15th international conference on Sciences and Techniques of Automatic control 1

& computer engineering - STA'2014, Hammamet, Tunisia, December 21-23, 2014 STA'2014-PID3359-PEC

Modelling and Model Predictive Control of a

DC-DC Boost Converter
S. Bououden, O. Hazil, S. Filali, and M. Chadli.

transient response time requirements are concerned, due to the

Abstract—In this paper we develop a model predictive highly non-linear characteristics of the DC-DC converters.
control (MPC) approach for regulating DC-DC Boost Obtaining an accurate model for many systems is difficult
converters. The proposed control strategy is implemented when they actually consist of many different operating modes
and tested using two models: an averaged non-linear which are transferred to each other based on logics, conditions
model for control purposes and a switched Buck-Boost or rules. Such systems are so called hybrid systems. A hybrid
circuit model as the controlled plant. The main objective of system consists of several dynamical equations representing
the paper is to design a Fuzzy Predictive Control to the system behavior at a certain operating condition(s). The
achieve desired voltage output without increasing DC-DC converter system can be easily modeled by hybrid
complexity of the hybrid model of DC-DC converter in modeling approaches [3-7]. These models can describe hybrid
various conditions. The proposed algorithm of fuzzy model systems perfectly and they can reduce the effects of
predictive controller is done by simulation using nonlinearity raised by the hybrid nature of the system. In this
MATLAB. Comparative analysis and results of FMPC and paper a step -down DC -DC converter is considered as a
MPC control is observed hybrid system and therefore it is modeled based on an existing
method for such systems, namely Mixed Logical Dynamical
model (MLD) [8].
Index Terms— Boost converter, averaged model, Takagi- However, it is very important to utilize a control method
Sugeno fuzzy model, Predictive controller. which coincides with such models. As a result of this fact,
there is much interest in developing more intelligent and
robust control structures. In [9] a fuzzy logic controller
I. INTRODUCTION scheme was proposed for controlling a DC-DC converter. In
the proposed scheme a genetic algorithm was used to enhance
S ome of the most important circuits within the family of
power circuits are the DC-DC converters. They are
extensively used in power supplies for electronic
the output quality of a PID fuzzy controller. Eduardo Mojica-
Nava and Nicanor Quijano [10] proposed a weighted control
equipment to control the energy flow between two DC technique that uses the replicator dynamics concepts to weight
systems (e.g. well-regulated DC-to-DC power converters are the operation of different controllers tuned to operate in
critical for mission success on space platforms). The control different modes. They applied their scheme to a step-down DC
objective of such devices is to maintain regulation of the - DC converter and showed its performance through
output voltage at the desired value [1],[2]. simulations. In [11], author derived a family of PI controllers
Typical control system configurations for power circuits with parameters that depend nonlinearly on the steady - state
include open-loop as well as closed -loop control strategies. In duty cycle. He used the Ziegler -Nichols method in the PI
both cases, PI controllers are utilized. The controller must controller around a constant operating equilibrium point. He
keep the DC -DC converter within a certain percentage of the applied the proposed scheme to a boost and a buck -boost
specified nominal operating point in the presence of converter. An unconstrained nonlinear predictive controller is
disturbances and modeling errors. Unfortunately, PI control formulated for a DC-DC converter in [12].
does not always fulfill the above mentioned control They used a control methodology that extends the concept of
specifications, especially when disturbance rejection and Generalized Predictive Control to nonlinear systems. Finally,
Tobias Geyer et. al. proposed a novel modeling and controller
design approach based on the hybrid system [1]. They
S. Bououden is with the Faculty of sciences and technology, University of proposed a hybrid model of system for the whole operating
Abbes Laghrour Khenchela, and Laboratory of automatic and robotic, range and designed a model predictive controller based on this
University Constantine1, Algeria; e-mail: [email protected]). model. They showed the advantages of their proposed scheme
S. Filali is with the Laboratory of automatic and robotic, University
Constantine1, Algeria. (e-mail: [email protected]).
using simulations.
O. Hazil is with the Laboratory of automatic and robotic, University This paper proposes a new control strategy for DC-DC
Constantine1, Algeria. . (e-mail:[email protected]). converters, in which the non-linear predictive control
M. Chadli, was with University of Picardie Jules Verne, MIS (EA 4029)33 rue approach [4] is used to calculate the control input. The non-
Saint-Leu, 80039 Amiens, France (e-mail: [email protected]).
linear characteristics of the controlled plant are explicitly
taken into account when the control input is calculated and
.

978-1-4799-5907-5/14/$31.00 ©2014 IEEE 643

Authorized licensed use limited to: Indian Institute of Technology (Ropar). Downloaded on October 05,2024 at 17:52:43 UTC from IEEE Xplore. Restrictions apply.
 2

thus, better performance and robustness can be achieved. The of Ton/( Ton +Toff) is the duty cycled d(t). Duty cycle is
fuzzy predictive control algorithm uses an averaged non-linear compared with a sawtooth signal v(t) of amplitude equals to 1.
model to predict the future behaviour of the switched Simulink Consequently 0”d”1.
Boost converter, which is the plant to be controlled. So, the studied converter has two working topologies
corresponding to its switch states. The first topology (Fig. 3-a)
The rest of the paper is as follows: In section 2 the corresponds to the on state of the switch (d=0) during a period
mathematical model of the converter are introduced. Section 3 fraction dTs the system can be presented by the following
presents the T-S fuzzy model of the DC-DC converter. Next, equations:
a predictive controller is designed and applied to the system in diL 1
= vin (1)
Section 4. Section 5 present simulation results illustrating the dt L
performance of the proposed control approach. Finally, dvC 1
Section 6 concludes the paper. =− vC (2)
dt CR
vs = vC (3)
II. AVERAGED MODEL OF BASIC PWM BOOST CONVERTER The second topology, presented by the (Fig. 3-b) corresponds
to the off state of the switch (d=1) during the rest of the
A. State-space of the boost converter sampling period d’Ts, for this case, the diode conducts, and the
dynamical equations of the converter are expressed as
This section shows the state-space averaged model of the diL 1 1 (4)
= vin − vC
boost converter, which includes the PWM (Pulse Width dt L L
Modulation) and has as control input the duty ratio d. The DC- dvC 1 1
DC converter can operate in both continuous conduction mode = iL − vC (5)
dt C CR
(CCM) and discontinuous conduction mode (DCM) .We
assume that the converter operates in continuous conduction vs = vC (6)
mode (CCM) and that the inductor current is always larger iL iL
than zero L
L
Figure 1 shows the schematic circuit diagram of a DC–DC vin vC
C
vs vin vC C R vs
boost converter and the relevant control signals. R
iL
(a) (b)
L D
vin sw C Fig 3. Equivalent circuit of the boost converter
vC R vs (a) switch on. (b) switch off

Choosing the state vector as x = [iL , vC ]T where vc is the

PMW d (t )
capacitor voltage; iL is the inductor current, the general
Fig 1. Schematic of the boost converter
equation that governs the operation of the boost converter is:
­ x = Ai x + Bi
u b (t ) v (t ) ® (7)
¯ y = Ci x
Where: i=1 to the first configuration described in Figure 2-(a)
and: i=2 the second configuration shown in Figure 2-(b).
d (t ) Where:
Ton Toff
ª0 0 º ªv in / L º
A1 = « », B1 = « 0 », C1 = [0 1]
T ¬ 0 − 1 / RC ¼ ¬ ¼
Fig 2. Waveforms of the PWM process ª 0 −1 / L º ªv / L º
A2 = « » , B 2 = « in », C 2 = [0 1]
In Fig.1, vs is the output voltage and vin the line voltage. The ¬1 / C − 1 / RC ¼ ¬ 0 ¼
output voltage must be kept at a given constant vref value, the So the combination of state-space representation of mode 1
diode D is on inverse polarization, R models the converter (on mode) and mode 2 (off mode) induces the following linear
load, while C and L represent, respectively, capacitor and state space representation:
inductor values, the switch sw was a power transistor ­ x = Ax + Bu
controlled by a binary signal ub. ® (8)
The binary signal that triggers on and off the switches is ¯ y = Cx
controlled by a fixed-frequency pulse width modulation With: A = dA1 + d ' A2 ; B = dB1 + d ' B 2 ; C = dC1 + d ' C 2 ;And:
(PWM) circuit (Fig. 2). The constant switching frequency is d' = 1− d, u = d
1/Ts, with Ts the switching period is given by the sum of Ton This gives:
(when ub =1) and Toff (when ub=0) and the ratio

644

Authorized licensed use limited to: Indian Institute of Technology (Ropar). Downloaded on October 05,2024 at 17:52:43 UTC from IEEE Xplore. Restrictions apply.
 3

ª 0 − (1 − d ) / L º ªv / L º systems obtained from specific information about the original

A=« B = « in C = [0 1]
− 1 / RC »¼ »,
, nonlinear plant.
¬ (1 − d ) / C ¬ 0 ¼ The ith rule of fuzzy model for the nonlinear discrete-time
systems has the following form:
B. Discrete-time state-space model Ri:If z1(k) is about M1i and zq(k) is about Mqi
x(t ) = Ai x(t ) + Bi u(t )
Then ­® for i=1,…, r (18)
The MPC is to be fed with a discrete time model which is ¯ y(t ) = Ci x(t )
easily obtained from the continuous model assuming that the in which Mji is the fuzzy set of zi(k) in rule Ri, r is the number
switching period Ts is much smaller than the time constants of IF-THEN fuzzy rules and zi(k) are the decision variable
associated with the circuit, the following discrete-time model
assumed measurable, x( k ) ∈ R n is the system state vector,
is obtained using the forward Euler approximation, from the
continuous time version (8): y ( k ) ∈ R p is the measurable output,
The global dynamic system is inferred as follows:
x( k + 1) = ( I + T s A) x (k ) + Ts Bd ( k ) (9) x (t ) = ¦i=1 µ i ( z (t ))( Ai x(t ) + Bi u (t ))
r
(19)
Using this assumption, the discrete-time state-space model of
y (t ) = ¦i =1 µ i ( z (t ))(Ci x (t ) )
r
the boost converter can be written as: (20)
ª x1 (k +1) º ª 1 − (1− d (k))Ts / Lºª x1 (k) º ªTs vin (k) / Lº Where
«x (k +1)» = «(1− d (k))T / C +
1− (Ts / RC) »¼«¬x2 (k)»¼ «¬ » (10)
¬ 2 ¼ ¬ s 0 ¼ ∏ qj=1 M ji ( zi (t )) (21)
µi =
The system in (10) is a nonlinear system since there are
¦ (∏ )
r q
i =1 j =1 M ji ( zi (t ))
products of two input signals. It is therefore mandatory to
make some sort of linearization at some point [15].
We can rewrite the non linear model of boost converter as Mji(zi(t)) is the grade of membership of zi(t) in Mji, The
follows: normalized activation function µi(zi(t)) in relation with the ith
sub model is such that
­ x = f ( x, u )
¦i=1 µi ( z(t )) = 1 , 0 ≤ µi ( z (t )) ≤ 1 ∀ i = 1,..., r
r
® (11) (22)
¯ y = g ( x, u )
A nonlinear system may also be represented by sectors
Considering that the system variables consist of the following
[16].The following Lemma will be used in the sequel of the
two components:
paper.
x=ǻx+x0 (12)
Lemma: Let f(x(t)): RĺR is a bounded function (Į” f(x(t))”ȕ),
d=ǻd+d0 (13)
it always exist tow functions,Ș1(x(t)) and Ș2(x(t)) and two
Where x0 and d0 represent the equilibrium values and ǻx and
scalars Į and ȕ such that:
ǻd are the perturbed values of state and duty cycle, where the
superscript ǻ represents incremental variations over the
f(x(t))= Į Ș1(x(t)) +ȕ Ș2(x(t)) (23)
equilibrium point.
With Ș1(x(t)) + Ș2(x(t))=1, Ș1(x(t))•0, Ș2(x(t))•0
Assume that the operating point is an equilibrium point which
And Į=max(f(x(t))), ȕ=max(f(x(t)))
must satisfy
x0 = f ( x0 , u0 ) = 0 (14)
f ( x(t )) − β α − f ( x(t ))
x = ∆x + x0 = f ([∆x + x0 ], [∆u + u0 ]) (15) η1 ( x(t )) = η 2 ( x(t )) =
α −β α −β
The linearized equations are:
­ ǻx(k+1 )=A′ǻx(k)+B′ǻu(k) We can rewrite the non linear model of boost converter as
® (16)
follows:
¯∆y(k)=C′∆x(k)
Where the matrices A’, B’ and C’ are given by: ª x1 (k + 1) º ª 1 Ts / L ºª x1 (k ) º ª Ts vC (k ) / L º
« x (k + 1)» = «− T / C 1 − (T / RC)»«x (k )» + «− T i (k ) / C»d (k )(24)
δf ; B′ =
δf ; C′ =
δg ¬ 2 ¼ ¬ s s ¼¬ 2 ¼ ¬ s L ¼
A′ = The obtained T-S fuzzy model represents exactly the nonlinear
δx x = x0
d = d0
δd x = x0
d =d0
δx x = x0
d = d0
system for x(k ) ∈ R n with 2l locals models where l represents
Finally we obtain the following system description: the number of local models.
ª∆x1(k +1)º ª 1 −(1−d0)T / Lºª∆x1(k)º ª(vin + x02)T / Lº
(17) Assuming that, I min ≤ i l (k ) ≤ I max and V min ≤ v c ( k ) ≤ V max
«∆x (k +1)» = «(1−d )T / C 1−T / RC »«∆x (k)» + « − x T /C »
¬ 2 ¼ ¬ 0 ¼¬ 2 ¼ ¬ 01 ¼ the system described by (24) is modeled with a four rules T-S
fuzzy system as follows:
III. T-S FUZZY MODEL OF BOOST CONVERTER R1:If il(k) is Imin and vc(k) is Vmin
Then: x(k+1)=A1x(k)+B1d(k)
R2:If il(k) is Imax and vc(k) is Vmin
The system (10) can be represented by linear subsystems Then: x(k+1)=A2x(k)+B2d(k)
called Takagi-Sugeno. The main feature of T-S fuzzy model is
to express nonlinear dynamics by means of a set of IF-THEN R3:If il(k) is Imin and vc(k) is Vmax
rules. The consequent parts of the rules are local linear Then: x(k+1)=A3x(k)+B3d(k)

645

Authorized licensed use limited to: Indian Institute of Technology (Ropar). Downloaded on October 05,2024 at 17:52:43 UTC from IEEE Xplore. Restrictions apply.
 4

R4:If il(k) is Imax and vc(k) is Vmax

Then: x(k+1)=A4x(k)+B4d(k) J the cost function to be minimized at each sampling period
penalizes deviations from the predicted output y(k+i/k) of a
The overall model of T-S fuzzy system can be given by the reference trajectory r(k+i/k), and variations of the control
following: vector ǻu(k)= u(k)- u(k-1), it is often given by the quadratic
form
x(k+1)=Ax(k)+Bu(k) (25) Hp H u −1 2

¦ ¦
2
With J (k ) = yˆ ( k + i k ) − r ( k + i k ) Q (i)
+ ∆u ( k + i k )
B = ¦i =1 µ i ( x(k )) Bi
4 i= H w i=0 R (i )
(26)
(28)
and Where Hu is the control horizon, Hw is first point of the
ª 1 − Ts / L º prediction horizon, Hp is last point of the prediction horizon,
A = A1 = A 2 = A3 = A 4 = «
¬T s / C 1 − ( T s / RC ) »¼ Hu < Hp, and ǻu(k+i/k)=0 if i<Hu. R(i), Q(i) are suitable
weighting matrixes.
ª TV /L º ª T sV min / L º The state-space representation allows us to write the predicted
B 1 = « s min », B2 = «− T I »,
¬ − T s I min / C ¼ ¬ s max / C ¼
output as follows
j
ª TV
B 3 = « s man
/L º
B4 =
ª T sV max / L º yˆ ( k + i k ) = C ( A j xˆ ( k k ) + ¦ A i −1 Bu ( k + j − i k ) (29)
», «− T I »
¬ − T s I min / C ¼ ¬ s max / C ¼
i =1

The membership function is such that: ‡ ƒ ”‡™”‹–‡ –Š‡ ‘†‡Ž ‘—–’—– ‘ˆ Ž‘ ƒŽ ’”‡†‹ –‹‘ ˆ‘”
ˆ—–—”‡‘‡–•‹ƒ–”‹šˆ‘”ǣ
− iL (k ) − I max iL (k ) − I min
µ1 = µ2 = Y (k ) = Ψx(k ) + Γu (k − 1) + Θ∆u (k ) (30)
I max − I min I max − I min
Š‡”‡
− vc (k ) − Vmax vc (k ) − Vmin ª CB º
µ3 = µ4 = ª CA º « »
Vmax − Vmin Vmax − Vmin « # » « H u −#1 »
« » «C ¦ Ai B »
Figure 4 shows the membership functions of the fuzzy model « CA H u » « i =0 »
« » « Hp »
Ψ=« # »Γ= «C Ai B »
« ¦
1
µ1 (iL (k )) µ 2 (iL ( k )) «CA H u +1 » »
« » «
i =0
»
« # » #
« H p −1 »
« CA H p » «C ¦ A i B »
¬ ¼
«¬ i = 0 »¼
0 ª CB " 0 º
I min I max
« C ( AB + B ) " CB »
Inductance current « »
1 « # % # »
« H u −1 i »
« C¦AB
µ 3 ( vC ( k )) µ 4 (vC ( k ))
" CB »
Θ=« i=0 »
« »
Hu

« C ¦ Ai B % C ( AB + B ) »
« i = 0# % # »
« H p −1 H p −Hu »
« C C ¦ Ai B »
0

«¬ ¦
Ai B "
Vmin
Capacitor voltage
Vmax
i=0 i=0
»¼
Fig 4: Membership functions of the T-S fuzzy model
The cost function (28) can be rewritten as:

IV. MODEL-BASED PREDICTIVE CONTROL J ( k ) = ε ( k ) T Q ε ( k ) − ∆u ( k ) T G + ∆u ( k ) T H ∆ u ( k ) (31)

Where:
The basic model to calculate the predictions is discrete state ε (k ) = r ( k ) − Ψ x( k ) − Γu (k − 1) (32)
model given by [9]:
G = 2Θ T Qε ( k ) (33)
­°x(k + 1) = Ax(k ) + Bu(k ) x(k)∈ Rn;u(k)∈ Rl (27) H = 2Θ T Q Θ + R (34)
®
°̄ y(k ) = Cx(k ) y(k)∈ Rm

646

Authorized licensed use limited to: Indian Institute of Technology (Ropar). Downloaded on October 05,2024 at 17:52:43 UTC from IEEE Xplore. Restrictions apply.
 5

Therefore the control law that minimizes the local cost MPC
35 Vref
function (28) is given by: Fuzzy MPC
1 −1 (35) 30
∆u ( k ) = H G
2
25

V. SIMULATIONS AND RESULTS 20

The performances of the proposed control design are

Vc
15
illustrated through simulations. The parameters nominal
values of the boost converter are illustrated in Table 1. 10
TABLE1
5
Parame
Description Numerical value
ters 0
L The inductance 200µH
C The capacitance 200µF -5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s)
R The load resistance 10ȍ
Vin Input voltage 10V Capacitor voltage vc
T The sampling 0.65ms
fs The frequency 200µHz 2
Imin Minimum current 0
Imax Maximum current 3
Vmin Minimum voltage 0 1.5
Vmax Maximum voltage 22
IL Inductor current

Figure 5 illustrates the transient simulation of the boost

0.5
converter with change in the reference voltage from 20V to
30V at time t=0.5 seconds. The waveforms depicted in the Fig
5 are the duty-cycle d capacitor voltage vc and inductor current 0
iL. The performances of the MPC controller fed with a
linearized system around equilibrium point of the converter
-0.5
are compared with the Fuzzy Model Predictive controller. We 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (s)
can notice for MPC controller fed with a linearized system
around equilibrium point that the output voltage settle to their Inductor current iL
desired value without any overshoot and the settling time of Fig 5: the transient simulation of the boost converter
the circuit as clearly seen in the graph is t=0.1 seconds. The
Fuzzy Model Predictive controller gives small overshoots and
the settling time of the circuit is much smaller. The best VI. CONCLUSION
performances are obtained for Hp=7, Hu=3 In this paper, we have presented a FMPC controller for the
boost converter operating in CCM mode. At first the non
0.4 linear state-space model is generated and linearized around
equilibrium point, and in second time we used the fuzzy model
0.35
based on sector nonlinearity concept, to approximate the
0.3 nonlinear model, this approximation allows us to obtain the
exact model of the system. The fuzzy model obtained was then
0.25
u(t) Duty cycle

used to design a MPC controller that ensures the output

0.2 voltage regulation with good performances.
0.15
REFERENCES
0.1
[1] H.Guldemir,” Modeling and Sliding Mode Control of Dc-Dc Buck-
Boost Converter”, 6th International Advanced Technologies Symposium
0.05
(IATS’11), 16 -18 May 2011, Elazõ÷, Turkey.
[2] A.AHMAD, K.ZHILIU And H.KINOSHITA, HIGH PERFORMANCE
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 ALGORITHMS FOR THE CONTROL OF BUCK DC-DC
Time (s) CONVERTERS’’, International Journal of Engineering Science and
Duty cycle d Technology Vol. 2(10), 2010, 5799-5812
[3] J.V. Inglés, P. Garcés, R. Leyva, “Robust LMI Control of a Buck-Boost
Converter with Low Ripple Propagation”, Preprints of the 2012 20th
Mediterranean Conference on Control & Automation (MED),
Barcelona, Spain, July 3-6, 2012

647

Authorized licensed use limited to: Indian Institute of Technology (Ropar). Downloaded on October 05,2024 at 17:52:43 UTC from IEEE Xplore. Restrictions apply.
 6

[4] C.E. Garc´ia, D.M. Prett, and M. Morari. “Model predictive control:
Theory and practice “, a survey. Automatica, 25:335–
348, 1989
[5] P. TATJEWSKI, M. NCZUK, “SOFT COMPUTING IN MODEL–
BASED PREDICTIVE CONTROL”, Int. J. Appl. Math. Comput. Sci.,
2006, Vol. 16, No. 1, 7–26.
[6] S. Bououden, M. Chadli, S. Filali, A. El Hajjaji, “Fuzzy Model Based
Multivariable Predictive Control of a Variable Speed Wind Turbine:
LMI approach,”Renewable Energy. Elsevier. vol. 37, no. 1. pp. 434-
439, 2012.
[7] G. L. Carneiro, R.K.H. Galvão, “MODEL BASED PREDICTIVE
CONTROL OF AN AIRCRAFT WITH ACTUATOR FAILURE IN A
TERRAIN FOLLOWING TASK”, 3rd CTA-DLR Workshop on Data
Analysis & Flight Control September 14-16, 2009, S. J. Campos, SP,
Brazil
[8] B. Hichem and M.Faouzi “Fast Nonlinear Model Predictive Control
using Second Order Valera Models Based Multi-agent Approach”
Institut Supérieur Des Etudes Technologiques de SFAX Ecole Nationale
d’ingénieur de Monastir Tunisia
[9] M. Lazar, B.J.P. Roset, W.P.M.H. Heemels, H. Nijmeijer, P.P.J. van den
Bosch, “INPUT-TO-STATE STABILIZING SUB-OPTIMAL
NONLINEAR MPC ALGORITHMS WITH AN APPLICATION TO
DC-DC CONVERTERS” , 2006 IFAC
[10] F. H. F. Leung, P. K. S. Tam, and C. K. Li, "The Control of Switching
DC-DC Converters – A General LQR Problem," IEEE Transactions on
Industrial Electronics, vol. 38, pp. 65–71, 1991.
[11] H. Sira-Ramirez, R. A. Perez-Moreno, R. Ortega, and M. Garcia-
Esteban, "Passivity-based controllers for the stabilization of DC-to-DC
power converters," Automatica, vol. 33, pp. 499-513, 1997.
[12] H. Sira-Ramirez, "On the Generalized PI Sliding Mode Control of DC-
to-DC Power Converters: A Tutorial," International Journal of Control,
vol. 76, pp. 1018–1033, 2003.
[13] T. Takagi and M. Sugeno. “Fuzzy identification of systems and its
applications to modeling and control”. IEEE Transactions on Systems,
Man, and Cybernetics, 15 :116–132, 1985
[14] K. Tanaka and H. O. Wang, “Fuzzy Control Systems Design and
Analysis: A Linear Matrix Inequality Approach”. New York: Wiley,
2001.
[15] K. Mahabir, G. Verghese, J. Thottuvelil, and A. Heyman, “Linear
averaged and sampled data models for large signal control of high
power factor converter,” in Proceedings of IEEE Power Electronics
Specialists Conference, pp. 372 – 381, San Antonio, Texas, 1990.
[16] S. Bououden, M. Chadli, S. Filali, A. El Hajjaji, (2012). “Fuzzy Model
Based Multivariable Predictive Control of a Variable Speed Wind
Turbine: LMI approach,” Renewable Energy. 37 (1) pp. 434-439.
[17] S. Bououden, O. Benelmir, S. Ziani, S. Filali, (2007).A new adaptive
fuzzy model and output terminal constraints in predictive control,
International Journal of information and systems sciences. 3 (1) 25–35.

648

Authorized licensed use limited to: Indian Institute of Technology (Ropar). Downloaded on October 05,2024 at 17:52:43 UTC from IEEE Xplore. Restrictions apply.