INT. 1. CONTROL, 1996, VOL. 65, NO. 2, 205-222
Dynamical adaptive pulse-width-modulation control of DC-to-DC
power converters: a backstepping approach
HEBERTT SIRA-RAMIREZt, MAURICIO GARCIA-ESTEBAN$ and
ALAN S. I. ZINOBERS
Adaptive regulation of pulse-width-modulation (PWM) controlled DC-to-DC power
Supplies is proposed using a suitable combination of dynamical input-output
linearization andthe backstepping’ controller design method. A nominal parameter,
input-dependent, state coordinate transformation of the average PWM converter
‘models leads toa type of pure parameter feedback canonical form associated with the
Fliess generalized observability canonical form of such average models. A back-
stepping design procedure can then be imimiediately devised which leads to a
1
HAO=} Md, for 0-< po <1 aay
0, for ui)<0
Finally, it should be noted that when the sampling period T's suficiently small, the
actual values ofthe state variables V and I, rather than thei average values, (, and <,,
may be used for the on-line solution of the computed duty ratio funtion y. This
procedure is precisely at the heart ofthe state average method for PWM designs. A
theoretical justification ofthis procedure has been given by Sira-Ramirez el. (1993).
Summarizing, a dynamical PWM controller achieving the asymptotic stabilization
of the average input inductor current of the boost converter circuit (1) to a desired
constant equilibrium value y = ¥, is given by
1, foryst 2)
where d= 1,4, denotes firstestimate of 40, 1,4. simple Lyapunov stability
Argument shows thatthe pseudo controller (25) an the update law
4 = Hx) 26)
yield a closed-loop stable system for which 2, is guaranteed to converge to zero.
Since x, is not really a control input, one defines the pseudo-control error variable
2, as the difference between x, and its required value, computed in (25). Let
x-[-o 2 Bhe-09- ai] en
By solving for x, from (27) and using (23), one obtains a new state coordinateDynamical adaptive PWM control of power converters 213
transformation defining, respectively, the stabilization error and the pseudo-control
error variables, 2, and 2,
8)
The corresponding inverse transformation is simply obtained as
maatY
° -
we nat $ 29)
* aval ”
‘The first equation of the transformed system may then be written as
Aeamay
1 a
(40, — dB} ls = +-(40,~40}) (30)
+ Le. ai « ) G0)
which can be briefly expressed as
—e.2,4(40-d0 w, ey
where
1 B® @)_%
wend) =[[g lg feranefte,-at]-2) 0 0 1] ee
lS ai +6 3) 90 1] 6
Note that the update laws corresponding to 40! and 40} will yield constant values
for such estimates. Note, moreover, that these two estimated parameters are not
needed in this first step of the backstepping calculation,
‘We let 1} denote the first component of the regressor vector W4(z, 2» 40}).
The first adaptation law (26) may then be rewritten, in terms of the new error
variables 2, and zp, as
&, 4) @3)
Step 2: We proceed to complete the state coordinate transformation (28) and (29) of
the original phase variables by considering now the differential equation for the
pseudo controller error z,. Using the definition of z, and z, and the first incremental
parameter update laws for the involved components of the vector 48, one obtains,
after long but straightforward manipulations, the following expression:
, 40) 0,) 6, 8,1 =n)", +)
‘» 50)-O,)
= O4(4(2), 2p, 40") — ,) — 48, (1 —p)* + Y) = 40,0412,
eile 6125 BOE) — BO) + 241 + W) +04(40, Wi + 48) ea)
where 3,2» 40s given by the second equation of (29), which we do not substitute24 H, Sira-Ramire: et al
just to avoid lengthy intermediate equations. In the rest of this section x, stands for
21,21, 40),
Tf We now equate the dynamics obtained in (34) for 2, to the dynamics of an
asymptotically stable behaviour for z,, given by
4=-Gtn @>0 @9
cone can immediately solve for the required control input derivative upon invoking,
once mote, the certainty equivalence principle. In this instance the unknown value of
the vector 49 will be replaced by a new vector of parameter estimates, denoted by 0
One then obtains
(= 0)8,
E exze eles —6421— (AP (2,
Binet a aa am
afl HOD (ABP le 20-5)—(1 +B), ea)
$.0,(x,~0,) + 4050 ,(1 —1)* (2, + Y) + OC, ~60)} 6)
where Bf, j = 2,3, represent the new estimates ofthe incremental parameter vector
components 49, j = 2,3, and x, is given by (29). The expression for the dynamically
controlled error variable z, (i. the closed-loop behaviour of z,) is found to be
~(40,— 4B)(%,— 09] + el(40, ~ 485) W) + (40, ~ 493) en
(40,— 40) 0,(1—W*(,+¥)
which can also be briefly expressed as
*sZq+ (40 — SOP wy, 25, Hs 40", AB") 38)
The regressor vector for the new estimation process is thus given by
) nent APY 6,20 BO) 0)
[Note thatthe dependence of w, on 40*is implicit through its dependence on as given
by the solution of (36).
‘Asin the previous step, an incremental parameter adaptation law forthe vector of
new estimates 40® can be devised to achieve simultaneously a stable adaptation
process and an asymptotic convergence to er0 of th pseudo-control error variable zy
Such a new incremental parameter update law is given by
BP = 202,248,040") (ao)
3.1.1. Summary of adaptive controller expressions for the boost converter. The adaptive
PWM controller is next summarized in terms of the original state variables of the
system. The constant Y stands for the desired value of the input inductor current 1(0).
The constants ¢, and ¢, are positive design constants, satisfying ¢,,c, > 2,Dynamical adaptive PWM control of power converters 215
The adaptive feedback regulated switch position function is synthesized as
{i for