BUCK/BOOST CONVERTER CONTROL WITH

FUZZY LOGIC APPROACH

Bor - Ren Lin and Chihchiang Hua
Department of Electrical Engineering
National Y unlin Institute of Technology
Touliu Citv. Yunlin Countv, Taiwan. R.O.C.

-
Abstract In this paper, the application of fwzy control to DC- measurements, these have to be fuzzied by fuzzification.
DC converters operating at finite switching frequency is studied. Similarly, since the converter cannot respond directly to fuzzy
Several control methods currently used for buck/boost converters controls, the fuzzy control sets generated by the fuzzy algorithm
are compared to the fuzzy converter control. Simulation results have to be defuzzified by defuzzification. The fuzzy control
for several control methods are presented. The simulations show algorithm consists of a set of fuz.zy control rules which are
that the fuzzy control method has better dynamic performance related by the concepts of fuzzy implication and the
and less steady state error. compositional rule of inference. These fuzzy control rules are
combined by using the sentence connectives and and also. A
I. INTRODUCTION fuzzy variable is expressed by natural language. For example, the
The modeling of networks which contain switches has error of the output voltage can be defined by linguistic variables
drawn much attention because of the unusual properties of "zero ( Z ) " , "positive small (PS)", "positive medium (PM)".
switches in comparison with other circuit elements. The "positive big (PB)", "negative small (NS)", "negative medium
difficulities in the modeling of switched networks are mainly due (NM)", "negative big (NB)", etc. The basic fuzzy set operations
to the nonlinear and time varying nature of switches. Most are union, intersection and complement. Let A and B be two
modeling in power electronics is intended to convert the
nonlinear and time varying model to an ideal or non-ideal switch
model [1]-[2]. Then, the state space method is used to solve the fuzzification
state equation for the system. Recently, sliding mode controlled
DC-DC converters have been introduced [3]. The main advantage
o f the sliding mode control is insensitivity to plant parameter
variations and external disturbances, which leads to invariant
dynamic and steady state response i n the ideal case.
Fuzzy set theory is a theory about vagueness or Reference signal
uncertainty and enables one to use nonprecise, ill-defined Fig. 1. Closed loop control configuration
concepts. The neural network and fuzzy controller have the
ability to learn to control a nonlinear systcm. When a neural fuzzy sets in U with membership function pA and p B
network is to control a plant, i t is necessarq to train the neural respectively shere U is the universe of discourse. The universe
network controller to learn the plant behavior. Recently the of discourse is a collection of fuzzy variable possible values {x
application of fuzz,y control has been successful in nuclear
reactor control 14-51. elevator control [ 6 ] , automatic train Definition 1: Union
operation systems [7-0] , automatic transmission control [IO], ) the union AUB
The membership function ~ ( A U Bof S
asymmetrical four quadrant power converter [ 111, and two- defined for all x E U by
dimensional motion control [12]. The fuzzy control does not need
an accurate mathematical model of a plant. Therefore, it IS P(AUB)(X)= max { PA(X).PB(X)1 (1)
applicable to a process where the plant model is unknown o r ill-
defined. The fuzzy control is also nonlinear and adaptive in
Definition 2: Intersection
nature which gives it robust performance under parameter ~ of the intersection
The membership function p ( 18)
variations and load disturbances. In this paper, a f u u y algorithm
A n B is defined lor all x E U bq
based on linguistic rules describing the operator control strategy
is applied to control DC-DC converters. Simulation results show
that the performance of the fuzzy conlrol in buckiboost
converters is somewhat better than controls based upon state
Definition 3: Complement
space and sliding mode methods.
The membership function pq of the complement of a
fuzzy set A is defined for all x E U by
11. FUZZY CONTROL ALGORITHM
Fig. 1 shows the basic closed loop fuzzy control
configuration. Since the converter produces nonfuzzy

0-7803-0891-3/93$03.0001993IEEE 1':
I 342
Fig. 2 shows these three basic fuzzy set operations. Normally, the change of control action. Table 1 shows the fuzzy model based
fuzzy rule has the if .. then.. structure, such as o n fuzzy rules. For example, if the error is NB and the change of
error is NB then the change of control output is NB.
If x is Land y is M then z is N
Table 1. Fuzzy model
where x and y are input fuzzy variables, z is the output fuzzy (Table entries are change in control output)
variable; L,M and N are fuzzy subsets in the universe of error
discourses X, Y and Z respectively. A fuzzy subset can be NB, NE NM NS Z PS P M PB
t i e r ror
NM, NS, Z , PS, PM and PB. Fig. 3 shows the basic fuzzy NB NB NB NE NB NM NS Z
partition of membership functions for error, change of error and NM NE NB NB NM NS Z PS
NS NE NB NM NS Z PS P M
z NB NM NS Z PS P M PB
PS NM NS Z PS P M PB PB
PM NS Z PS PM PB PB PB
PE Z PS P M PE PB PB PB

NB: Negative Big , NM: Negative Medium, NS: Negative Small

, Z: Zero, PB: Positive Big , PM: Positive Medium, PS: Positive
Small

There are n fuzzqyrules defined as

R I : if x 1:s L1 and y is MI then z is NI

R?: i f x iii L2 and y is M2 then z is N2
R3: if x its L 3 and y is M3 then z i s N3

.......................................,...,,
Rn: if x i!j Ln and y is Mn then z is Nn
I I

These individual rules are combined by using the union operator.

R = R1 U R2 U R3 U ........ U Rn

A fuzzy controller should compute the actual output control

Fig. 2. Basic fuzzy operations : Union, Intersection and
signal lor a spec1fic input signal from a given set of fuzzy rules.
Complement
The compositional operator is applied to infer the output from the
given process sitates x, y and the fuzzy relation R. A fuzzy
relation R from PI to B is a fuzzy subset of Cartesian product U x
FFB NY NS 2 PS PM PB
V, where A E U and B E V. The fuzzy relation R can be defined
as follows :

Once the relation R is defined, the an actual input fuzzy set A '
can be used to compute the resulting output fuzzy set B'. There
are two basic inference rules to compute the resulting output
(b) err
fu7.zy set B', max-min and max-product, in Fig. 4.
qrLB NY NS Z PS PM PB

(4 err
Fig. 3 . Basic fuzzy membership functions (a) Triangular
membership function (b) Trapezoidal membership function ( c )
Bell-shaped membership function

1343
 . .
step 2) Calculate the error and change of error.
step 3) Determine the fuzzy subset and membership function for
error and change of error.
step 4) Determine the change of control action according to the
ind~vidualfuzzy rule.
step 5) Calculate the actual change of control action by the
defuzzification operation.
step 6) Send the change of control action to control the plant.
step 7) Go to step 1.

111. MATHEMATICAL MODEL OF BUCK/BOOST

CONVERTER
Fig. 5 shows the buckiboost converter system with the
controller.
R. R d n

lbl

Fig 4 Ba5ic l u z q inference rules (a) max-min inference (b)

max-product inlerence

These rules can be extended for any number of fuzzy inputs

and outputs. Since the control action z cannot directly control the I
I -- .
V..
I,

plant, the defuzzification operation can be performed by either

the mean of maxima or the center of gravity method. The mean
of maxima generates a control action which represents the mean
value of all local control actions whose membership functions Fig.5. Fuzzy logic converter conlrol system
reach the maximum. In other words, if' the set possesses a single
The actual output voltage V, is compared to the reference
maximum, this value is the control action takrn. If several local
voltage Vref to produce an error signal that is used to determine
maxima appear in the set, then the control action to be executed
is the mean value of the n z, at which the membership function is the switch duty cycle.
the maximum value. The state-space averaging method is very useful for
analyzing the low-frequency. small-signal performance of
switching circuits. I t is applicable when the power switching
f zi period is short compared to the response time of-the output. For
-
z, = i = l
n
the subsequent discussions, the method is illustrated by
considering the bucWboost converter with a second order filter,
and assuming continuous current jn the inductor.
Thc centroid defuzzif ication method is uniquely
determined lrom the f u l l y response set R as follows

$z dddz
(conhnuous case) (8) Switch on :
z, =
p z ) dz
X = A1.X + B 1 * V d c
n

switch off :

The average-state equation is then a weighted average of the two

where n is the number of quantization levels of the output. Then
the output signal zo is applied to control the inverter. state equations (10) and ( 1 I ) based an thc duty cycle ratio d.
The fuzzy control algorithm is described as follows :
step 1) Sample the output signal of the plant.

1344
where A = d*A1 + (l-d)*A2 and B = d*B1 + (I-d)*B2. Then the
20 ...........
small signal linealized behavior of the bucWboost converter was
described [ 131 to solve the problem. Another approach is to use
difference equations for the mathematical models of discrete 19.5 ................ ...
systems to solve the problem. The fuzzy logic controller is used
to update the duty cycle for desired output voltage.
19 ..............

IV. SIMULATION RESULTS

005 0.052 0054 0.056 0058 006
The simulation parameters for the bucWboost converter
are RL=O.~Q, Q=O.lQ, Ro=118Q, Rc=1.18Q, Rt=0.2Q, (a)
L=0.42mH, C=14MpF, vdc=20 volts and duty cycle d = 0.2, 0.5
20 :
and 0.6. Fig. 6 shows the output voltage transient behavior by
load change (from R&OQ to &=20Q) in bucWboost converter
with d=0.2. Fig. 7 shows the output voltage transient behavior by 19 5 -
load change (from b = 2 O Q to b = l O Q ) in bucWboost converter
with d=0.5. Fig. 8 shows the output voltage transient behavior by
load change (from b = ? O Q to &=4OQ) in bucWboost converter 19 -
with d=0.6. An examination of the data for the curves shows that
the fuzzy control method has less steady state error and better
transient performance. The f'uzzy control method has good
performance compared to the other methods in this simulation.

0 05 0 052 0 054 0 056 0 058 0 06

(C)
47
0 05 0 051 0 052 0053 0 054 0 055 Fig 7 Buck/boost converter transient response from 20 Q load to

;1.------j
5 . . . . . . . . . .; . ......
(a)

. . { . . . ........ . . . . . . . . . . ~. .................
~
10 R load with d=O 5(a) fixed tolerance band method (b) sliding
mode method (c) fuzzy logic approach

4.9 ......... j ......................

4.8 ....................
% ...............................

.:... ........................................
31.5
3151 h
.................... .j..
..... ........

........ .........
i
I I
30.5 .......... .: . . . . . . . . . ..:............ . . . . . . . . . .
47
0 05 0 051 0 052 0 053 0 054 0 055

30 r
2 9 0 5 h 0 5 2 0054 0056 Om8 006 '
vi P+ (a)
5 h

4 . 9 -. . . . . . . . . . . . . .; . . . . ..:.. . . . . . . :...... . . . . .

_
4.8 . . . . . . . . . . ; ......... .: . . . . . . . . ..; ........... < ............:

4 71 1
0 05 0 051 0 052 0 053 0 054 0 055
(c)
Fig 6 BucWboost converter transient response from 40 Q load to
20Q load with d=O 2(a) fixed tolerance band method (b) sliding
mode method (c) fuzzy logic approach

1345
 32
continuously variable transmission", in Proc. Int. Congress on
31.5 _........... .:. ............:.. . . . . . . . . . ..:. ........... .:. ........... .:. . . _ TransportationElectronics, Dearbom, MI. 1988.
[ 1 I ] L. E. Borges da Silva, G.-E. April and G. Olivier,"Real time
.:.
............ .j.. .......... ........... .:.. . _ fuzzy adaptive controller for an asymmetrical four quadrant
power converter", IAS, 1987, pp. 872-%78.
............ .:. ........................ .:. . ._ [ 121 Liang-Jong Huang and Masayoshi Tomizuka, "A self-paced
fuzzy tracking controller for two-dimensional motion control",
IEEE System, amn, and Cybernetics, vol. 20, no. 5, 1990, pp.
1 I 1 5 1 124.
29.5

V. CONCLUSION
The main disadvantage of fuzzy control is the lack of a
systematic procedure for designing the control rules and choosing
membership functions. A new research area might be to use
neural networks to adjust the membership functions and decide
the control rules. Since fuzzy control is related to linguistic rules,
this control can be combined with expert systems to create more
knowledge based control systems.

REFERENCES
[ 11 G . C. Verghese, M. E. Elbuluk and J. G. Kassakian, " A
general approach to sampled data modeling for power electronic
circuits" IEEE Trans. Power Electronics, Vol 1 No. 2, PP. 76-89,
April 1986.
121 G. Eggers. "Fast switches in linear networks" IEEE Trans.
Power Electronics, Voll, No. 3 , pp. 129-140, July 1986.
[3]B. J. Cardoso Fo. etc. "Analysis of swrtching frequency
reduction methods applied to sliding mode controlled DC-DC
convcrters" APEC 1992, pp. 403-410.
[4] J. A . Bernard, "Use of rule-based system for process control",
IEEE Control System Mag., vol. 8, no. 5, pp. .3-13, 1988.
[5] M. Kinoshita, etc. "An automatic operation method for
control rods in BWR plants", in Proc. Specialists Meeting on In-
Core Instrumentation and Reactor Core Assessment, Cadarache,
France, 1988.
161 F. Fujitec, "FLEX-8800 series elevator group control
system", Fujitec Co., Ltd., Osaka, Japan, 1988
[71 S. Yasunobu and S. Miyamoto, and H. Ihara, "Fuzzy control
for automatic train operation system", in Proc. 4th
IFAC/IFIP/IFORS Int. Congress on Control in Transportation
Systems. Baqden. April 1983.
[8] S. Yasunobu and S. Miyamoto. "Automatic train operation by
predictive fuzzy control", i n Industrial Application of Fuzzy
Control, North-Holland. 1985,pp. 1-18.
[9] S. Yasunobu, S. Sekino, and T. Hasegawa, "Automatic train
operation and automalic crane operation systems based on
predictive fuzzy control", in Proc. 2nd IFSA Congress, Tokyo,
Japan, July, 1987, pp. 835-838.

1346