Lehman Bass Extensions Averaging To Power Electronic Systems IEEETPE 1996
Lehman Bass Extensions Averaging To Power Electronic Systems IEEETPE 1996
Lehman Bass Extensions Averaging To Power Electronic Systems IEEETPE 1996
Abstruct- This paper extends averaging theory for power stability results which rely on abstract averaging theory (see
electronic systems to include feedback Controlled converters. New references in [9]) that partially combine the results of [6] and
averaging techniques based on the integral equation description 171 with the theory of Filoppov [SI.
provide theoretical justification for commonly used averaging
methods. The new theory provides a basis for answering fun- It is the purpose of this paper to introduce averaging
damental questions about the averaging approximation. A ripple techniques that are general enough to encompass both time
estimate expression is presented, along with the simulation results discontinuity and large classes of state discontinuity, without
for a feedback controlled boost converter. utilizing the (difficult) theory of Filippov. Because the proofs
are straightforward (essentially relying on the Fundamental
I. INTRODUCTION Theorem of Calculus and Gronwall’s inequality), insight on
both transient and asymptotic behavior of PWM feedback
TATE space averaging techniques are commonly used in
controlled dc-dc converters is obtained. The results of this
the analysis and control design of pulse width modulated
paper begin to provide theoretical justification for commonly
(PWM) power electronic systems [I]-[3]. However, it was
used averaging techniques. In addition, this work points out
not until recently that rigorous mathematical justification [ 3 ] ,
some shortcomings in the averaging technique (which to
[4] was given that theoretically explained the applications of
our knowledge have not been documented before). Some
these averaging techniques. As [3] and [5] have pointed out,
readers may question whether there is a significant contribution
the theoretical development of PWM systems lags far behind
in writing a paper that theoretically justifies models that
the many practical control applications.
have been in use for so many years. However, we believe
In [ 3 ] ,classical Russian averaging techniques [6], [7] are
that it is vital to bridge theory with practice in order for
shown to be applicable to several types of PWM power elec-
future fundamental contributions to be made. In fact, the
tronic systems, such as open loop dc-dc converters. Besides
theoretical results of this paper have led to the discoveries of
using these classical averaging techniques to prove stability.
new, more accurate switching-frequency-dependent-averaged
[31 also gives a ripple estimate for improving the accuracy of
models [IO], published in a separate paper.
the averaging technique, even for systems with large ripple.
Section I1 reviews some of the mathematical issues asso-
However, the application of the results of [3] is limited to
ciated with state discontinuous systems. The primary theoret-
systems with time discontinuities.’
ical contribution of this paper is contained in two theorems
In fact, the classical averaging theory used in [3] is not
presented in Section 111. Section IV discusses the practical
applicable when there are state discontinuities. This is sig-
implications of the results of Section 111 and gives numer-
nificant because all feedback controlled converters are state
ical examples and computer simulations. Section V draws
discontinuous. In [ 3 ] , the argument is made that smooth
conclusions.
commutation models can be used in place of the discontinuous
Heaviside unit step function to avoid any state discontinuity
in the mathematical system model. In essence, this idea 11. THEORETICAL PRELIMINARIES
was introduced by Filoppov [SI to justify what is meant by The difficulty in mathematically justifying averaging
solutions to state discontinuous differential equations. The approximation techniques of state discontinuous differential
work of [9] continues this line of thinking by presenting equations can be best explained through an example. Consider
the state discontinuous differential equation
Manuscript received September 10, 1993; revised December 7, 1995.
B. Lehman was supported by the National Science Foundation under a
Presidential Faculty Fellowship, CMS-959268. +
k ( t ) = f(z) bu(d(z)- tri ( t :7’)) (2.1)
B. Lehman is with the Department of Electrical and Computer Engineering,
Northeastern University, Boston, MA 021 15 USA. where x E R”, b E R”,f : R” i R” and d : 72” i R are
R. M. Bass is with the School of Electrical and Computer Engineering, both continuous functions with 0 5 d(x) 5 1, and U ( . ) is the
Georgia Institute of Technology, Atlanta, GA 30332-0250 USA.
Publisher Item Identifier S 0885-8993(96)03562-4. Heaviside step function, i.e., u ( s ) = 1 for s 2 0 and u ( s ) = 0
‘In this paper, a system with “time discontinuity” is described by a for s < 0. The function tri ( t : T ) = ( t / T )- f l o o r ( t / T ) =
differential equation whose right-hand side is discontinuous with respect ( t mod) TIT is shown in Fig. 1. Equation (2.1) is a typical
to time. A system with “state discontinuity” is described by a differentia1
equation whose right-hand side is discontinuous with respect to a state representation of a feedback controlled PWM Buck converter
variable. [3]. The theory presented in [3], however, only applies to
0885-8993/96$05.00 0 1996 IEEE
LEHMAN AND BASS: EXTENSIONS OF AVERAGING THEORY FOR POWER ELECTRONlC SYSTEMS 543
E4 to)+ 1;
[f(z(s))
every z1 E R, x2 E R, there are constant positive k , satisfying
l/fi(.l) -f,(z2)11 5 k z l / z l-z21/.The functions d,: Rn -+R
+ bu(d(z(s)) - tri (s, T ) ) ds
] (2.2) are the duty ratios and will also be assumed locally Lipschitz in
R with Lipschitz constant m,. Furthermore, they will always
is unique and satisfies state differential equation (2.1) almost satisfy 0 5 d L ( z ) 5 1.
everywhere. Hence, when no chattering occurs in the system, Along with (3. I ), consider the corresponding "averaged"
the "standard" solution to (2.1) can be derived and will integral equation
be equal to the solution of integral equation (2.2) almost
everywhere. y(t; to, Y(t0)) =y(t)
where f t , and d, are as previously defined and y E R". This and a constant K >0 such that for 0 < T 5 TO
section will discuss the conditions under which solutions to
(3.2) can approximate solutions to (3.1). Since (3.2) is both
continuous and autonomous, its analysis is much simpler than
that of discontinuous and nonautonomous (3.1). For example,
if f L and d, have continuous partial derivatives with respect to
z,then the stability properties of (3.2) may be determined
by examining the eigenvalues of the linearization of (3.2) for any t E [ t o , L ] .
about each steady state. No such simple statement can be said Step 3: Immediately from Step 1, Step 2, and Gronwall's
about determining the stability of (3.1). The two theorems inequality, this implies that for t E [to, L ] , L > t o , and
presented in this section extend the results of [3] to the state 0<TsTo
discontinuous case, i.e., to the feedback control case.
ll4t) - Y ( t ) l l I(Ilz(t0) - y(to)ll + 6 ) e K ( L - t o )
A. Chattering where 6 i 0 as T 4 0. This implies that on any arbitrarily
By representing state discontinuous differential equations by large but bounded time interval, if z(t0) = y(to), then x ( t ) and
a corresponding integral equation, it is possible to rigorously y(t) can remain arbitrarily close to each other for a sufficiently
explain averaging approximations in power electronic systems. small switching period.
However, it will always be necessary to assume that the Step 4: Assume that z(to) = y ( t 0 ) and that y(t) ap-
models under consideration have a finite number of right- proaches a uniformly asymptotically stable equilibrium point,
hand side state discontinuities on any bounded time interval y3. Then, there will always exist a sufficiently small To =
and that each discontinuity is Lebesgue integrable. This, To(6)such that, for 0 < T 5 TO
however, is not always true for mathematical models of power
electronic systems. For example, when systems are switching lIz(t) - Y ( t ) / l < 6, t 2 t o .
infinitely often (chattering), there exists no compact time
Furthermore, this result will remain valid for initial conditions
interval in which the right-hand side of the state discontinuous
) y ( t o ) l l I p, where p > 0 is sufficiently
that satisfy l I ~ ( t 0 -
differential equation is continuous. Hence, a unique solution
to a corresponding integral equation will not exist in the usual small.
Step 4 basically states that if averaging can be proven
sense unless the theory of differential inclusions [SI is used.
on a finite time interval, then it can always be extended
In this paper, we will always assume that the system is
to an infinite time interval in the special case when the
not chattering. The physical implication of this assumption is
that power electronic switches turn on and off only once each averaged solution approaches a uniformly asymptotically sta-
PWM switching period. Conditions for guaranteeing this are ble equilibrium point. This statement has been proven by
presented in [13] and will not be discussed here. However, it many authors [3], [6], 171, [12] and is standard to averaging
is important to note that the averaging results presented below, theory.
are only valid when chattering does not occur. Once Step 2 is completed, Steps 3 and 4 will immediately
follow. However, it turns out that, for PWM systems, com-
B. Theoretical Results pleting Step 2 is extremely difficult and relies on some very
recently developed mathematical tools [ 111, [ 121. Keeping the
We begin this section by outlining the general averaging
above algorithm in mind, it is now possible to prove the
procedure that will be taken in this paper to justify the
main results of this paper. The proof of Theorem 3.1 relies
approximation of (3.1) by (3.2).
on several Lemmatta, which are presented in the Appendix.
Given a nonautonomous, integral equation [such as (2.1) or
(3.1)] z ( t ) = z(to) + Ji g ( s , ~ ( s )T ,)d s , consider the cor-
Theorem 3.1: Let z ( t )and y(C) denote the solutions to (3.1)
and (3.2), respectively. Then, for any constant L > t o and for
responding autonomous "averaged' integral equation y(t) =
any constant 7 > 0, there exists a TO= To(q, L) > 0 and a
y(t0) + JL g(y(s))ds, where g(.) is an "average value" of
constant K > 0 such that, for 0 < T 5 To,
y ( t , ., .) and J ( . ) does not depend on time, t , or on the
switching period, T .
ll4t) - Y ( t ) l l I (Ilz(t0) -y(to)ll+rl) exp{K(t-to)} (3.3)
Step 1: Take the difference between the two integral equa-
tions to obtain for all t E [to, L].
Proof of Theorem 3.1: For simplicity, define operators
3:R" --f R" and W : R" -+ R" as
Step 2: Show that for any 6 > 0, however small, and any
L > t o , however large, there will always exist a To = To(6,
L) . u ( d i ( z ( s ) )- tri (s, T ) )ds (3.4)
~
LEHMAN AND BASS: EXTENSIONS OF AVERAGING THEORY FOR POWER ELECTRONIC SYSTEMS 545
II(J-x)(t) - (J-lc)(t>ll I
So(N + l)(t - t o )
E; lo
N t
Under the assumption of no chattering, x ( t ) , the solution to
(3.1) will be continuous. Therefore, it is well known (Theo-
+M IIU(ha(&(S)) - tri(s, TI)
,=I
rems 24.4 and 24.5, [14]) that x ( t ) can be approximated by - u(d,(i.,(s)) - tri (s, T))llds. (3.9)
+
piecewise constant functions. Construct N 1 such piecewise
constant functions & ( t ) E R",z = 0, 1, . .. , N , such that Using Lemma AS. there will always exist a TO= To(a,, L )
for any t E [to, L], 0 5 [ d , ( z ( t ) )- d Z ( & ( t ) ) ] 5 S, for such that, for 0 < 'I' 5 TO
i = 1, 2, . . . , N , where 6, > 0 are a set of positive constants.
Furthermore, choose &(t) such that for any
Il(J-x)(t) - ( J 4 ( t ) I l 5 So(N + 1)(t - t o )
N
t E [to, L ] , I l f L ( x ( t ) )- f,(Za(t))lI I 6, also, for
i = 0, 1, 2, . . . , N . Since f i ( . ) and rip(.) are Lipschitz
+M +
[a, S,(t - t o ) ]
2=1
functions, such &(t)can always be constructed for arbitrary
6, > 0. Define ( J Z ) ( t ) as
=0 + Yl(6) (3.10)
[SO, . . . , 6 ~and
( J ? ) ( t=z(to)
) +
lot fo(&(s)) ds
zero as 6, 4 0.
] y1(S) is a positive constant that approaches
N
+ &,(t- t o )
2=1
Yz(6),
I
for any t E [to, L ] . Clearly yz(6) 40 as 6, + 0.
Consider now the inequality
u ( d ; ( & ( s ) ) - tri ( s , T))llds (3.8)
II(Jx)(t) - (Wx)(t)II 5 Il(Jz>(t) - ( W ( t ) l I
for any t E [ t o , L ] . However, &(.) and 6, have been cho- + II(JZ)(t) (wz)(t)lI
-
sen so that d,(Z.,(t)) 5 d , ( x ( t ) ) 5 d , ( & ( t ) ) +
S,, i = + II(W?)(t) - (Wx)(t)II (3.12)
I, 2, . . . , N for any t. Define N new piecewise constant
functions, h % ( Z - i ( t ) )where
, h,(&(t))= min(1, d,(Z,(t))+ which is true for all t. Using the above discussion and
S,}; i = 1, 2 , . . . , N . Note that d,(&(t)) 5 d , ( z ( t ) ) 5 Lemma A.4, there exists a TO= To(cr,0,L ) such that, for
~
546 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 11, NO. 4, JULY 1996
0 < T 5 To, Then, using the notation defined in (3.4)-(3.6) and (3.1 1)
Noting that f ;(.) and di (.) are Lipschitz and that 0 5 d; (.) I
1, one obtains
sufficiently small if all three of the following conditions are Remark 3.6: The above theorem gives conditions in which
satisfied: the interval in Theorem 3.1 can be made infinite. For the
1) there exists no chattering in the system; case when y ( t ) apprloaches a uniformly asymptotically stable
2) TO<< e--kc(Lpto), where k, are the Lipschitz constants equilibrium point, y.., the difference, Ilz(t) - y ( t ) l l , can be
for f,(.); made arbitrarily small1 for all t 2 t o assuming I Iz(t0)- y(to) I I
3) To << where m, are the Lipschitz constants and the switching period are sufficiently small.
for d,(.). Remark 3.7: Suppose f,(.) and di(.) have continuous par-
This is not to say that for every system in question, the tial derivatives. Then, for an equilibrium point, ys, of (3.2) to
switching period must be chosen so that 1)-3) are satisfied. be uniformly asymptotically stable, it is possible to check that
For example, if solutions to (3.2) decay exponentially to an
equilibrium point, then condition 2) can often be relaxed. It is
important to remark that condition 1) must always be fulfilled
or else the solutions of (3.1) will not be defined in the usual
{
Det S I - - 8.f0 (Ys
ay
-
sense.
Remark3.4: Based on the Theorem 3.1 and the above
discussion, it is possible to determine general conditions that
suggest the improvement of the accuracy of approximation have all solutions with R e ( s ) < 0.
between the original (3.1) and the approximate (3.2) system. Remark 3.8: Theorem 3.2 guarantees that under the proper
Clearly, the approximation becomes better as the switching conditions, when (3.2) is stable, then so is (3.1). Unlike
period becomes smaller, but also, as Remark 3.3 notes, the (3.2), however, the solution to (3.1) will not in general
approximations will tend to improve for systems with smaller approach an equilibrium point as t + 00, since (3.1) is a
Lipschitz constants, i.e., the smaller k, and m, are, the time varying integral equation. In general, the solution to
more accurate the averaging technique will tend to be (for (3.1) will (assuming it is stable) approach a periodic orbit.
general systems) and the better for linear systems than for However, this periodic orbit will not necessarily be in the
nonlinear systems. Additionally, as Theorem 3.2 suggests vicinity of the equilibrium point of the averaged equation,
below, if the averaged system is stable, then the averaging unless T is sufficiently small. In fact, (the theory clearly
approximations will also improve. Conversely, if the averaged shows that) it is possible to construct examples in which
system is unstable, the averaging approximation tends to (3.1) has an asymptotically stable periodic orbit for all T ,
worsen. Finally, as is clear from (3.3), a necessary condition but is only in the vicinity of the equilibrium point of (3.2)
for the solutions of (3.2) to approximate the solutions of (3.3) when T + 0 (see Section IV). This behavior becomes more
is that the initial conditions of the two systems must be chosen pronounced in feedback controlled (as opposed to open loop)
in appropriate neighborhoods. PWM dc-dc converters due to the nonlinearities, and is not
Remark 3.5: One of the main advantages of the averaging noted in [3] and [9]. We further explain this phenomenon in
technique is that nonlinearities are maintained in the averaged [lo].
system. Hence, the approximation of (3.1) by (3.2) is valid Remark 3.9: In Theorems 3.1 and 3.2, the feedback signals
even when the states, x, become large, which would not be are compared with tri(t, T ) , shown in Fig. 1. However, all
true if a linearization technique were to be used. The averaging the above theorems remain valid for triangle waves as shown
approximation is, therefore, valid for large signals. in Fig. 2 also, provided that they are rescaled to vary between
As stated earlier, when the solution to the averaged equa- zero and one (see Section IV). Furthermore, it is not necessary
tion approaches a uniformly asymptotically stable equilibrium to compare each d, (.) with the same function with the same pe-
point, the solutions of (3.1) and of (3.2) will remain close riod. For instance, in (3.1) we might have U(&(.) - tri (., T,))
to each other on an infinite time interval for a sufficiently instead of U(&(. ) - tri (., T ) ) ,where T, might not equal T3,
small switching period. The following theorem is an immediate for i # j. As long as each T, is sufficiently small, all previous
consequence of this fact. The proof is almost identical to results will remain valid.
Proposition 4 of [3] or Theorem 2.2 of [12], and therefore,
is omitted. C. Ripple Estimate
Theorem 3.2: Let ~ ( tand ) y ( t ) denote the solutions to (3.1)
It is often desirable to obtain an estimate on the ripple of
and (3.2), respectively, and let ys E R (ys # y ( t 0 ) ) denote
the system, which will be denoted in this paper as Q ( t ,T , .).
a uniformly asymptotically stable equilibrium point. Suppose
Then, practical applications of averaging tell us that a better
that y(t) i ys as t 4 00.
approximation of the solution to (3.1) will be given by
Then there are constants Po(q) and To(q)such that, for any
v > 0, any I l 4 t o ) - Y(t0)ll < P, 0 5 P < Po < v> and
O<T<To
where ~ ( tand ) y(t) are the solutions of (3.1) and (3.2),
Il4t) - Y(t)ll < r/ (3.19)
respectively, T is thie switching period, and Q ( t ,T , .) is the
for all t 2 t o . ripple estimate obtained by the following algorithm.
548 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 11, NO. 4, JULY 1996
(4
Fig. 2. Other possible triangle waves.
Consider only the right-hand sides of (3.1) and (3.2). Let verter was considered in [3]:however, the theory developed
z(t0) = y ( t o ) , and replace every z ( s ) and y(s) in (3.1) and in [3] does not extend to closed-loop operation (as do the
(3.2) by the constant c E 72". Now take the difference between theorems in this paper). Assuming the converter is operating
(3.1) and (3.2) to obtain in the continuous conduction mode, the closed loop (rescaled)
system description is given by
N
(3.21)
1 r7
E
b= [f]
As the switching period becomes smaller, the amplitude where the components of x ( t ) = [ i ~ ( tvc(t)lT
), are the
of q ( t ,T , .) will also become smaller and ripple of the inductor current and capacitor voltage. Note, that since the
system will become negligible. Additionally, an adjustment triangle wave in Fig. 3 varies from 0.7-3 V, it is necessary to
on the initial condition can be made by solving the equation rescale the system into (4.1) so that Theorems 3.1 and 3.2 can
+
.(to) = y ( t o ) Q ( t , T , & t o ) ) , for y(t0) in terms of .(to). be applied. This is easily done scaling the duty ratio function
using the minimum (trimin = 0.7 V) and maximum (trimax
The general expression for the ripple estimate (3.23) is an
important contribution of this work and has been used in [lo] = 3.0 V) values of the triangle wave:
to help model the effects of switching at lower frequencies.
g(z) - trimin
d(x) =
IV. APPLICATION
EXAMPLE trimax - trimin
Consider the PWM boost converter with feedback control where g(z) is defined in Fig. 3. For this specific g(z), we have
structure as shown in Fig. 3. Open-loop operation of this con- VREF= 0.312.3, k l = 0.412.3, and = -0.112.3.
LEHMAN AND BASS: EXTENSIONS OF AVERAGING THEORY FOR POWER ELECTRONIC SYSTEMS 549
L = 50p.H i&) I II
E = 5V
(V)
9-
8.5-
8.
7.5.
7.
6.5.
6.
5.5.
5.
4
4.5. I I
24, 56, 7 : ~ lObu 12\u 15'0~ 17u
: 20011 22\11 25bu 275u 3 OU t ( S )
Fig. 4. Simulated start up transient response of capacitor voltage for (4.1) and (4.2) for different values of switching frequency. __ fa = 1 MHz;
_-_- fs = 100 kHz; - - - - f. = 50 kHz; - - - - - average.
1- 9-
900111- 8 5-
100m- 4 5-
0- 4
111 t(s)
Fig. 5. Simulated start up transient response of both capacitor voltage and inductor current for (4.1) when switching frequency equals 100 kHz. __
vc(t); - - - - iZ(t).
1- 9.
900m- 8.5.
800111- 8-
7oom- 7.5-
600111- 7-
500111- 6 . 5 -
400111- 6.
300m- 5.5-
Zoom-
100m- 4.5.
5.
d
0- 4-
Fig. 6. Simulated start up transient response of both capacitor voltage and inductor current for (4.2) and ripple correction (4.3) when switching frequency
equals 100 KHz. __ vcavg(t); - - - - vcr1p ( t ),.- - - - - alaVg(t);- - - - - - d r l p ( t ) ; .
and (3.2), respectively. Then for any t E [ t o , L],L 2 t o Proof of Lemma A.3: Without loss of generality, assume
that t o = 0 (initial time can always be redefined so that this
is the case.) By definition
Ilx(t)ll L I l ~ ( t 0 ) l lexp
%(I? - tri(t, T ) )=
1 t E [nT,nT DT] +
+
0 t E [aT DT, ( n f 1)TI
Ilv(t)ll L IlY(t0)ll exp n = 0, 1, 2, . . . . (A.7)
Assume that D f 0. (The case when D = 0 is trivially
where k , are the Lipschitz constants of fi, previously defined. proved since both the left and right-hand side of (A.6) are
ProofofLemma A.2: By (3.1j identically equal to zero). Suppose 0 5 t _< DT. Then
552 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 11, NO. 4, JULY 1996
+ /MT
[u(D- tri(s, 5")) - D]d s .
I1 (A.lO) N
for any t o 5 t l 5 t 2 5 L.
M s:[u(D-tri(s, T ) ) - D ] d s a n d J h T [u(D-tri(s, 7'))-
D ] d s = J,
t-MT
[ u ( D- tri (s, T ) )- D] d s . Note that Since z ( t ) is a piecewise constant function, there will
always exist a sequence to = ao < a1 < a2 < ... < up = t ,
t I L, and a set of constants { c 3 } ;j = 1, 2, ... , p , with
cI = 5 ( t ) on the interval t E [u3-1, a 3 ] ,such that
l
3
s)) 5
f z ( ~ ( s ) ) d r ( z (ds P ; t E [to,LI. J = 1, 2, ... , p (A.20)
r=l
(A. 14) the proof is complete. Q.E.D.
LEHMAN AND BASS: EXTENSIONS OF AVERAGING THEORY FOR POWER ELECTRONIC SYSTEMS 553
By Lemma A.4, for any y > 0, there exist a TO= T 0 ( y , L ) Richard M. Bass (S’82-M’82-SM’94), for a photograph and biography, see
such that, for 0 < T 5 To, p. 98 of the January 1996 issue of this TRANSACTIONS.