fundamental properties of reset control systems
Orhan Beker
a
, C.V. Hollot
b,
, Y. Chait
c
, H. Han
b
a
Maxtor Corporation, Shrewsbury, MA, USA
b
ECE Department, University of Massachusetts Amherst, Amherst, MA 01003, USA
c
MIE Department, University of Massachusetts Amherst, Amhrest, MA 01003, USA
Received 21 January 2002; received in revised form 11 June 2003; accepted 4 January 2004
Abstract
Reset controllers are linear controllers that reset some of their states to zero when their input is zero. We are interested in their feedback
connection with linear plants, and in this paper we establish fundamental closed-loop properties including stability and asymptotic tracking.
This paper considers more general reset structures than previously considered, allowing for higher-order controllers and partial-state
resetting. It gives a testable necessary and sucient condition for quadratic stability and links it to both uniform bounded-input bounded-state
stability and steady-state performance. Unlike previous related research, which includes the study of impulsive dierential equations, our
stability results require no assumptions on the evolution of reset times.
? 2003 Elsevier Ltd. All rights reserved.
Keywords: Reset actions; Stability analysis; Nonlinear control systems
1. Introduction
In this paper we study the control system depicted in
Fig. 1 which consists of a reset controller R connected in
feedback with a plant transfer function P(s).
1
A reset con-
troller is a linear time-invariant system whose states, or sub-
set of states, reset to zero when the controller input e is zero.
Motivation for reset control comes from two sources. First,
from the limitations of linear feedback control systems im-
posed by Bodes gain-phase relationship. Second, from the
favorable sinusoidal describing function of reset controllers
which promise relief from Bodes constraint. Indeed, a reset
integrator, also referred to as a Clegg integrator (CI), has a
describing function similar to the frequency response of a
linear integrator but with only 38.1
j
x
j
]
. Examples
of reset controllers include the CI: A
r
= 0; B
r
= 1; C
r
= 1;
A
p
= 0 and the FORE:
A
r
=b, B
r
= 1, C
r
= 1, A
p
= 0. (2)
In both of these cases, n
p
= n
r
= 1. We assume that plant
P(s) in Fig. 1 accounts for any linear pre-compensation.
In fact, the design of reset control systems as developed in
Zheng et al. (2000) and Horowitz and Rosenbaum (1975)
involves the synthesis of both linear compensator C(s) and
reset controller R. Typically, the linear compensator is used
to stabilize and shape the loop to satisfy classical Bode spec-
ications at high and low frequencies. The reset controller
is then designed to meet overshoot constraints. We assume
P(s) strictly proper and adopt the realization:
x
(t) = A
(t) + B
u(t),
,
(t) = C
(t),
where A
R
n
, B
R
n
1
and C
R
1n
. The
closed-loop reset control system can then be described by
the IDE
x(t) = A
c
x(t) + B
c
w(t), x(t) M(t),
x(t
+
) = A
R
x(t), x(t) M(t),
908 O. Beker et al. / Automatica 40 (2004) 905915
,(t) = C
c
x(t) + d(t),
e(t) = w(t) C
c
x(t), (3)
where x = [x
r
]
,
A
c
,
_
A
C
r
B
r
C
A
r
_
, B
c
,
_
0
B
r
_
,
A
R
,
_
I 0
0 A
p
_
, C
c
,
_
C
+n
r
: e(t) = 0, (I A
R
) = 0}. (4)
If x(t
) M(t
), then x(t
a
reset time. From (4) they satisfy: x(t
) M(t
) x(t
+
)
M(t
+
+n
r
: C
c
= 0, (I A
R
) = 0}. Our
rst theorem gives a general Lyapunov-like stability con-
dition similar to those in Bainov and Simeonov (1989),
Lakshmikantham, Bainov, and Simeonov (1989) and Ye
et al. (1998). However, while these consider more general
IDE, they also make assumption on the reset times. For an
IDE, the reset times may form an increasing sequence with
a nite limit: lim
i
t
i
=
)=1
t
)
. This phenomenon
considerably complicates analysis, and the IDE literature
makes assumption so as to avoid this behavior. A prevalent
approach is to assume a lower bound on the reset intervals;
i.e., for some o 0, t
i
o, i N. This assumption guar-
antees continuation of solutions over R
+
. In Ye et al. (1998)
and Lakshmikantham et al. (1989), the authors explicitly as-
sume such continuation. In contrast, we do not require any
explicit assumptions on the reset times. Satisfaction of the
Lyapunov condition is enough.
Theorem 1. Let J(x) : R
n
R be a continuously-diere-
ntiable, positive-denite, radially unbounded function such
that
J(x) ,
_
cJ
cx
_
A
c
x 0, x = 0, (6)
J(x) ,J(A
R
x) J(x) 60, x M. (7)
Then,
(i) there exists a left-continuous function x(t) satisfying
(5) for all t 0,
(ii) the equilibriumpoint x=0 is globally uniformly asymp-
totically stable.
Proof. We begin with a few denitions. Let O be a neigh-
borhood of the equilibrium point x = 0. We rst dene a
ball of radius r as B
r
, {x O : x 6r} and its bound-
ary
B
r
, {x O : x = r}. Similarly, we dene [-level
set of O as O
[
, {x O : J(x) 6[} and corresponding
boundary
O
[
,{x O : J(x) =[}. We now proceed with
the proof.
(i) Without loss of generality we consider an initial con-
dition x
0
M. Since (5) initially behaves as a linear sys-
tem, a solution exists over some time interval. To prove
its continuation over all of R
+
, we avoid trivialities and
assume an innite sequence of reset times {t
k
}
k=1
with
O. Beker et al. / Automatica 40 (2004) 905915 909
accumulation point 1; i.e., t
k
1. We now show that
x(t
k
) 0 as k . In doing so, we will have proved con-
tinuation of solution since x(t) = 0 continues to be a solu-
tion of (5) for all t 1. From (6) and (7), J(t
k+1
) J(t
k
)
for all k so that x(t
k
) tends to
O
c
as k for some
c 0. Thus, the base-linear system associated with (5) has
bounded velocity x(t) on (t
k
, t
k+1
), and, since t
k
0, x(t
k
)
must approach the xed points of the reset map A
R
; that is,
(I A
R
)x(t
k
) 0. From this and (4) we see that x(t
k
) tends
to the boundary of M. But, on this boundary, the dynamic is
dictated by the base-linear system so that x(t
k
) approaches
the base linears equilibrium set. Eq. (6) insures this set to
be trivial so that x(t
k
) 0.
(ii) We will now show that the equilibrium point
x = 0 is stable. Given an c 0, let : = min
x
B
c
J(x) 0
and take x
0
O
:
. Inequalities (6) and (7) imply that
J(x(t)) 6J(x
0
) 6:. Hence, x(t) O
:
x(t) B
c
. There
exists a o(c) such that : = max
x
B
o(c)
J(x) and conse-
quently x(t) B
o(c)
x(t) O
:
. Therefore, x
0
B
o(c)
x
0
O
:
x(t) O
:
x(t) B
c
. This proves that x = 0 is
a uniformly stable equilibrium point.
Now, given o, and any initial condition x
0
B
o
, by (6)
and (7), J(x(t)) decreases when x Mand non increasing
when x M. Since J(x) is bounded from below by zero,
there exists a c 0 such that
J(x(t)) c, t . (8)
To show c = 0 we proceed by contradiction and suppose
c 0. By continuity of J(x) there exists a d0 such that
B
d
O
c
. (8) implies that the trajectory x(t) lies outside
the ball B
d
for all t. Let = sup
d6x6o
J(x). Then,
0 by (6) and Js continuous dierentiability, for t
i
t
6t
i+1
J(x(t)) 6J(x
0
) (t t
0
). (9)
The right-hand side of (9) eventually becomes negative,
thereby contradicting (8). Therefore, J(x(t)) 0 as t
. Since J is radially unbounded, o can be chosen arbi-
trarily large. Therefore, the equilibrium x = 0 is globally
uniformly asymptotically stable.
Remark 2. (1) Typically, Lyapunov-like conditions for
IDEs are weaker than (6), requiring
J 0 just on the
smaller set x M. Condition (6) requires decrease over
all of the state space, and this insures two things. First,
the stability of the base-linear dynamic, and second, that
the base-linear and reset control systems share the same
Lyapunov function. Technically, this stronger condition
helps render the non-trivial xed points of the reset map A
R
unattractive.
(2) Theorem 1 does not rule out accumulating reset times.
As argued in the proof of item (i), if reset times do accu-
mulate at time 1, then lim
t1
x(t) = 0.
In the next subsection we specialize to quadratic Lya-
punov function J(x) = x
0
n
p
P
p
](sI A
c
)
1
_
_
0
0
n
p
I
n
p
_
_
(10)
is strictly positive real
3
where I
n
p
denotes an identity matrix
of size n
p
n
p
and 0
n
p
denotes a matrix of zeros of size
n
p
n
p
.
We now specialize to quadratic Lyapunov functions.
Denition 4. The reset control system (5) is said to be
quadratically stable if there exists a positive-denite sym-
metric matrix P such that J(x) = x
Px satises conditions
(6) and (7).
We now state our quadratic stability result.
Theorem 5. The reset control system (5) is quadratically
stable if and only if it satises the H
[
-condition.
Proof. (Suciency) We show that (5) is quadratically
stable. To this end, consider the quadratic Lyapunov candi-
date J(x) = x
(A
c
P + PA
c
)x 0; x = 0 (11)
and
x
(A
R
PA
R
P)x 60; x M. (12)
Dene
M={ R
n
+n
r
: C
c
= 0} and let O be a matrix
whose columns span
M. Since M
M, (12) is implied
by
O
(A
R
PA
R
P)O60. (13)
3
A square transfer function matrix X(s) is called strictly positive real
if: (i) X(s c)s elements are analytic in Re[s] 0 and (ii) X
(s c)
+ X(s c) 0 for Re[s] 0 for some c 0 (Khalil, 1996). Here
denotes the complex conjugate transpose.
910 O. Beker et al. / Automatica 40 (2004) 905915
Using the structure of A
R
and C
c
in (3), a straightfor-
ward computation shows that inequality (13) holds for some
positive-denite symmetric matrix P if there exists a [ R
n
p
and positive denite P
p
R
n
p
n
p
such that
_
0 0
n
p
I
n
p
P =
_
[C
0
n
p
P
p
. (14)
Thus, to show quadratic stability, it suces to nd a
positive-denite symmetric matrix P such that (11) and
(14) hold. From the KalmanYakubovichPopov (KYP)
Lemma; e.g., see Khalil (1996), such P exists if H
[
(s) in
(10) is SPR for some [.
(Necessity) Suppose (5) is quadratically stable. Then,
there exists a positive-denite symmetric matrix P such that
(11) and (12) hold. The continuity of J, together with
(12), implies that x
(A
R
PA
R
P)x 60 whenever x
M
which in turn implies that (14) holds for some [ R
n
p
and
positive denite P
p
R
n
p
n
p
. The strict positive realness of
H
[
(s) in (10) then follows from (11), (14) and the KYP
Lemma.
The H
[
condition can be checked using linear matrix
inequalities. When the reset controller has a single state
(n
p
=1, n
p
=0), one can take P
p
=1 without loss of general-
ity and the H
[
condition reduces to a single-variable search
for an SPR function as illustrated with a simple example.
Example 6. Consider the unforced reset control system
(5) with P(s) = 1}s and R
b
(s) = 1}(s + 1). To check the
H
[
-condition we use (10) and form the stable transfer func-
tion H
[
(s) = (s + [)}(s
2
+ s + 1). Given any [ (0, 1), it
is easy to show that the real part of H
[
(jc) is positive for
all c0. Hence, the system is quadratically stable from
Theorem 5. A quadratic Lyapunov function verifying (6)
and (7) is
J(x) = x
_
1.5 0.5
0.5 1
_
x.
Remark 7. In this remark we address the prevalence of
quadratically stable reset control systems.
(1) Some stable reset control systems are not quadratically
stable. For example, in Hu, Zheng, Chait, and Hollot (1999),
it was shown that a Clegg integrator exponentially stabilizes
the plant P(s)=((3+:)s+1)}(s
2
+3s:) for : (6.1, 1.1).
However, computation shows the H
[
-condition is only sat-
ised for the smaller range : (3, 0). Thus, there is a dif-
ference between the classes of reset control systems that
are stable and quadratically stable. It is interesting to note
that the quadratically stable systems in this example exactly
coincide to those P(s)s that are both minimum phase and
stable.
(2) In spite of the previous example, it does appear that the
class of quadratically stable systems is rich as exemplied
in the next two examples: (a) Consider those reset control
systems whose base-linear transfer functions have the classic
second-order form: 1(s) = (c
2
n
)}(s
2
+ 2c
n
s + c
2
n
) where
, c
n
0. Also, assume the reset controller is a FORE (with
pole s =b). In Section 5, we will show this class of reset
control systems to be quadratically stable for all b 0. This
is encouraging given the ubiquity of control systems hav-
ing this type of complementary sensitivity functions which
arise when integral action is required and one loop-shapes a
stable, minimum-phase plant. This class also covers the ex-
ample in Beker et al. (2001a) which demonstrates that reset
control satises specications unachievable by any linear
stabilizing compensator. (b) The experimental demonstra-
tions of reset control in Zheng et al. (2000) and Chen (2000)
were veried to be quadratically stable, and, their associ-
ated loop transfer functions were non-trivial. For example,
in Zheng et al. (2000), the transfer function was 14th-order
and had a pair of complex right-half plane zeros.
(3) The class of quadratically stable reset control systems
require their base-linear systems to be stablethis follows
immediately from the H
[
-condition. This begs the question:
Do there exist stable reset control systems with unstable
base-linear system? Said another way, can mere application
of reset stabilize a linear, unstable feedback loop? The an-
swer is yes and the example comes from Hu et al. (1999)
where P(s)=(3.1s+1}s
2
+0.3322s0.1). This plant cannot
be stabilized by a linear integrator, but it can be stabilized by
a CI. This stability is not deducible from the H
[
-condition
but from the techniques in Hu et al. (1999).
Remark 8. As mentioned, Horowitz and his co-workers
(see Krishnan and Horowitz (1974) and Horowitz and
Rosenbaum (1975)) incorporated FOREs into control sys-
tem design by advocating a two-step process in which a lin-
ear controller was rst designed followed by selection of the
FOREs pole. In Horowitz and Rosenbaum (1975), specic
guidelines were provided which explicitly link the design of
the FORE to the linear compensation. However, considera-
tions of closed-loop stability were not addressed. The ques-
tion, then, is whether quadratic stability can be incorporated
into this design scheme. Assuming FORE as the reset ele-
ment, one possibility comes from the following expression
for H
[
(s): H
[
(s)=(1}(s+b)[)S(s)+[. Design of the linear
compensator insures base-linear stability and consequently
the stability of H
[
(s). Therefore, for quadratic stability it
suces to guarantee Re{(1}()c+b) [)S()c)} [ for
all c0. For xed [ and b it seems possible to express
the above as a useful constraint on the linear loop shape.
This would allow one to bring quadratic stability directly
into the design process.
The H
[
-condition is useful in establishing other proper-
ties of reset control systems. In the next subsection we will
address one such property, UBIBS stability.
3.3. UBIBS stability
In Chen (2000) a bounded-input bounded-output (BIBO)
stability result was given for a special class of reset control
O. Beker et al. / Automatica 40 (2004) 905915 911
systems that utilize FOREs. It was assumed that the reset
intervals were lower-bounded. We generalize this result to
a larger class of reset control systems, extend it to UBIBS
stability and remove the assumption on reset times. We now
consider the forced reset control system described by the
IDE in (3) and recall the denition of UBIBS stability. In
the following denotes the usual Euclidean vector norm
and
,sup
t
x(t).
Denition 9. The reset control system (3) is said to be uni-
formly bounded-input bounded-state (UBIBS) stable, if,
for each p 0, there exists j 0 such that for each ini-
tial condition x
0
and each bounded input w(t), the solution
x(t, x
0
, w) continues over R
+
and
x
0
p, w
p x(t, x
0
, w) j
for all t 0.
Remark 10. Our denition of UBIBS is strengthened by
assuming solutions exist over R
+
. For the zero-input case,
recall that quadratic stability automatically insures this
occurs; see Theorem 1(i).
For UBIBS stability, we require the following assumption
on the reset controller.
Assumption 11. The state matrix A
r
in (1) satises
A
r
=
_
A
r
11
A
r
12
0 A
r
22
_
.
That is, x
p
does not explicitly depend on the non-resetting
states x
p
.
Remark 12. Assumption 11 holds if all states of R are re-
set. Examples include CI, FORE and their generalization
addressed in Zheng (1998).
We now state our result on UBIBS stability.
Theorem 13. If Assumption 11 holds and (5) is quadrat-
ically stable, then the reset control system (3) is UBIBS
stable.
Proof. Without loss of generality we assume an innite
number of resets. We begin by writing down the base-linear
dynamics
x
(t) =
_
A
C
r
B
r
C
A
r
_
x
(t) +
_
0
B
r
_
w(t),
,
(t) =
_
C
(t) ,[x
p
(t)x
r
(t)]
. Here x
p
(t) and x
r
(t) denote
the states of P(s) and R
b
(s), respectively. We now dene
:(t) ,x(t) x
+n
r
: w(t) C
c
x
(t) C
c
= 0,
(I A
R
)(x
(t)x
r
(t)]
so that
x(t) 6:
(t) +x
p
(t) +:
r
(t) +x
r
(t). Since the
H
[
-condition implies asymptotic stability of the base-linear
system, x
(t) +:
r
(t) is
bounded.
Since H
[
(s) in (10) is SPR, then by the KYP Lemma
there exists a symmetric, positive-denite matrix P, a real
vector q R
n
+n
r
and an c 0 such that
A
c
P + PA
c
=q
q cP,
_
0 0
n
p
I
n
p
P =
_
[C
0
n
p
I
n
p
.
Therefore, we can partition
P =
_
P
1
P
2
P
2
I
n
p
_
(16)
with P
1
0 and P
2
=
_
[C
(t)+:
r
(t) is bounded. Therefore, for t (t
i
, t
i+1
], we
compute
J(t) = :
(t)(qq
(t) :
p
(t) :
p
(t)]
(t) :
p
(t)
P
1
_
:
(t):
p
(t)
+2:
p
(t)[C
(t) + :
p
(t):
p
(t). (19)
Denoting the last n
p
states of x
as x
p
and evaluating (19)
at t = t
+
i
we get
J(t
+
i
) 6J(t
i
) + 2x
p
(t
i
)([C
(t
i
) +x
p
(t
i
)). (20)
We now show the second term on the right-hand side of
(20) is bounded by k(1 e
:t
i
) for some k, : 0. Since
the base-linear system is stable, it follows that its response
x
p
(t
i
) is bounded. The term[C
(t
i
) is also bounded.
912 O. Beker et al. / Automatica 40 (2004) 905915
Indeed, recall that w(t
i
) = C
x
p
(t
i
) = C
(t
i
) + C
x
p
(t
i
);
therefore, C
(t
i
) 6w(t
i
) +C
x
p
(t
i
). Since input
w(t) is bounded and x
p
(t) is the response of the stable linear
system (15), [C
(t
i
) is bounded. We now claim that
x
p
(t
i
) is bounded. To see this, assume not. Then, from (20),
J(t
+
i
) J(t
i
) as i . Combining this with (18), gives,
for large i,
J(t) e
c(tt
i
)
J(t
+
i
), t (t
i
, t
i+1
],
J(t
+
i
) J(t
i
).
This implies J(t) 0 as i ; hence, :(t
i
) 0
and x
p
(t
i
) x
p
(t
i
). This is a contradiction; hence,
x
p
(t
i
) is bounded. Finally, we show that x
p
(t
i
)s deriva-
tive is bounded. From Assumption 11 and (1) x
p
(t
i
) =
e
A
r
22
t
i
_
t
i
0
e
A
r
22
o
B
p
e( o + t
i1
) d o. Thus, dx
p
(t
i
)}dt
i
=
A
r
22
x
p
(t
i
). Therefore, dx
p
(t
i
)}dt
i
is bounded. Since
x
p
(t
+
i1
) = 0, the boundedness of x
p
(t
i
) and its derivative
implies the existence of constants K
1
and : 0 such that
x
p
(t
i
) K
1
(1 e
:t
i
) for all i.
Having established this bound on x
p
(t
i
), we now return
to (20) and write J(t
+
i
) 6J(t
i
) + (1 e
:t
i
)K for some
constant K. Together with (18) we obtain
J(t
i+1
) 6e
c(t
i+1
t
0
)
J(t
0
)
+K
i
)=1
e
c(t
i+1
t
)
)
(1 e
:t
)
).
Let s
i
,
i
)=1
e
c(t
i+1
t
)
)
(1 e
:(t
)
t
)1
)
). We now claim
the existence of a B0 such that s
i
B for all i. If so,
then J(t) is bounded which is enough to complete the proof.
Since e
c(t
)
t
)1
)
1, then, to prove the claim, we rst write
s
i
=
i
)=1
e
c(t
i+1
t
)
)
(1 e
c(t
)
t
)1
)
)B
1
()) where B
1
()) ,
(1e
:(t
)
t
)1
)
}1e
c(t
)
t
)1
)
). Now, its easy to show there
exists a
B0 such that B
1
())
B for all ). Thus,
s
i
B
i
)=1
e
c(t
i+1
t
)
)
(1 e
c(t
)
t
)1
)
)
B.
To nish, we take B =
B.
Example 14. As an illustration of Theorem 13, consider
the reset control system with P(s) = (s + 1)}(s(s + 0.2))
and R
b
(s) = 1}(s + 1). To establish its boundedness to
a step input r(t) we invoke Theorem 13. Since the reset
controller is a FORE, Assumption 11 holds. Using (10) we
form H
[
(s) = (s
2
+ (0.2 + [)s + [}(s + 1)(s
2
+ 0.2s + 1)).
Clearly, H
[
(s) is stable, and for [ = 0.25 the real part of
H
[
()c) is positive for all c0. Hence, the H
[
-condition
is satised and the step response is bounded.
Having established conditions for inputoutput we now
draw our attention to the steady-state performance of reset
control systems.
4. Steady-state performance
In this section we study the steady-state behavior of reset
control systems and establish an internal model principle
similar to that found in linear control systems.
4.1. Asymptotic tracking and disturbance rejection
Consider the setup in Fig. 1 where the reference signal
r is generated from the initial-condition response of linear
system M(s). We say that P(s) contains an internal model
of r if P(s) contains the persistent modes of M(s). Similar
to linear control systems, our next result shows that a reset
control system asymptotically tracks a reference signal r;
i.e., lim
t
e(t) = 0, provided P(s) contains an internal
model of r.
Theorem 15. Suppose P(s) contains an internal model of r,
n(t) d(t) 0 and the H
[
-condition is satised. Then, the
reset control system described in (3) achieves asymptotic
tracking of the reference input r.
Proof. Recall that P(s) is realized by {A
, B
, C
}. We
assume this realization contains an internal model of the
reference input; i.e., there exists an initial condition :
0
, such
that r(0) = C
:
0
and
:(t) = A
:(t), :(0) = :
0
,
r(t) = C
:(t). (21)
We dene the state x
(t) , x
(t) x
r
(t)
x(t) = A
c
x(t), x(t) M,
x(t
+
) = A
R
x(t), x(t) M,
,(t) = C
c
x(t) + r(t),
where M= { R
n
+n
r
: C
c
= 0, (A
R
I ) = 0}. The
asymptotic tracking problem is now expressed as an asymp-
totic stability problem for the unforced system. By Theorem
5, asymptotic stability of the unforced system is guaranteed
if H
[
-condition is satised. This concludes the proof.
Remark 16. (1) Theorem 15 holds when an internal model
is contained in P(s); it is not enough for it to appear in the
reset controller R
b
. In fact, asymptotic tracking may not
occur if the internal model is present only in R
b
. For ex-
ample, suppose r is constant, the reset controller is a CI
(R
b
=1}s) and P(s) is integrator-free. If the system asymp-
totically tracks while continually resetting, the CIs output
will not persist. But, a persistent plant input is required to
track a constantthus, the internal model must be contained
in P(s).
(2) Theorem 15 holds equally well for output disturbance
rejection. That is, if P(s) contains an internal model of d,
the plant output satises lim
t
,(t) = 0.
O. Beker et al. / Automatica 40 (2004) 905915 913
(3) The linear loops considered in Examples 6 and 14
both contain integrators. Thus, each of these reset control
systems enjoy asymptotic tracking of constant reference
signals.
(4) In a typical feedback control problem (e.g., servo
tracking), one is interested in tracking reference signals in
the presence of sensor noise. Given the internal model prin-
ciple of Theorem 15, a natural question is how tracking er-
ror e is aected by sensor noise n. We ask this with the
understanding that in general, reset elements do not enjoy
superposition. Nevertheless, as a corollary to Theorem 15
(see Beker (2001)), we have that the steady-state tracking
error depends only on the sensor noise, for P(s) with internal
model of the reference signal.
5. A class of second-order base-linear systems
In this section we follow-up on Remark 7.2a and
show there exists a rich class of reset control systems
that are quadratically stable. To begin, consider the feed-
back system in Fig. 1 where the reset controller is a
FORE (with pole b). Assume the linear loop has trans-
fer function P(s) = ((s + b)c
2
n
}s(s + 2c
n
)) resulting
in a base-linear system with complementary sensitiv-
ity function 1(s) = c
2
n
}(s
2
+ 2c
n
s + c
2
n
). This trans-
fer function has classical second-order form and is often
encountered in feedback control systems where inte-
gral control is used and response is second-order dom-
inant. This setup is described by the IDE in (5) with
data:
A
c
=
_
_
2c
n
1 1
0 0 b
c
2
n
0 b
_
_
, B
c
=
_
_
0
0
1
_
_
,
A
R
=
_
_
1 0 0
0 1 0
0 0 0
_
_
, C
c
=
_
c
2
n
0 0
. (22)
Our next result shows this class of reset control system to
be quadratically stable.
Theorem 17 (See Beker et al. (1999) for proof ). The re-
set control system described in (5) and (22) is quadrat-
ically stable for all positive b, c
n
and . Consequently,
this reset control system is UBIBS stable and enjoys the
asymptotic tracking and disturbance rejection properties
of Theorem 15.
Remark 18. (1) Examples 6 and 14, as well as the reset con-
trol system in Beker et al. (2001a), have the second-order
base-linear structures considered here. As a result, their
quadratic stability is ensured by Theorem 17.
(2) A useful extension to Theorem 17 would consider
P(s) that are second-order dominant; i.e., P(s) = ((s + b)
c
2
n
}s(s + 2c
n
)(s} + 1)) where 1}(s} + 1) models a
high-frequency parasitic. A relevant question is whether
quadratic stability (or equivalently, the H
[
condition) is ro-
bust in the presence of this dynamic. This question is yet
unanswered.
6. Conclusion
This paper shows that quadratic stability plays an impor-
tant role in reset control systems, similar to that in linear
feedback. That is, quadratically stable reset control systems
are inputoutput stable and have an internal model property
useful in asymptotic tracking and disturbance rejection. For
linear systems, quadratic stability is tested via a Lyapunov
equation. For reset control systems, it is deduced from a con-
strained Lyapunov equation, or equivalently, from an SPR
conditionthe H
[
-condition. All stable linear systems are
quadratically stable, but not so for reset control; see Re-
mark 7.1. Nevertheless, the H
[
-condition has been valuable
in establishing stability for some high-order experimental
systems and is always satised for the important class of re-
set control systems described in Section 5. A possible topic
for further research is to explore the use of non-quadratic
Lyapunov functions. One step in this direction has been
taken in Beker (2001) where passivity formalism has been
applied.
Another topic for future research is concerned with the
performance of reset control systems to sensor noise. While
the UBIBS stability result in Theorem 13 is applicable to
the sensor noise input, we would ultimately like to eval-
uate the system gain from sensor noise to plant output.
Some groundwork has been laid in Beker, Hollot, and Chait
(2000) and Beker et al. (2001b) where we have studied the
steady-state response of reset control systems to sinusoidal
sensor noise and in some simple examples have computed
the steady-state gain. These gains agree well with results ob-
tained using a sinusoidal describing function approximation
of the reset controller.
Finally, other areas of future research include robustness,
controller synthesis and performance limitations. As men-
tioned in Remark 18.2, robustness of Theorem 17s result
to hi-frequency parasitics is an open issue. Generalizing this
to a more general norm-bounded uncertain dynamic would
also be of interest. In Remark 8 we discussed the incor-
poration of the H
[
-condition into controller synthesis. This
appears to be a worthwhile objective allowing one to bring
quadratic stability directly into the design process. Lastly,
boundaries dening the potential benets and cost for using
reset control have yet to be drawn. On one hand we have
the example in Beker et al. (2001a) showing benet. On the
other, we have instances, such as in Hu et al. (1999), where
reset destabilizes a stable base-linear system. A formal study
of the performance limitations of reset control systems ap-
pears ripe and challenging.
914 O. Beker et al. / Automatica 40 (2004) 905915
Acknowledgements
This material was based upon work supported by the
National Science Foundation under Grant No. CMS-
9800612. We would also like to thank the reviewers for
their constructive comments which helped improve the
paper.
References
Bainov, D. D., & Simeonov, P. S. (1989). Systems with impulse eect:
stability, theory and application. New York: Halsted Press.
Beker, O. (2001). Analysis of reset control systems. Ph.D. thesis,
Department of Electrical and Computer Engineering, University of
Massachusetts, Amherst, MA.
Beker, O., Hollot, C. V., & Chait, Y. (2000). Forced oscillations
in reset control systems. In Proceedings of the 39th IEEE
conference on decision and control, Sydney, Australia December 2000
(pp. 48254826).
Beker, O., Hollot, C. V., & Chait, Y. (2001a). Plant with integrator: An
example of reset control overcoming limitations of linear feedback.
IEEE Transactions on Automatic Control, 46(11), 17971799.
Beker, O., Hollot, C. V., & Chait, Y. (2001b). Stability of limit-cycles in
reset control systems. In Proceedings of the 2001 American control
conference, Arlington, VA, June 2001 (pp. 46814682).
Beker, O., Hollot, C. V., Chen, Q., & Chait, Y. (1999). Stability of a reset
control system under constant inputs. In Proceedings of the American
control conference, San Diego, CA, June 1999 (pp. 30443045).
Bobrow, J. E., Jabbari, F., & Thai, K. (1995). An active truss element and
control law for vibration suppression. Smart Materials Structures, 4,
264269.
Branicky, M. S. (1998). Multiple Lyapunov functions and other analysis
tools for switched and hybrid systems. IEEE Transactions on
Automatic Control, 43, 475482.
Bupp, R. T., Bernstein, D. S., Chellaboina, V.-S., & Haddad, W. M.
(2000). Resetting virtual absorbers for vibration control. Journal of
Vibration and Control, 6, 6183.
Chen, Q. (2000). Reset control systems: Stability, performance and
application. Ph.D. thesis, Department of Mechanical and Industrial
Engineering, University of Massachusetts, Amherst, MA.
Chen, Q., Chait, Y., & Hollot, C. V. (2001). Analysis of reset control
systems consisting of a FORE and second-order loop. ASME Journal
of Dynamic Systems, Measurements, and Control, 123, 279283.
Chen, Q., Hollot, C. V., & Chait, Y. (2000). BIBO stability of reset control
systems. In Proceedings of the 39th IEEE conference on decision and
control, Sydney, Australia, December 2000 (pp. 251256).
Clegg, J. C. (1958). A nonlinear integrator for servomechanisms.
Transactions A.I.E.E., 77(Part II), 4142.
Decarlo, R. A. (1988). Variable structure control of nonlinear
multivariable systems: A tutorial. IEEE Proceedings, 76(3), 212232.
Feuer, A., Goodwin, G. C., & Salgado, M. (1997). Potential benets
of hybrid control for linear time invariant plants. In Proceedings
of the American control conference, Albuquerque, June 1997
(pp. 27902794).
Haddad, W. M., Chellaboina, V., & Kablar, N. A. (2000). Active
control of combustion instabilities via hybrid resetting controllers. In
Proceedings of the American control conference, Chicago, IL, June
2000 (pp. 23782382).
Hollot, C. V., Zheng, Y., & Chait, Y. (1997). Stability analysis for
control systems with reset integrators. In Proceedings of the 36th
IEEE conference on decision and control, San Diego, CA, December
1997 (pp. 17171719).
Horowitz, I., & Rosenbaum, P. (1975). Non-linear design for cost
of feedback reduction in systems with large parameter uncertainty.
International Journal of Control, 24(6), 9771001.
Hu, H., Zheng, Y., Chait, Y., & Hollot, C. V. (1997). On the zero-input
stability of control systems having Clegg integrators. In Proceedings
of the 1997 American control conference, Albuquerque, NM, June
1997 (pp. 408410).
Hu, H., Zheng, Y., Chait, Y., & Hollot, C. V. (1999). On the stability of
control systems having Clegg integrators. In D. E. Miller, & Li Qiu
(Eds.), Topics in control and its applicationsa tribute to Edward
J. Davison (pp. 107115). Berlin: Springer.
Khalil, H. K. (1996). Nonlinear systems. Englewood Clis, NJ:
Prentice-Hall.
Krishnan, K. R., & Horowitz, I. M. (1974). Synthesis of a non-linear
feedback system with signicant plant-ignorance for predescribed
system tolerances. International Journal of Control, 19(4),
689706.
Lakshmikantham, V., Bainov, D. D., & Simeonov, P. S. (1989).
Theory of impulsive dierential equations: Series in modern applied
mathematics, Vol. 6. Singapore: World Scientic.
Lau, K., & Middleton, R. (2000). On the use of switching control for
systems with bounded disturbances. In Proceedings of the 39th IEEE
conference on decision and control, Sydney, Australia, December 2000.
Singer, N. C., & Seering, W. P. (1990). Preshaping command inputs to
reduce system vibration. Transactions of the ASME, 112, 7682.
Tsypkin, Y. Z. (1984). Relay control systems. Cambridge: Cambridge
University Press.
Ye, H., Michel, A. N., & Hou, L. (1998). Stability analysis of systems with
impulse eects. IEEE Transactions on Automatic Control, 43(12),
17191723.
Zheng, Y. (1998). Theory and practical considerations in reset
control design. Ph.D. thesis, Mechanical and Industrial Engineering
Department, University of Massachusetts, Amherst, MA.
Zheng, Y., Chait, Y., Hollot, C. V., Steinbuch, M., & Norg, M.
(2000). Experimental demonstration of reset control design. Control
Engineering Practice, 8(2), 113120.
C.V. Hollot received the Ph.D. in elec-
trical engineering from the University of
Rochester in 1984. He then joined the De-
partment of Electrical and Computer Engi-
neering at the University of Massachusetts,
receiving the NSF PYI in 1988. His research
interests are in the theory and application of
feedback control.
Yossi Chait received his B.S. degree in
Mechanical Engineering from Ohio State
University in 1982, and his M.S. degree
and Ph.D. degree in Mechanical Engi-
neering from Michigan State University in
1984 and 1988, respectively. Currently he
is an associate professor at the Mechani-
cal Engineering Department, University of
Massachusetts, Amherst.
Dr. Chait has numerous publications in the
area of robust control design. He has been
active in Quantitative Feedback Theory
teaching and research for the past fteen years. In recognition of this work,
he was an Air Force Institute of Technology Distinguished Lecturer and
was a Dutch Network Visiting Scholar at the Laboratory for Measurement
and Control, Delft University, and Philips Research Laboratories, a visiting
appointment at Tel Aviv University, an Academic Guest, Measurement
and Control Laboratory, Swiss Federal Institute of Technology, ETH,
Zurich, Switzerland and a Lady Davis Fellow, Department of Mechanical
Engineering, the Technion, Haifa, Israel. Dr. Chait has consulted for
industry in a broad range of applications, for example in automatic
welding, real time particle analyzers and vibrations reduction. His recent
research focuses on congestion control of the Internet and modeling of
feedback mechanisms in biological systems such as the Hypothalamus
O. Beker et al. / Automatica 40 (2004) 905915 915
PituitaryThyroid axis and Circadian rhythms. For more details visit
http://www.ecs.umass.edu/mie/labs/dacs/.
Huaizhong Han was born in Guichi, China
in 1975. He received B.S. and M.S. in en-
gineering from University of Science and
Technology of China, Hefei, in 1998 and
2001, respectively. He is currently a Ph.D.
candidate at the University of Massachusetts
Amherst conducting research in the control
of communication networks.
Orhan Beker received his B.S. degree in
Electrical and Electronic Engineering from
Bogazici University, Istanbul, Turkey in
1996; M.S. and Ph.D. in Electrical and
Computer Engineering from University of
Massachusetts at Amherst in 1999 and
2001 respectively. Since 2001 he has been
a servo engineer at Maxtor Corporation
in Shrewsbury, MA. His research interests
include control theory and applications,
specically nonlinear control applications
in hard disk drives.