Robust Control For DC-DC Buck Converter Under Parametric Uncertainties
Robust Control For DC-DC Buck Converter Under Parametric Uncertainties
net/publication/327020779
CITATIONS READS
5 517
7 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Kevin E. Lucas on 14 August 2018.
KEVIN E. LUCAS M.*,† , ERICK M. ROCHA*, WALTER BARRA JR.*, DAVID A. VACA B.† RENAN L. P. MEDEIROS¥,
MAYSA S. ALVES*, MAICKSON P. V. LEÃO*
*
Federal University of Pará, UFPa, Faculty of Electrical Engineering, Campus do Guamá, Belém, Pará,
CEP: 66075-900, Brazil
†
Escuela Superior Politécnica del Litoral, ESPOL, Facultad de Ingeniería Eléctrica y Computación, Campus
Gustavo Galindo Km. 30.5 Vía Perimetral, P.O. Box 09-01-5863, Guayaquil, Ecuador
¥
Federal University of Amazonas, UFAM, Department of Electricity, Av. Rodrigo Octávio, 6200, Coroado I,
CEP: 69080-900, Brazil
Abstract DC-DC Buck Converters are widely used in applications demanding the supply of a constant load voltage. Howev-
er, the dynamic behavior of the buck converter may change due to uncertainties such as load and input voltage variations, as
well as variations in the components of the buck converter itself. Because of these variations and the nonlinear behavior due to
switching, makes the controller design presents some challenges on obtain the desired performance. In order to guarantee stabil-
ity of the system and to preserve its dynamic performance under parametric uncertainties, in this paper a robust control tech-
nique, based on parametric uncertainties, is studied. The controller is designed by robust pole-placement technique based on a
desired performance in closed-loop, specified by a family of characteristic polynomials. The design procedure is formulated as a
set of linear constraints solved via linear programming methods. A classical controller, based on pole-placement technique, is
used for performance comparison. For performance assessment, several tests were carried out in Matlab/Simulink environment.
By calculating the ISE (integral square error) performance index, the obtained results show that the proposed controller obtain
robust performance.
Keywords buck converter, dc-dc converter, robust control, parametric uncertain, linear programing.
Resumo Conversor CC-CC do tipo Buck são largamente utilizados em aplicações demandando a regulação da tensão na car-
ga. No entanto, seu comportamento dinâmico pode sofrer variações devido à incertezas provocadas por variações de: tensão de
entrada, carga e dos componentes do conversor. Considerando estas incertezas e o comportamento não-linear, devido ao chave-
amento do conversor, o projeto do controlador apresenta alguns desafios para obter o desempenho desejado. Com o objetivo de
garantir a estabilidade robusta e preservar o desempenho dinâmico, é proposto neste trabalho, a investigação de uma estratégia
de controle robusta, baseada em incertezas paramétricas. O controlador é projetado por uma técnica robusta de alocação de polos
baseada em um desempenho desejado em malha fechada, especificado por uma família de polinômios característicos. O proce-
dimento de projeto é formulado como um conjunto de restrições lineares resolvidas através de métodos de programação linear.
Um controlador clássico, baseado na técnica de posicionamento de pólos, é usado para comparação de desempenho. Testes
computacionais, em ambiente Matlab/Simulink, foram realizados para avaliação de desempenho da metodologia proposta. Os
cálculos do índice de desempenho ISE (erro quadrado integral) mostram que o controlador proposto apresenta desempenho com
melhor desempenho robusto.
characteristics, and their control problems are well to guarantee robustness of poles in closed-loop in a
acknowledged as a challenging issue (Middlebrook & specified region through the roots of a characteristic
Cuk, 1976), (Perry et al, 2017), (Guo et al, 2009). polynomial interval. Therefore, parametric uncertain-
The main reasons are: load variations, power source ties of a mathematical model are analyzed in control-
variations, inevitable parameter uncertainties and ler design through interval analysis concepts.
exogenous disturbances. All these uncertainties could In this paper, the design of a robust controller by
lead to instability and performance degradation. pole-placement in the interval domain is presented.
The controller design is usually based on a linear Robust pole-placement is based on a specified by a
approximation of nonlinear dynamics of the convert- family of characteristic polynomials. Robust poles of
er. Standard linear controllers, such PID controllers the closed-loop system are obtained by linear pro-
for instance, are tuned in order to regulate the output gramming (Bhattacharyya et al, 1995),
voltage of converters (Forsyth, 1998), (Middlebrook (Bhattacharyya, 1992), (Bhattacharyya et al, 1995).
& Cuk, 1976). However, such conventional control- Theory analysis shows that the proposed control-
lers may experience a sensible performance deterio- ler maintain the stability and performance of systems
ration under varying operating conditions (Middle- against possible variations in their operating condi-
brook & Cuk, 1976). In order to assure robustness tions and load disturbances.
against parameter uncertainties, linear controllers A classical controller, based on pole-placement
must be designed by using robust control theory. technique through the Diophantine equation, is used
Therefore, it is advisable to design controller to cope for performance comparison. The simulations have
with a pre-specified range of parameter uncertainties, been carried out in MATLAB/SIMULINK.
load variations, or a wide variation of operating The rest of the paper is organized as follows. In
points. In addition, the performance of the DC-DC Section II, system modeling and problem statement
converters is also affected by external disturbance, are discussed. In Section III, the design procedure of
which may cause instability that usually appear due to the proposed controller is presented based on the
the measurement noises (Salimi et al, 2011). There- problem formulation in Section II. Section IV is
fore, the controller design process must ensure the aimed to present all simulation results to justify the
performance robustness. robustness of the proposed controller. Finally, the
Recently, several advanced control techniques conclusions are presented in Section V.
have been investigated for DC-DC converter system,
which provides powerful tools in system modeling
and robust controller design. These control tech- 2 Dynamics of a Buck Converter System
niques for DC-DC converter systems can be found in
the literature, such as robust parametric control A schematic diagram of buck converter is shown
(Bhattacharyya et al, 1995), adaptive control (Tan et in Figure 1, which comprises an dc voltage source
al, 2006), sliding mode voltage control (SMVC) Vi , a PWM gate drive controlled switch d , a diode
(Labbe et al, 2014), model predictive control (MPC)
(Kim et al, 2014) and multivariable robust control D1 , a filter inductor L , a filter capacitor C , a load
(Medeiros et al, 2018), to name a few. resistance RLoad and f s is a switching frequency.
In research on dynamic systems with parametric
uncertainties, the techniques that deal with this prob- The model also includes the parasitic element, the
lem have been studied extensively over the last 40 inductor resistance rL where the uncertain values are
years (Bhattacharyya, 1992), (Bhattacharyya et al, given in Table 1.
1995). In this context, control strategies, which aim
to implement adaptive, predictive and fuzzy control
algorithms, have been widely studied for the resolu-
tion of control problems of systems with parametric
uncertainties. However, adaptive and fuzzy strategies,
being able to offer efficient and satisfactory results,
can make the implementation and operation of the
Figure 1. Buck converter topology
process more complex (Barra Junior et al, 2005).
Another way to deal with parametric uncertain-
Table 1. Uncertain values of the buck converter.
ties in systems to be controlled, it is the robust pole-
placement technique. In this technique, the robust
Parameter Value Parameter Value
controller designed must assign the closed-loop poles
to a specific region of the complex plane, S, versus Vi 15V 15% Vo 5V
the parametric uncertainties related to the mathemati- rL 0.05 15% RLoad 4 50%
cal modeling of the system. In Lordelo & Ferrreira
(2002), uncertain systems are represented by transfer L 2mH 15% C 2000uF 15%
functions whose coefficients belong to real intervals fs 5kHz
and the robust pole-placement is developed in order
XXII Congresso Brasileiro de Automática – CBA2018, Brazil, 2018
By referring to the Kirchhoff’s voltage law and by pole-placement from the solution of a Diophantine
Kirchhoff’s current law, the dynamic model of buck equation. Then, in order to deal with the parametric
converter in continuous conduction mode (CCM) can uncertainties, a robust control by robust pole-
be written as follows: placement is proposed. Average model of buck con-
diL verter with voltage mode control is presented in Fig-
L dt d Vi vc rL iL (1)
ure 3.
dv v
C c iL c
dt RLoad
Where is iL the average inductor current, vc is the
average capacitor output voltage, and the duty cycle
d 0,1 represent the control signal, which is taken
as the conduction PWM signal.
Due to the non-linearity introduced by static
switching, the analysis can be divided into two differ-
ent operation intervals for each period in CCM. Fig-
ure 2 shows the state operation of the buck converter
in CCM. Figure 3. Voltage Model Control of buck converter.
P n r n r 1 o (14) closed-loop system has its poles within the roots
space of interval-desired polynomial, ensuring the
x a0 a1 ar b0 b1 br (15) robust stability (Lordelo and Ferreira, 2002).
It is known in the literature that when r n 1
the above equations admit a solution for the control- nr nn ar d nbr nr
n a n a d b d b
ler coefficients. Therefore when r n 1 , there al- n r 1 n r 1 n 1 r n r 1 n 1 r n r 1 (20)
ways exist a pole-placement controller [8].
The elements of M depend only on the coeffi- no ao dobo
o
o
cients of the numerator and denominator of G ( s ) .
The robust controller coefficients are calculated
nn dn by using the linear programming technique as shown
nn 1 nn d n 1 dn below,
nn 1 d n 1
min f x (21)
Ac x b
nn dn
M (16)
n nn 1 do d n 1 where,
o
no do A b
Ac , b max (22)
A bmin
no do where f x is an arbitrary linear function in x , x is
The desired specifications for the closed-loop the vector with the parameter of the controller to be
voltage control for the buck converter are set to: Set- optimized, bmin and bmax are, respectively, the ine-
tling time tset 100ms and overshoot SP 5% . qualities relative to lower and higher limits of the
closed-loop system. The matrix A corresponds to
open-loop plant coefficients.
3.2 Robust Control Technique
Considering the case where the plant, defined in
3.3 Robust Control Design by linear programming
(9), is subject to parameter uncertainty, the transfer
function plant become an interval transfer function as According to the uncertainties of the plant de-
follows: fined in Table 1, the transfer function in (4) become
n
i an interval transfer function as shown belong,
n( s ) ni s
G( s) i 0
V (s) n
G(s) c 2 o
n
i
d (s) di s (17)
i 0 D( s) s d1s d o
ni ni ni , d i d i d i , i no no no (23)
The characteristic polynomial family of the
d1 d1 d1 , d o d o d o
closed-loop system is determined by equation (11).
In order to guarantee the robustness of the The values of the desired interval polynomial in
closed-loop system against uncertainties, it is consid- closed-loop are obtained by varying the performance
ered that the parameters of the desired polynomial, established for the classical controller as follows,
described by equation (12), assume values such that t set 100ms 25%
their coefficients belong to an interval, as shows in (24)
equation (18), delimiting a desired region for the SP 5% 25%
robust pole-placement. Thus, the desired polynomial According to equation (24), the interval values
in closed-loop corresponds to an interval polynomial of damping coefficients and natural frequency are
family, which includes the desired nominal polyno- d 0.52,0.86 and n 43.47,72.45 , respectively.
mial. The adoption of interval values theses parameters
nr
( s) nr s n r 1s n r 1 o allow increase the damping of dominant oscillatory
(18) mode.
i i i , i The controller structure and the closed-loop de-
To satisfy the stability and performance condi- sired polynomial are shown in equations (25) and
tions, it is necessary and sufficient to satisfy the con- (26), respectively.
dition shown in equation (19). 2
K d s K p s Ki a2 s 2 a1s ao
( s) d F ( s) (19) C (s) (25)
s s b2 s 2 b1
According to Keel & Bhattacharyya (1997) and
Equation (19), for the robust controller design is for-
mulated a linear inequalities set, which restricted the
plant and desired polynomial coefficients in the de-
fined intervals, as shown in Eq. (20). Thus, the
XXII Congresso Brasileiro de Automática – CBA2018, Brazil, 2018
( s ) 4 s 3 s 2 s 1 s o
4 3 2
i i i , i 0,1, 2, 3, 4
s a s b
(26)
( s ) s 2n s n
2 2
, n n n
where a and b are non-dominant poles of the sys-
tems, therefore their characteristic do not influence in
the system behavior. The location of these non-
dominant poles is at the discretion of designer.
4 1
3 203.75, 339.58
2 2.04, 3.41 10
5 Figure 4. Load variation over time.
(27)
1 1.57, 2.62 10
5
o 4.41,10.68 10
8
5 Conclusion