Type-2 fuzzy logic control of a DC/DC buck converter

J. Solano Martinez. D. Hissel. M-C. Pra.

*University of Franche Comt, FEMTO-ST UMR CNRS 6174

90010 Belfort, France
{jsolanom,daniel.hissel,marie-cecile.pera}@univ-fcomte.fr

Abstract: This paper presents a fuzzy logic controller (FLC) that performs the output voltage regulation
of a DC/DC buck power converter. As type-1 FLC are well known in this application, the objective is to
consider a type-2 FLC. The voltage controller is implemented, evaluated and compared (using a type-1
and a type-2 FLC) by simulation and by experimentation.
Keywords: Type-2 fuzzy logic, voltage control, buck converters, simulation and experimental validation.

1. INTRODUCTION
A classical structure of fuzzy logic controller (FLC) as
presented in Hissel (1998) is implemented to compare the
performance of type-2 and type-1 FLCs. The voltage
controller is implemented, evaluated and compared (using
type-1 and type-2 fuzzy logic) by simulation and by
experimentation. The objective of this work is to study the
viability of the use of type-2 fuzzy logic to perform the
voltage control of a buck power converter. The optimization
of the fuzzy logic system is then not considered.
This paper is organized as follows: Section 2 presents some
motivations to consider type-2 fuzzy logic controllers and
presents the membership functions used in fuzzy logic
systems. Section 3 is devoted to the design of the voltage
controller. Finally, in Sections 4 and 5 simulation and
experimental results are provided and conclusions and
prospects are discussed.
2. TYPE-2 FUZZY LOGIC AND UNCERTAINTY
Fuzzy logic allows using linguistic labels and human
knowledge to represent or model physical process and to
design their control systems. Fuzzy control has been used in
different applications such as voltage control in power
converters or speed control in electric machines.
The concept of type-2 fuzzy logic systems, which handle the
uncertainty, was introduced by Zadeh (1975). The first type-2
logic system was developed and presented only 23 years later
by Karnik and Mendel (1998). Type-2 fuzzy logic has been
used in different applications such as control of electrical
machines Barkati et al. (2008), Barkat et al. (2011) speed
control of diesel engines Lynch et al. (2006), power system
stabilizers Robandi et al. (2008), or transformer diagnosis
Florez et al. (2008).
2.1 Membership functions
Membership functions (MFs) establish a relationship between
numerical values and linguistic labels. Type-1 fuzzy MFs
(T1-MF) represent the membership degree for a variable x.
Type-2 fuzzy MFs (T2-MF) also consider an uncertainty U of
the membership degree. T1-MFs are a particular case of T2-
MFs where U is 0. Membership functions are classified as:
Singleton, if is 0 but in a single point of the domain is 1.
Interval type-1, if is 0 except in the interval [lb, rb] where
=1 if lb < x < lb
Type-1, if is a crisp value which varies between 0 and 1
Type-2, if is represented by a secondary degree :

Fuzzy sets as used in type-1 fuzzy logic do not represent the 1. Interval type-2 MFs if is an interval type-1 MF
uncertainty. This is the reason why Zadeh (1975) proposed to 2. General type-2 MFs if is a type-1 MF
represent it by using type-2 fuzzy sets. Uncertainty is
involved with any situation with some lack of information: it These membership functions are illustrated in Appendix A.
may be incomplete, imprecise, fragmentary, not fully reliable, Interval T2-MFs are easy to implement and then have been
vague, contradictory or deficient in some other way Klir used in almost all the works about type-2 fuzzy logic. For
(1995). Different authors define and classify different types simplicity reasons, interval type-2 MFS has been also
of uncertainty. The classification proposed in Ioos (2009) selected in this research.
illustrated in (Fig. 1) fits to describe uncertainty in electrical
engineering.
Uncertainty is also present in fuzzy logic systems as
explained by Mendel (2001): the uncertainty 1) about the
meaning of the words that are used in the rules, 2) about the
consequence that is used in a rule, 3) about the measurements
that activate the fuzzy logic systems and 4) about the data
used to tune the fuzzy logic system parameters
Fig. 1. Sources of uncertainty. - Adapted from Ioos (2009)
 3. BUCK CONVERTER VOLTAGE CONTROLLER
A DC/DC buck converter is used to couple a DC voltage
source to a load, while controlling the quantity of energy
supplied by the former by reducing the voltage. The buck
converter structure as presented on (Fig. 2) allows transit of
energy from the source to the load but not vice versa
The output voltage of the buck converter depends on different
parameters and operation conditions as: the input voltage, the
duty cycle, the switching frequency and, the inductance,
capacitance and load values. Generally, the switching
frequency, the inductance and the capacitance values are
constant. The input voltage can be also considered as
constant. As the load is imposed, the output voltage is
controlled by controlling the duty cycle.
3.1 Overview
The voltage controller requires one single input which is the
error signal (e) between the reference and the measured
voltage (Vref and V0). The output of the controller is the duty
cycle of the converter (c). The controller computes an output
for a sample time denoted k. The sampling period (Tech) is
constant. The PWM is performed at a constant switching
frequency (fsw).
The objective is to evaluate different fuzzy logic controllers
using the same structure as shown in (Fig. 3). The difference
between the FLC is the uncertainty in their membership
functions. The first FLC has an uncertainty of 0% (T1-FLC),
the second and the third consider an uncertainty of 20% and
50% respectively (IT2-FLC). This choice permits to evaluate
and compare type-2 and type-1 FLC under similar conditions.
*
c ( k ) = c (k ) g m (6)
Finally, the relative change is integrated to find the duty cycle
of the converter by using an integrator as defined by (7).
c (k ) = max(1, min (0, c (k ) + c (k ) Tech )) (7)
The whole fuzzy logic controller is illustrated in (Fig 4).
3.3 Fuzzy logic system design (rules and MFs)
Table 1 presents the considered classical anti-diagonal rule
base. The seven membership functions are: Negative High
(NH), Negative Low (NL), Negative Very Low (NVL), Zero
(Z), Positive Very Low (NVL), Positive Low (PL) and
Positive High (PH). The input and output membership
functions are defined as classically done in this real-time
application: triangular MFs for the two input fuzzy sets and
singleton MFs for the output fuzzy set.
The two input fuzzy sets (for e and de) are composed of
seven membership functions. The MFs are symmetric around
the zero axes and are defined in two steps by using only three
parameters:
1. Seven type-1 membership functions (Uncertainty =0%) are
defined by two parameters as illustrated in (Fig. 5).
The MF Z has its peak at x = 0 and its right base at x = pvlx
The MF PVL has its left base at x = 0, its peak at x = pvlx and
its right base at x = plx
The MF PL has its left base at x = pvlx, its peak at x = plx and
its right base at x = 1
The MF PH has its left base at x = plx, its peak at x = 1

3.1 Voltage controller definition
The error between the reference and measured voltage is
defined in (1)
e( k ) = V ref (k ) V0 (k )
The error variation is also considered and is defined in (2)
de(k ) = e(k ) e(k 1)
The uncertainty U in the membership functions is the third
considered parameter.
To construct the interval type-2 MFs, the type-1 membership
(1) functions are modified as illustrated in (Fig. 6). Here, three
different uncertainties were chosen: 0%, 20% and 50%.
(2)

The fuzzy logic controller handles normalized inputs (domain

[-1, 1]) as defined in (3) and (4).

e( k ) Fig. 2. Buck power converter

e* (k ) = max 1, min ,1 (3)
e
n
de(k )
de* (k ) = max 1, min ,1 (4)
den Fig. 3. Voltage controller structure
A FLC defines the relative change in the duty cycle as
defined in (5)
*
(
c ( k ) = f e* (k ), de* (k ) ) (5)

The output of the fuzzy controller is denormalized using a

denormalization factor gm to obtain the relative change in the
duty cycle as defined by (6).
Fig. 4. Fuzzy logic controller detail
Table 1. Fuzzy rules 1. Simulation setup: The fuzzy controller and the buck
converter are implemented in Matlab Simulink.
2. Experimental setup: The fuzzy controller is implemented
in Matlab and uploaded to a dSPACE programmable
controller which generates the signal to drive the transistors.
The experimental setup is presented in (Fig.7.)
The same parameters used in simulation are used in
experimentation as resumed in Tables 2 and 3.
4.1 Evaluation of the controller
The evaluation of the controller response is performed by
imposing an initial condition and three different operation
conditions:
1. Initialization: The output voltage reference is initially fixed
at 2.8 [V]
2. Start up: At t = 0 [s] the voltage reference changes to 5[V].
Fig. 5. Type-1 fuzzy membership functions The load is constant during this period (27.69).
3. Load regulation (-): At t = 1 [s]. The switchable load
resistance is switched off. (Load resistance from 27.69 to
120 )
4. Load regulation (+): At t = 1.5 [s]. The switchable load
resistance is switched on again. (Load resistance from 120
to 27.69 )
To compare the three fuzzy controllers, the Integral Absolute
Error (IAE) criterion is retained. IAE is defined in (8)
t
IAEV (t ) = Vref (t ) V0 (t ) dt (8)
0

Fig. 6. From type-1 to interval type-2 membership functions Table 2. Buck converter - Validation parameters
Description Parameter Value
3.9.2.1 Parameters Series Inductance L 1 [mH]
The choice of the parameters (plx and pvlx) defining the fuzzy Parallel Capacitance C 1 [mF]
sets is based on previous works about fuzzy control of power Fixed load resistance R0 120[]
converters Hissel (1998): Switchable load resistance Rch 36 []
To guarantee a fast convergence of the controller, the PWM Switching frequency fsw 15 [kHz]
parameters ple and pvle which define the error (e) Input voltage Vi 10.4 [V]
membership functions are relatively close to zero. Values of Reference output voltage Vref 5 [V]
ple = 0.15 and pvle = 0.3 are retained.
Table 3. Fuzzy logic controller - Validation parameters
To reduce the noise amplification of the controller, the Description Parameter Value
parameters plde and pvlde which define the error differential
Processor sampling time Tech 100 [s]
(de) membership functions are relatively close to one. Values
of plde = 0.4 and pvlde = 0.7 are retained. Error normalisation en 3 [V]
Error differential normalisation den 5 [V]
Finally, and for simplicity reasons, the output fuzzy set is
Denormalization factor gm 0.004
composed by singleton membership functions. The singletons
are uniformly distributed around the domain of the fuzzy sets.
The selection of these parameters is obviously subjective but
at least it permits to have a set of parameters to evaluate and
compare the considered type-2 fuzzy logic controllers.
The considered fuzzy sets (for uncertainty values of 0%, 20%
and 50%) are presented in Appendix B.

4. FUZZY LOGIC CONTROLLER VALIDATION

The proposed voltage controller is evaluated by two methods:
by computer simulations and experimentally. Fig. 7.Experimental setup

4.1 Results
Simulation and experimental results are illustrated in (Figs. 9
and 10) and resumed in Table 4. The voltage regulation is
performed as expected but the results are not in agreement (in
simulation the best results were found with an uncertainty of
0.5, in experimental the best results were found with an
uncertainty of 0.2). The difference between simulation and
experimentation results could be explained by the fact that in
simulation we do not model the noise, which is always
present in experimentation.
[V] V0
6 6
Fig. 9. Simulation results
5. CONCLUSIONS
In this paper, the voltage control of a DC/DC buck converter,
one of the classical applications of fuzzy control in electrical
engineering, was presented, developed and validated using
type-2 fuzzy logic. Simulation and experimental results
suggest that T2-FL can be used in this particular application.
However, it is too early to affirm that T2-FL controllers are
better performing than T1-FL controllers. Future research at
FEMTO-ST will be focused on the study of the tuning and
optimization of the parameters of type-2 fuzzy logic systems.
[V] V0

5 5

4 U=0% 4 U=0%
U=20% U=20%
3 3
U=50% U=50%
2 2
0 0.5 1 1.5 time [s] 2 0 0.5 1 1.5 time [s] 2

[ -] Duty cycle [ -] Duty cycle

0.8 0.8
U=0% U=0%
U=20% U=20%
0.6 U=50% 0.6 U=50%

0.4 0.4

0.2 0.2
0 0.5 1 1.5 time [s] 2 0 0.5 1 1.5 time [s] 2
[V s] V - Integral absolute error [V s] V - Integral absolute error
0.3
0.4
U=0%
0.3 U=0%
U=20% 0.2 U=20%
0.2 U=50%
U=50%
0.1
0.1

0 0
0 0.5 1 1.5 time [s] 2 0 0.5 1 1.5 time [s] 2
[V] V0 Start-UP detail [V] V0 Start-UP detail
6 6

5 5

4 U=0% 4 U=0%
U=20% U=20%
3 3
U=50% U=50%
2 2
0 0.02 0.04 0.06 0.08 0.1 0 0.02 0.04 0.06 0.08 0.1
time [s] time [s]
[V] V0 Load from 27.7 to 120 [] [V] V0 Load from 27.7 to 120 []
6 6
U=0% U=0%
U=20% U=20%
5.5 5.5
U=50% U=50%

5 5

4.5 4.5
0.96 0.98 1 1.02 1.04 1.06 1.08 1.1 0.96 0.98 1 1.02 1.04 1.06 1.08 1.1
time [s] time [s]
[V] V0 Load from 120 to 27.7 [] [V] V0 Load from 120 to 27.7 []
5.5 5.5

5 5

4.5 U=0% 4.5

U=0%
U=20% U=20%
4 4
U=50% U=50%
3.5 3.5
1.5 1.55 1.6 1.65 1.7 1.5 1.55 1.6 1.65 1.7
time [s] time [s]
Fig. 10. Experimental results
Table 4. Validation results - IAE
Simulation Experimental
Type-1 FLC U=0% 0.2194 0.2086
Type-2 FLC U=20% 0.2201 0.2059
Type-2 FLC U=50% 0.2166 0.2581
REFERENCES
Barkat, B. Nouri, H. and Tlemcani A. (2011). Non-
interacting adaptive control of PMSM using Interval
Type-2 Fuzzy Logic Systems. Fuzzy Systems, IEEE
Transactions on, Vol. 19, No. 5, pp. 925-936.
Barkati, S. E. Berkouk, E. and Boucherit, M. (2008).
Application of type-2 fuzzy logic controller to an
induction motor drive with seven-level diode-clamped
inverter and controlled infeed. Electrical Engineering
(Archiv fur Elektrotechnik), Vol. 90, No. 5, pp. 347_359.
Flores, W. Mombello, E. Jardini, J. and Ratta, G. (2008) A
novel algorithm for the diagnostics of power
transformers using type-2 fuzzy logic systems. In:
Transmission and Distribution Conference and
Exposition, 2008. IEEE/PES, Bogota, Colombia; pp. 1-5.
Hissel. D. (1998) Contribution la commande de dispositifs
lectrotechniques par logique floue Procdures de
rglage sur site par plans d'expriences et mthodologie
Taguchi. PhD thesis, Institut National Polytechnique de
Toulouse.
APPENDIX A. CLASSIFICATION OF THE MEMBERSHIP FUNCTIONS
(a)
Fig. 10. Membership functions (a) singleton, (b) Interval type-1, (c) Type-1.
(a)
Fig. 11. Membership functions (a) Interval type-2, (b), General type-2.
Iooss, B. (2009) Contributions au traitement des incertitudes
en modlisation numrique. HDR thesis, Universit
Toulouse III.
Karnik, N. and Mendel, J. (1998). Introduction to type-2
fuzzy logic systems. (1998) In: Fuzzy Systems
Proceedings, IEEE World Congress on Computational
Intelligence. vol.2 pp. 915-920.
Klir, G and Yuan, B. (1995) Fuzzy sets and fuzzy logic:
theory and applications. Prentice Hall PTR Upper Saddle
River, NJ, USA.
Lynch, C. Hagras, H. and Callaghan, V. (2006). Using
uncertainty bounds in the design of an embedded real-
time type-2 neuro-fuzzy speed controller for marine
diesel engines. In: Fuzzy Systems, IEEE International
Conference on, pp. 1446-1453.
Mendel, J. (2001). Uncertain rule-based fuzzy logic systems:
introduction and new directions. Prentice-Hall, Upper
Saddle River, NJ.
Robandi, I and Kharisma, B. (2008). Design of interval type-
2 fuzzy logic based power system stabilizer. In: Proc. of
world academy of science engineering and technology.
pp. 2070-3740.
Zadeh, L. (1975). The concept of a linguistic variable and its
application to approximate reasoning_I* 1. Information
sciences, Vol. 8, No. 3, pp. 199_249.
(b) (c)
(b)
APPENDIX B. MEMBERSHIP FUNCTIONS DESIGNED FOR THE VOLTAGE CONTROLLER

(a)

(b)

Fig. 12. Type-1 fuzzy membership functions (U=0%) to represent e (a) and de (b)

(a)

(b)

Fig. 13. Type-1 fuzzy membership functions (U=0%) to represent e (a) and de (b)

(a)

(b)

Fig. 14. Type-1 fuzzy membership functions (U=0%) to represent e (a) and de (b)