Advanced Controllers for Quadratic Boost

Converter - A Case Study

J. Vijay Prabhu, M.Ashok Bhupathi Kumar, P. Damodharan, K.Vijayakumar
Department of ECE
IIITDM Kancheepuram, Chennai 600127, India.

Abstract—In this paper, model predictive control and sliding reverse recovery loss in the output diode. The non-minimum
mode control for quadratic boost converter is studied for its phase characteristics of boost converter due to the presence of
applications in grid connected solar photo voltaic systems. In a right half plane (RHP) zero results in high-bandwidth control
model predictive control, the control problem considered is
a constrained optimization problem. The objective function is design. The narrower conversion ratios results in low voltage
formulated for reference tracking and steady-state performance gain and lower power density which makes the boost converter
of quadratic boost converter. The optimal control law is a inefficient for high-voltage or high-power applications [3].
piecewise affine function obtained by satisfying the constraints Many boost derived topologies are discussed in literature to
of the quadratic boost converter. In sliding mode control, a overcome these disadvantages. The boost derived topologies
fixed frequency assisted sliding mode control law is designed
which comprises of voltage and current loop. The current loop like multistage DC-DC boost converters can be realized by
is controlled by sliding mode control, whereas the voltage loop cascading or interleaving conventional boost converter mod-
is controlled by the traditional proportional integral controller. ules. The voltage gain of multistage structures are linear
Simulation is carried out in MATLAB for a 90 W quadratic boost or exponential. Quadratic boost converter (QBC) is a multi-
converter. The steady-state and reference tracking performance stage topology which is a cascade connection of two boost
of the converter is analysed and the results are presented.
converters. The circuit complexity of QBC can be reduced by
Index Terms—model predicitve control, sliding mode control,
quadratic boost converter, multi-parametric programming integrating the switches of the cascaded boost converter into
one switch. For low-power application where voltage gain is
I. I NTRODUCTION limited, QBC can operate with wider conversion ratios when
compared to pulse-width modulated (PWM) boost DC-DC
Renewable energy sector in India over the years, supports converters [4].
the government plan of sustainable growth and emerged as The control schemes discussed in literature can be classified
an integral part to meet the nation’s energy needs in the into two; the linear and the nonlinear controllers. Also, based
grid connected power generation capacity. Renewable energy on the mathematical model of the plant used, controllers are
contributes to about 12.96 percent in the national electricity designed using the averaged and the non-averaged model.
installed capacity. Renewable energy based decentralized and The tuning of proportional integral (PI) controller use the
distributed applications benefits millions of people in Indian linearized state-space averaged model of the converter [5]. The
villages and other energy needs in an environment friendly PI controller is designed by selecting a cross-over frequency
manner. India voluntarily committed to reduce emission inten- smaller than the switching frequency and a desired phase
sity of its GDP by 20 to 25 percent within 2020, contributing margin of 45 to 60 degrees are used. The step response of this
to the climate change mitigation [1]. This enables the power trivial design shows overshoot up to 5 percent and a settling
electronics technology for integrating more renewable energy time of few switching periods in the range 5 to 30. Moreover,
into the grid, like solar and wind. The grid-connected photo PI controller can be applied only to single-input, single-output
voltaic (PV) power plants which varies in size from few (SISO) unconstrained problems. Optimal control input is not
hundreds of watts to megawatts bridges the gap in demand possible and mechanical saturation of control input results in
and supply of power. The overall efficiency of grid-connected destabilization of systems [6].
technology currently available is in the range of 93 to 95%.
To overcome the limitations of classic PI controller, control
Transformerless PV technologies are used to achieve high
strategies like linear quadratic regulator (LQR) is discussed in,
efficiency [2] and DC-DC converters are integrated to these
[7]. In LQR, the discrete-time averaged model is linearized
inverter topologies to maintain the required DC-link voltage.
locally with an outer estimation loop required to employ
DC-DC converters are mostly used to transfer unidirectional an integrator. LQR is applicable to multiple-input, multiple-
power and maintain the DC-link voltage at the desired level. output (MIMO) systems without using any decoupling. The
In these converters the power flow is from source to load disadvantages of LQR are the model is locally linearized, so
because of unidirectional switches like power MOSFETs and the operating range is limited and constraints are not handled.
diodes are used. The conventional boost converter which is Nonlinear controllers like fuzzy logic [8] and feedforward
used to maintain the required DC-link voltage has many control [9] uses the averaged model of the converter assuming
disadvantages like low efficiency due to hard switching and

978-1-7281-0419-5/19/$31.00 © 2019 IEEE

a lossless converter which is an ideal condition. The H∞
controller discussed in [10] is a nonlinear controller used
to control power converters and the closed-loop stability is
verified using the Lyapunov function. The limitations of linear
and non-linear controllers are difficulty in controller design,
tuning and robustness to load parameter variations. This en-
ables the use of advanced controllers like model predictive
control (MPC) and sliding mode control (SMC).
MPC implemented successfully in process industries is an
ideal candidate to control the DC-DC converter topologies and
several algorithms are proposed in recent years [11]- [13]. Fig. 1. Topology of the Quadratic Boost converter.
MPC in its simplest form is a dead-beat controller used in
controlling the basic DC-DC converter topologies discussed
inductors L1 and L2 are used to charge and discharge energy in
in [14] and for buck converter complex MPC strategy is
the continuous conduction mode (CCM). The filter capacitors
introduced in [15]. More complex MPC strategies are dis-
are C1 and C2 , where capacitor C2 and load resistor R are
cussed in [16], [17] for the boost converter and in [18] for
connected in parallel. QBC operates in two modes which
the quadratic boost converter. For real time implementation
depends on the position of the switch–S is ON, mode 1 and S
of MPC an explicit model predictive control is proposed in
is OFF, mode 2. The voltage gain of QBC is given by V0 /Vin
[19] which calculates the control law off-line. The control
= 1/(1-D)2 .
law is calculated by minimizing or maximizing the formulated
objective function subject to linearized model of the converter
B. Continuous-Time State-Space Model
and constraints on input, states and output variables. The
control law is a piece-wise affine (PWA) function which is In Mode 1, as shown in Fig. 2 when the switch S is ON (S
evaluated online, since the values of PWA function are stored =1), the inductor voltages across L1 and L2 and the capacitor
as a lookup table in the hardware. This reduces the memory currents through C1 and C2 are given by (1)-(2) and (3)-(4)
requirement and the computational time required to solve the respectively,
control problem.
diL1
Due to posterior constraints of classical control schemes, VL1 = L1 = Vin (1)
controller research is much focused on non-linear controllers dt
such as hysteresis controller, one cycle controller and sliding diL2
mode controller. The sliding mode controller (SMC) grabs V L 2 = L2 = VC1 (2)
dt
much attention due to its variable structure system, sen-
dVC1
sitiveness to the parameter variation and good stability in iC1 = C1 = −iL2 (3)
wide variation around input and output variables. The major dt
constraint associated with the SMC is its infinite switching dVC2 −V0
iC2 = C2 = (4)
frequency which is practically unattainable. Hence, in order dt R
to overcome this practical difficulty and realize a SMC with
In Mode 2, as shown in Fig. 3 when the switch is OFF
performance as close as real controller, certain techniques
(S=0), the inductor voltages across L1 and L2 and the capacitor
have to be employed. One among the techniques is to use
currents through C1 and C2 are given by (5)-(6) and (7)-(8)
hysteresis assisted SMC which is proposed in [20]. This
respectively,
method generates variable switching frequency which in turn
increases the switching losses and impacts the active and pas-
sive components. To overcome the aforementioned problems,
a fixed frequency assisted sliding mode control law is designed
which comprises of voltage and current loop. The current loop
is controlled by SMC, whereas voltage loop is traditional PI
controller. The current mode control is more advantageous
over voltage mode control in terms of overload protection,
better stability and transient response.

II. C ONVERTER M ODELING

A. Topology Fig. 2. Mode 1 operation of QBC.
The circuit topology of QBC is shown in Fig. 1. The boost
derived converter has four power semiconductor switches, diL1
one MOSFET switch S, and three diodes D1 , D2 , D3 . The VL1 = L1 = Vin − VC1 (5)
dt
 kept at a minimum and maximum of 0 A and 5 A respectively.
The inductor current constraint defined with a maximum value
protects the QBC from overcurrent and avoid the saturation of
the inductor. The discrete time small-signal model of QBC
with sampling time Ts = 20 µs is given by (10),
)
xk+1 = Ae xk + Be uk
(10)
yk = C xek + De uk

where A ∈ Rn×n is the state matrix, B ∈ Rn×m is the input

Fig. 3. Mode 2 operation of QBC. matrix, C ∈ Rq×n is the output matrix, x ∈ Rn is the vector
of states, u ∈ Rm is the vector of control inputs and y ∈ Rq
diL2 is the vector of outputs. For the simulation parameters given
VL2 = L2 = VC1 − V0 (6) in Table I, the matrices A,B,C, and D are,
dt
dVC1
 
iC1 = C1 = iL1 − iL2 (7) −0.153 0.0347 0.0026 −0.0007
dt  0.0093 −0.0212 −0.0016 0.0004 
A=  −0.0529 0.1205 −0.0339 0.0093 

dVC2 V0
iC2 = C2 = iL2 − (8) 0.0147 −0.0335 0.0093 −0.0025
dt R    
C. State-Space Averaged Small-Signal Model of QBC 5.179 0
 1.8837     0 
The state-space averaged small-signal model [21] is ob- B=  159.5211  , C = 0 0 0 1 , D =  0 
  
tained by assigning the state variables as the inductor currents 640.1379 0
and capacitor voltages iL1 , iL2 ,VC1 ,VC2 respectively. Perturbing
and linearizing the input voltage Vin , duty cycle D and state
TABLE I
variables gives the state-space small-signal model of the form, S IMULATION PARAMETER VALUES
e˙ = Ae
x x + Be
u where A and B are the state and input matrices
respectively. Parameters Values
The small-signal model is obtained by neglecting the higher
order non-linear AC terms and is expressed in (9), Input Voltage, Vin 40 V
  −(1−D)

Output Voltage, V0 300 V
ei˙ L1 0 0 0

L1  
   eiL1

˙     Inductor, L1 220 µH
 eiL2   0 1 −(1−D)  
0 L2 L2  eiL2 

  
= Inductor, L2 820 µH
+
 
 e˙   (1−D)
 
V C1   −1
0 0
 Ve 
C1  Capacitor, C1 = C2 11 µF
C1 C1
   
˙   
 Ve
V C2

e C2 Resistor, R 1 kΩ
(1−D) −1
0 C2 0 C2 R
Switching Frequency, fs 50 kHz
 Vin 1 
L1 (1−D) L1
 
Vin
 " #
 L2 (1−D)2
 0 de A. Model Predictive Controller Design

 −Vin

 e (9)
0 V The polytopic state and input constraints for the discrete
 in

 C1 R(1−D)4
  time system in (10) are defined as xk ∈ X, uk ∈ U, where
−Vin
C2 R(1−D)3 0 X ⊆ Rn , U ⊆ Rm are polyhedra [22]. For the constraints
on state and input variables the corresponding half-space
representation are expressed as in (11),

III. C ONTROL P ROBLEM F ORMULATION X = { x| Hx 6 K} , U = { u| Lu 6 M } (11)

The control problem is formulated as a constrained opti- where H, L are the identity matrices and
mization problem. The hard constraint defined as inequality   K, M are the

I I
for duty cycle is kept at minimum and maximum of 0 and constraint matrices defined by, H = , L= ,
−I −I
1 respectively. The hard constraint defined as inequality for    
xmax umax
output voltage is kept at minimum and maximum of 0 and K= , M=
−xmin −umin
300 V respectively. The soft constraints on inductor current is
 The objective function is chosen as quadratic to provide a u ∈ U = Hu uk 6 Ku ,
smooth control action. For quadratic objective function and    
linear constraints, the MPC is formulated as in (12), Hx 0 Kx
x̃k 6
0 Hu Ku
N
X −1 | {z } | {z }
xk T Qx xk + uk T Qu uk H̃∆u K̃∆u


min (12)
k=0
∆umin 6 ∆u 6 ∆umax
subj.to. xk+1 = Axk + Buk
For the linear state-space model the multi-parametric
yk = Cxk + Duk
 quadratic problem (mpQP) with norm-2 cost function is given
x0 = x(t)
 by (15),  
xk ∈ X 1 T
min U HU (15)
uk ∈ U

 U 2

where Qx and Qu are the cost matrices. If the output equation subj.to. GU 6 W + Sx0
is assumed to be linear, i.e. y = Cx + Du , the reference
tracking MPC problem is formulated as in (13), with H = H T  0, is a positive definite matrix, U is a vector
of optimization variables, x is a vector of parameters. The
N
X −1   matrices are defined as G ∈ Rm×s , W ∈ Rm , S ∈ Rm×n ,
min kQy (yk − yref )kp + kQu uk kp (13) where m represents the number of rows in constraints, s
k=0 dimension of optimization variables and n dimension of the
subj.to. xk+1 = Axk + Buk parameters. To solve the mpQP the parametric techniques are
discussed in [22] and the solution of mpQP is the optimized
yk = Cxk + Duk control law U* given as a piece-wise affine function derived

x0 = x(t) with state variables as initial conditions.
 The closed-loop implementation of MPC is carried out using
xk ∈ X 

the Receding Horizon Principle (RHC) which ensures closed-
yk ∈ Y   loop stability. The control law is computed for a finite horizon,

uk ∈ U

N=30 and apply only the first element u0 of the control law
sequence U*. Many methods are discussed in literature to
where, X ⊆ Rn , Y ⊆ Rq , U ⊆ Rm are polyhedrons.
compute the cost function value, but there are no standards.
The minimization of inputs in the objective function will The output cost parameter is divided into grids and for each
force the input to zero which results in output not reaching the grid space step response is obtained with initial conditions.
reference. With the formulation in (13), there exists a steady- The computational cost of MPC are chosen as Qy = 100 and
state offset and to provide offset-free tracking the optimization Q∆u = 1 respectively, if the response converges and has less
problem is reformulated using ∆u formulation. The control or no oscillations.
inputs are minimized over their increments. i.e. ∆u = uk −
uk−1 . The reformulated reference tracking MPC problem is B. Sliding Mode Controller Design
given as in (14),
N −1 
Basically, the SMC consists of two phases, one to bring
the control signals to the defined sliding surface which is
X 
min kQy (yk − yref )kp + kQ∆ u ∆uk kp (14)
k=0
known as reaching phase and the second phase is to maintain
        the equivalent control law onto the sliding surface known as
xk+1 A B xk B sliding phase. In this paper, the input inductor error is chosen
subj.to. = + ∆uk
uk 0 I uk−1 I |{z} as sliding surface to track the output voltage by maintaining
| {z } | {z } | {z } | {z } ũk
x̃k+1 Ã x̃k B̃
the input inductor in specified limits. The chosen sliding
  surface is given as,
  xk
yk = C D + |{z}
D ∆uk
| {z } uk−1 S = iL1 − iref (t) (16)
C̃ D̃
  Where the iref represents the reference current of input inductor
x(t) which is generated with the help of PI controller in the voltage
x̃t =
u(t − 1) control loop and iL1 is of QBC input current. The control law
H̃∆u x̃k 6 K̃∆u , H̃y ỹk 6 K̃y , of equivalent control is devised by differentiating the sliding
surface and equating it to zero. The main objective of defined
where, sliding surface is to maintain the reference current and input
x ∈ X = Hx xk 6 Kx , inductor current same by making the current error as zero
with exponential decay. The control law of equivalent control IV. S IMULATION R ESULTS
is given as,
A. Steady-State Performance of MPC and SMC
1 diref
Deq =1− (vi − L1 ) (17) The simulation of MPC is performed by using the open
vC1 dt source Multi-Parametric Toolbox 3.0 (MPT 3.0) [23], which
This equivalent control is acting as sliding phase to maintain runs on MATLAB platform and is a standard tool for opti-
the error as zero and to reach the sliding surface a curbing mization and parametric programming. The simulation results
control is defined as given in (18) shown in Fig. 4 demonstrates the steady-state performance of
the MPC controlled QBC. For an output voltage reference of
Dc = KS + δ sign(S) (18) 300 V, the response shows no overshoot with settling time
In order to generate the fixed frequency from the sliding mode of 20 µs. The sliding mode controller is simulated using
controller using pulse width modulation the combination of MATLAB software and the result is shown in Fig. 5, which
equivalent control and curbing control must be compared with shows no overshoot with settling time of 10 ms.
ramp generator. So, the addition of equivalent control and
curbing control should be in the range of zero and one and 400
the final control law is given as,
 300

1
 Deq + Dc ≥ 1
200
D = Deq + Dc 0 < Deq + Dc < 1 (19)
100

0 Deq + Dc ≤ 0

0
1) Stability Condition: The lyapunov function is used to 0 0.5 1 1.5 2 2.5
check the stability condition of the derived control law which 10-4

is given in (20)
1 Fig. 4. Steady-state performance of MPC.
V = S2 (20)
2
The time derivative of lyapunov function needs to be negative 400
definite for making the proposed control law as asymptotically
300
stable and its time derivative is given as,
 
vi vC1 diref 200
V̇ = S Ṡ = S − (1 − Dn ) − (21)
L1 L1 dt
100
Substituting the (18) in (21),
0
  0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
vi vC1 diref
V̇ = S − (1 − (KS + δ sign(S))) − <0
L1 L1 dt
(22) Fig. 5. Steady-state performance of Sliding Mode Control.
The stability condition gets satisfied by making the K > 0
and δ > 0.
B. Reference Tracking of MPC and SMC
2) Existence Condition: This existence condition deter-
mines the reaching ability of derived control law onto the The controllers are tested for step-up and step-down changes
designed sliding surface. The equation 8 and 9 together in output voltage reference. The MPC performance for the step
represents the reachability condition which are given as, change in output voltage reference is shown in Fig. 6. At time
t = 100 µs, the output voltage reference changes from V0,ref =
lim Ṡ < 0 (23) 300 V to 240 V.
Ṡ + →0
340
lim Ṡ > 0 (24) V0
Ṡ + →0 320 Vref

300
Substituting (16) in (23) and (24)
280
diref vi
− <0 (25) 260
dt L1
240
diref vi vC1
− + >0 (26) 220
0 0.5 1 1.5 2 2.5
dt L1 L1
10-4
With the help of (25) and (26) it can be concluded that the
existence conditions are satisfied. Fig. 6. Reference Tracking of MPC.
 [6] Tobias Geyer, Georgios Papafotiou, and Manfred Morari, “On the
optimal control of switch-mode dc-dc converters,” 2004 International
Workshop on Hybrid Systems: Computation and Control, Philadelphia,
PA, USA, 2004, pp. 342-356.
[7] R. Priewasser, M. Agostinelli, C. Unterrieder, S. Marsili and M. Huemer,
“Modeling, Control, and Implementation of DC–DC Converters for Vari-
able Frequency Operation,” IEEE Transactions on Power Electronics,
vol. 29, no. 1, pp. 287-301, Jan. 2014.
[8] T. Gupta, R. R. Boudreaux, R. M. Nelms, and J. Y. Hung, “Implemen-
tation of a fuzzy controller for dc-dc converters using an inexpensive
8-bit microcontroller,” IEEE Transactions on Industrial Electronics,vol.
44, no. 5, pp. 661-669, Oct. 1997.
[9] M. K. Kazimierczuk and L. A. Starman, “Dynamic performance of
Fig. 7. Reference Tracking of SMC.
PWM dc/dc boost converter with input voltage feedforward control,”
IEEE Transactions on Circuits and Systems I: Fundamental Theory and
Applications,vol. 46, no. 12, pp. 1473-1481, Dec. 1999.
The response shows MPC tracks the change in output [10] A. Kugi and K. Schlacher, “Nonlinear H∞ controller design for a dc-to-
dc power converter,” IEEE Transactions on Control Systems Technology,
voltage reference with no overshoot and with settling time of vol.7, no.2, pp.230-237, April 1999.
20 µs. At time t = 200 µs, the output voltage reference changes [11] J.M.Maciejowski, Predictive Control with Constraints.: Prentice Hall,
from V0,ref = 240 V to 300 V and MPC tracks the change in 2002.
[12] B. Stellato, T. Geyer and P. J. Goulart, “High-Speed Finite Control Set
reference. The reference tracking performance of sliding mode Model Predictive Control for Power Electronics,” IEEE Transactions on
controller is shown in Fig. 7. The output voltage changes from Power Electronics, vol. 32, no. 5, pp. 4007-4020, May 2017.
300 V to 240 V at time t = 200 ms and from 240 V to 300 [13] P. Karamanakos, T. Geyer and R. Kennel, “On the Choice of Norm
in Finite Control Set Model Predictive Control,” IEEE Transactions on
V at time t = 400 ms and the change in reference is tracked Power Electronics, vol. 33, no. 8, pp. 7105-7117, Aug. 2018.
with no overshoot. [14] Y. Qiu, H. Liu, and X. Chen, “Digital average current-mode control of
PWM dc-dc converters without current sensors,” IEEE Transactions on
V. C ONCLUSION Industrial Electronics,vol. 57, no. 5, pp. 1670-1677, May 2010.
[15] T. Geyer, G. Papafotiou, R. Frasca, and M. Morari, “Constrained optimal
For the QBC employing MPC as the controller, the steady- control of the step-down dc-dc converter,” IEEE Transactions on Power
Electronics,vol. 23, no. 5, pp. 2454-2464, Sept. 2008.
state and reference tracking simulation results shows the opti- [16] L. Cheng et al., “Model Predictive Control for DC–DC Boost Converters
mal performance. The output voltage is tracked for changing With Reduced-Prediction Horizon and Constant Switching Frequency,”
reference satisfying the constraints. To reduce the model IEEE Transactions on Power Electronics, vol. 33, no. 10, pp. 9064-9075,
Oct. 2018.
mismatch and better performance the non-linearity of QBC [17] A. G. Beccuti, S. Mariethoz, S. Cliquennois, S. Wang, and M. Morari,
can be modelled using hybrid systems modeling approaches “Explicit model predictive control of dc-dc switched-mode power sup-
like mixed integer quadratic programming (MIQP) and piece- plies with extended Kalman filtering,” IEEE Transactions on Industrial
Electronics, vol.56 , no.6 , pp. 1864-1874, June 2009.
wise affine (PWA) model. The MPC of constrained systems is [18] Vijay Prabhu J. and Damodharan P., “Explicit Model Predictive Control
nonlinear, which necessitates the use of Lyapunov stability the- of Quadratic Boost Converter for High Step-Up Applications,” 2018
ory to analyze stability. The nonlinear model of QBC is used IEEE International Conference on Power Electronics, Drives and Energy
Systems (PEDES), Chennai, India, 2018, pp. 1-5.
in sliding mode control and the simulation result shows the [19] A. G. Beccuti, M. Kvasnica, G. Papafotiou and M. Morari, “A Decen-
robust steady-state and reference tracking performance. The tralized Explicit Predictive Control Paradigm for Parallelized DC-DC
optimal and robust performance of MPC and SMC observed Circuits,”IEEE Transactions on Control Systems Technology, vol. 21,
no. 1, pp. 136-148, Jan. 2013.
in the simulation results suggests that the controllers can be [20] S. Oucheriah and L. Guo, “PWM-Based Adaptive Sliding-Mode Control
used in transformerless grid connected PV power plants. for Boost DC-DC Converters,” IEEE Transactions on Industrial Elec-
tronics, vol. 60, no. 8, pp. 3291-3294, Aug. 2013.
[21] Robert W. Erickson and Dragan Maksimovic, Fundamentals of Power
R EFERENCES Electronics.: Springer Science Business Media, 2007.
[1] Ministry of New and Renewable Energy, “Annual Report 2014- [22] Francesco Borrelli, Constrained Optimal Control of Linear and Hybrid
2015,” MNRE,Govt. of India, Feb.15-17,2015. Accessed on: Mar. Systems.: Springer, 2003, vol. 290.
15,2019.[Online].Available:https://mnre.gov.in/file-manager/annual- [23] M. Herceg, M. Kvasnica, C. N. Jones and M. Morari, “Multi-Parametric
report/2014-2015/EN/content.htm Toolbox 3.0,” 2013 European Control Conference (ECC), Zurich, 2013,
[2] W. Li, Y. Gu, H. Luo, W. Cui, X. He and C. Xia, “Topology Review and pp. 502-510.
Derivation Methodology of Single-Phase Transformerless Photovoltaic
Inverters for Leakage Current Suppression,” IEEE Transactions on
Industrial Electronics, vol. 62, no. 7, pp. 4537-4551, July 2015.
[3] M. Forouzesh, Y. Shen, K. Yari, Y. P. Siwakoti, and F. Blaabjerg, “High-
efficiency high step-up dc–dc converter with dual coupled inductors
for grid-connected photovoltaic systems,” IEEE Transactions on Power
Electronics, vol. 33, no. 7, pp. 5967-5982, July 2018.
[4] Oswaldo Lopez-Santos, Luis Martinez-Salamero, Germain Garcia, Hugo
Valderrama- Blavi, and David Alejandro Zambrano-Prada, “Steady-state
analysis of inductor conduction modes in the quadratic boost converter,”
IEEE Transactions on Power Electronics,vol. 32, no. 3, pp. 2253-2264,
March 2017.
[5] A. Urtasun and D. D. Lu, “Control of a Single-Switch Two-Input
Buck Converter for MPPT of Two PV Strings,” IEEE Transactions on
Industrial Electronics, vol. 62, no. 11, pp. 7051-7060, Nov. 2015.