Received April 28, 2022, accepted May 12, 2022, date of publication May 27, 2022, date of current

version June 3, 2022.

Digital Object Identifier 10.1109/ACCESS.2022.3178425

Implementation of Exact Linearization Technique

for Modeling and Control of DC/DC Converters
in Rural PV Microgrid Application
RODRIGO ALIAGA1 , (Student Member, IEEE), MARCO RIVERA 1 , (Senior Member, IEEE),
PATRICK WHEELER 2 , (Fellow, IEEE), JAVIER MUÑOZ 1 , (Member, IEEE),
JAIME ROHTEN 3 , (Member, IEEE), FADIA SEBAALY4 , (Member, IEEE),
ARIEL VILLALÓN1 , (Graduate Student Member, IEEE), AND ANDREW TRENTIN 2
1 Department of Electrical Engineering, Universidad de Talca, Curicó 3340000, Chile
2 Department of Electrical and Electronic Engineering, University of Nottingham, Nottingham NG7 2RD, U.K.
3 Department of Electrical and Electronic Engineering, Universidad del Bío-Bío, Concepción 4051381, Chile
4 École de Technologie Supérieure, Université de Québec, Montréal, QC H3C 1K3, Canada

Corresponding author: Marco Rivera ([email protected])

This work was supported in part by the Agencia Nacional de Investigación y Desarrollo de Chile (ANID), Fondecyt, under Project
1191028; in part by the Fondap Solar Energy Research Center (SERC) Chile under Grant 15110019; in part by the British Embassy in
Chile, Fondecyt Regular, under Grant 1220556; and in part by CLIMAT AMSUD under Grant 210001.

ABSTRACT The inclusion of solar systems in rural microgrids is becoming increasingly important to supply
energy for irrigation, electric motors, lighting and other. This paper presents the implementation of an exact
linearization technique for the modeling and control of a DC/DC converter for use in a microgrid based
on a photovoltaic (PV) generation system where non-linear converters are used. The basic advantage of
this technique is in linearizing the converter model, thus allowing different operating points to be considered
under different conditions. This paper presents a general description of the implemented microgrid topology.
The exact linearization theory adapted for power converters is applied to both a Single-Ended Primary-
Inductor converter (SEPIC) to extract energy from PV modules and to a Boost converter to increase the
voltage. Experimental results are also presented to validate the effectiveness of the proposed control.

INDEX TERMS DC/AC, DC/DC, micro-inverter, renewable energy, rural-microgrid.

I. INTRODUCTION performance quantification of the control strategy applied.

A microgrid is a system that allows the interconnection Among those linear control strategies for the inner con-
of a variety of loads and different electrical sources such trol of a VSC in a microgrid, PI controllers and hystere-
as PV, wind and biodiesel. Microgrids can operate either sis control may be included [2]. In general, these linear
connected to the main electrical grid or isolated. In their control strategies for VSCs in microgrids have practical
operation, robustness must be guaranteed. Considering the limitations. For example the tuning of control parameters
nonlinearities of many loads and power converters, they are valid only for certain operating points, affecting the
must work under different operating conditions. In this microgrid stability under large-signal external disturbances,
context, the rural-microgrid concept implies the electri- [3]–[5]. Model-based control strategies appear to be an option
fication of places where there is no existing access to to address the previously mentioned limitations. One example
electricity [1]. of these strategies is predictive control, but these model-based
There are several control strategies for voltage source con- control schemes also have their disadvantages such as the
verters (VSCs) that may be used in microgrids. The tradi- need of knowing the system parameters [6]. Based on this
tional approach is known as voltage-oriented control (VOC) challenge, this work proposes an exact linearization tech-
which mainly consists of linear control loops. This gives nique [7]–[10] for the control of the power electronic convert-
simple and analytical control synthesis and well-known ers used for energy conversion in a microgrid. The benefits
of the exact linearization technique are easy implementation,
The associate editor coordinating the review of this manuscript and robustness and control stability across with a wide operating
approving it for publication was Pinjia Zhang . range.

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
VOLUME 10, 2022 56925
 R. Aliaga et al.: Implementation of Exact Linearization Technique for Modeling and Control of DC/DC Converters

FIGURE 1. Microgrid topology.

The energy source used in this work is solar PV, which nonlinear DC/DC converter. In this context, papers such as [7]
implies the use of DC/DC converters to raise or lower the validate the implementation of the exact linearization tech-
input voltage depending on the case. There are different nique applied to the control of a Buck-Boost converter. The
power converter topologies used in the conversion of PV exact linearization is applicable to PV systems (and other
energy, which can be divided into non-isolated and isolated energy sources) [21], where an algorithms is used to track
converters. In the first group there are converters such as the maximum power point, such as Perturb and Observe
Buck, Boost, Cuk, Sepic and Zeta [11]. In the case of isolated (P&O) [22] or Incremental Conductance (IC) [23], [24].
converters, the Flyback, Forward, Push Pull, Half-Bridge and In this implementation, the P&O algorithm is used with a
Full-Bridge converter [12] can be highlighted. A non-isolated modification, which consists of providing the current as a
converters use a smaller number of components than isolated reference to be followed by the control, instead of the PV
converters, which makes them more economical. However, module. The novelty of this work is in the application of the
they have a larger size compared to isolated converters [13]. exact linearization technique for a SEPIC and Boost converter
In terms of control, the most important objectives in PV in a simple and systematic way, obtaining the same models
converters are the current and voltage control, maximum thanks to the generality of the method. Together with the
power point tracking (MPPT), synchronization with the elec- exact linearization, the use of a power balance technique
trical network, power quality, anti-islanded protection, energy is proposed, with the aim that the converters can operate
storage and monitoring of PV modules [14]. Regarding the with different types of loads into the entire power converter’s
control techniques of the fundamental electrical variables operating region. The final result is reflected in the versatility
(voltag current), works such as [15]–[18] present the strate- and robustness of the algorithm for microgrids.
gies that allow meeting the objective of control and maxi- The motivation for the work presented in this paper is
mization of the use of photovoltaic energy. In this context, to include renewable energy in rural microgrids, where the
works such as [19] describe the use of a voltage-current energy may be required in AC or DC form at different volt-
cascade loop to control the energy extracted and injected age levels (12/24 V and 110/220 V). In the proposed topol-
into the electrical grid. Reference [20] compares the most ogy indicates that the SEPIC converter charges the battery,
used maximum power point tracking algorithms, concluding tracks the maximum power point (MPP) and supplies the
that the classical methods are more reliable under uniform Boost converter. The Boost converter is required increase
irradiance conditions, while the intelligent algorithms present the voltage to the needed AC level, Figure 1. Additionally,
a better performance under different irradiance conditions once the DC link voltage is well regulated by the Boost
thanks to the increased speed of tracking, sensing and data converter, the inverter can be used to supply AC loads. All
storage. of the power converters used in this paper are nonlinear, and
The SEPIC and Boost converters can be controlled by therefore the control is not necessarily an easy task. Thus,
the exact linearization technique. The idea behind exact lin- the exact linearization-based control is proposed to be able to
earization is to redefine the power converter’s input as a func- manipulate and control the variables in the entire operating
tion of variables and parameters to find a linear relationship region avoiding the intrinsic nonlinearities of these power
between a new input and the output. In this method is not converters. This paper is organized as follows: Section II pro-
required to consider the dynamics of other variables except vides a general description of the system; Section III details
those of interest, therefore no reduced model is needed. the SEPIC converter model with the linearization process;
The linearization can be found by two main ways: input Section IV presents the boost converter, its linearization,
states linearization and input-output linearization. The result internal and external loop models; Section V details the sys-
obtained with the exact linearization is a transfer function tem stability analysis; Section VI introduces the experimen-
that allows the use of a linear controller such as a simple tal results; Section VII presents the comparison and finally
PI controller, highly simplifying the task of controlling the section VIII concludes this work.

56926 VOLUME 10, 2022

R. Aliaga et al.: Implementation of Exact Linearization Technique for Modeling and Control of DC/DC Converters

II. SYSTEM DESCRIPTION

A. DESCRIPTION OF THE TOPOLOGY
The general topology is shown in Figure 1, and consists of
a SEPIC converter responsible for extracting the energy and
storing it in a battery bank. Then, a Boost converter is used
to rais the voltage from 12/24 V to a reference voltage for the
DC link. Thus, this topology allows the connection of both,
FIGURE 3. SEPIC converter.
DC and AC loads. Finally, by using a full-bridge inverter, the
conversion is made to supply AC energy. The inverter [25] The SEPIC converter can reduce or increase the voltage,
output is connected to an LCL [26] filter to obtain a sinusoidal which allows different voltage levels at the maximum power
waveform with a lower total harmonic distortion (THD) [27]. point of the PV panels. It is therefore possible to connect a
The configuration includes two different voltage levels, high power panel (for example 300 Wp) which has a Vmpp
a low voltage to include typical batteries of 12/24 V, and over 30 V, or also connect a low power panel (for exam-
a higher voltage to be used in the AC voltage 110 V. The ple 35 Wp) with a Vmpp around 17 V.
proposed control technique is applied to these two power
converters (SEPIC and Boost). In both converters similar A. SEPIC CONVERTER MODEL
transfer functions are obtained for the design of the current
The SEPIC converter switching states [36] are defined con-
control loops.
sidering both when the switch is closed Sws = 1 and when the
switch is open Sws = 0. Therefore, the equation that models
B. INVERTER AND LCL FILTER
the current in the inductor for both switch states is:
An inverter with an LCL filter [28] (Figure 2) was imple-
mented in order to generate the sinusoidal alternating voltage disL1
L1 = vpv ds − vo (1 − ds ). (1)
to power single-phase AC loads. Sinusoidal Pulse Width dt
Modulation (SPWM) is used with a switching frequency of where ds is the unifying variable in the SEPIC converter
5 kHz, and with a maximum modulation index equal to 1 [29]. model.
In the proposed topology, the Boost converter output voltage
is regulated and controlled, for the DC link. The inverter can B. SEPIC EXACT LINEARIZATION
also be controlled, although it has not been included in this Once the equation of the average model for the SEPIC con-
paper [30]–[33]. The switching frequency can be increased verter has been found, the linearization of the system can
to reduce the filter size. However the results given in [34] are proceed. The exact linearization method requires definition of
adopted and the parameters listed in Table 1 are employed for the new inputs variables, in this case the new input variables
the AC filter design. as usc and ubc for the SEPIC and Boost converter, respectively.
Between the new inputs and the desired output, the power
converter behavior will be linear, as shown in equations
(5) and (11).
Considering (1) and defining a new control input for the
SEPIC converter, usc as:
disL1
usc = L1 (2)
dt
so:
FIGURE 2. Inverter and LCL filter.
usc = vpv ds − vo (1 − ds ). (3)

III. SEPIC CONVERTER AND MPPT Solving for ds , the expression of the linearization is
Figure 3 shows the SEPIC converter topology [35], which is obtained.
analyzed in this paper. The Single-Ended Primary-Inductor usc + vo
converter has the following characteristics that make it a good ds = . (4)
vpv + vo
choice for extracting energy from PV modules:
• The output voltage can be higher or lower than the input Applying the Laplace transformation to the equation (2),
voltage. the following equation can be obtained:
• The input current is constant, with slow dynamics related isL1 1
to the weather conditions. H (s) = s
= . (5)
uc sL1
• The capacitor Cs isolates the input from the output,
providing some protection against short circuits in the This transfer function represents a simple integrator, thus
load. allowing a PI controller to properly follow the reference.

VOLUME 10, 2022 56927

 R. Aliaga et al.: Implementation of Exact Linearization Technique for Modeling and Control of DC/DC Converters

C. CONTROL BLOCK DIAGRAM the isL1,ref current that must be perturbed to evaluate the
As seen above, a linear equivalent plant for the converter power drawn from the PV module. This is possible since the
can be formed. The general control scheme for the inductor isL1,ref current is proportional to the current of the PV module,
L1 current is shown in Figure 4(a). Figure 4(b) shows the con- as shown in Figure 3. The currents according to Kirchhoff’s
trol block in detail; this is the block that is implemented by the current law are related according to:
digital controller (STM32), linearizing the current control and
ipv = iCis + isL1,ref . (6)
generating the trigger signal. usc is the signal that comes from
the PI controller and the modulating signal ds , is obtained In steady-state under ideal conditions when no current
from (4). This is compared with the high frequency carrier flows to the capacitor, the currents of the PV module, ipv and
signal, generating the MOSFETs’ control signal in the SEPIC the current isL1,ref , are equal, ipv = isL1,ref . This justifies the
converter. proposed P&O algorithm, which generates the current isL1,ref
as a reference for the control.
The MPPT algorithm is in charge of finding the exact
point of maximum power, i.e., it finds the values of vpv and
ipv that maximize the power extraction. However, the MPPT
algorithm does not control the SEPIC converter, but only
gives the reference to reach the MPP. Therefore, the SEPIC
nonlinear based control is in charge of reaching the current
reference by manipulating the switching pattern Sws , whose
algorithm must deal with the dynamics and nonlinearities of
the SEPIC converter. Thus, the MPPT algorithm is decoupled
with respect the SEPIC nonlinear control, where both of them
are required for proper operation.
FIGURE 4. Exact linearization, a) simplified-equivalent control block,
b) detailed control block.
IV. BOOST CONVERTER
The linearization process ensures a linear behavior The topology shown in Figure 5 is the non-synchronous Boost
between the new input and the output for every control loop. converter [37] used in this work. The reason for this choice
Small errors on the filter parameter values and sensed vari- this topology to raise the voltage is based on demonstrating
ables may lead to minor errors in the controller behavior. that the linearization technique can be applied on different
Therefore, it is preferred to include an integrator in the con- converters, the linearization technique makes it easy to use a
troller to ensure zero steady state error and deal with any SEPIC or a Boost converter in the voltage-Boost stage.
nonlinear feedback errors. The PI controllers (such as the
one chosen in this work) requirements are very low; does
not represent a significant cost for the digital processor and
includes a integrator which helps to bear these minors errors.
The internal loop corresponds to the current control loop,
and must be complement the external one to extract the
maximum power from the PV module. To do this, the P&O
algorithm is used, which is in charge of delivering the current
reference. The implementation of the external loop to track FIGURE 5. Boost converter topology.
the MPP is detailed below.

D. MPPT AND EXACT LINEARIZATION A. INTERNAL LOOP TRANSFER FUNCTION

In this work, the P&O algorithm is used to track the MPP in As with the SEPIC converter, the unification equation for the
the PV curve [22]. The operation principle of the algorithm Boost converter is for both states:
consists of applying a perturbation to the PV voltage and dibL
assessing whether the extracted power of the PV module is L = vbi − vbo (1 − db ). (7)
dt
higher or lower than the power in the previous sampling time.
where db is the unifying variable in the Boost converter
If it is higher, the algorithm has to track the same path as the
model.
power curve; otherwise, it has to change direction. Finally,
Considering (7) and defining ubc as:
when the previous power is equal to zero or lower than a
small delta of established power, the algorithm has identified dibL
the MPP, therefore, P&O will be oscillating around this point ubc = L (8)
dt
until there is a variation in the power curve of the PV module.
where:
The control scheme to extract the energy from the PV
modules is shown in Figure 6. The P&O algorithm determines ubc = vbi − vbo (1 − db ) (9)

56928 VOLUME 10, 2022

R. Aliaga et al.: Implementation of Exact Linearization Technique for Modeling and Control of DC/DC Converters

FIGURE 6. SEPIC working with P&O.

then solving for db : The input power is equal to the input voltage (considered
constant for simplicity) multiplied by the input current, which
ubc − vbi
db = 1 + . (10) is equal to the current in the inductor L:
vbo
Pbi = Vib ibL (13)
Applying Laplace transforms to (8), the following be
obtained: and the output power is:
ibL 1 1 dubo ub
H (s) = = . (11) Pbo = C + o (14)
ubc sL 2 dt R
Equation (10) allows a function for db to be obtained, where:
which is compared to the triangular signal, as presented in the
control block shown in Figure 8 (a) and (b). A cascade control ubo = (vbo )2 (15)
is shown, with the internal block in charge of controlling the
current ibL , while the external block is in charge of controlling making equivalent the equations (13) and (14):
the output voltage of the Boost converter. This external block 1 dubo ub
generates the reference signal for controlling the current ibL . Vib ibL = C + o. (16)
2 dt R
The inner loop is the fast loop and the external loop is slower,
by at least a factor of 10. Applying Laplace transformers to (16), the following can
In this analysis, the modeling of the general system in order be obtained:
to control the output voltage of the Boost converter was been 1 U b (s)
shown. It should be remembered that the output voltage of Vib ILb (s) = CsUob (s) + o . (17)
2 R
the Boost converter is the voltage of the DC link, as shown
in Figure 1. So the input-output energy balance be used to The previous expression allows finding of the transfer
find a transfer function that allows tuning voltage control in function that relates current ibL to voltage vbo,ref :
the Boost converter. In real scenarios power converters have
Uob (s) RV b
an efficiency of less than 100 %. However it is well known
b∗
= RC
i . (18)
that power converter efficiency is high, normally over 96 %. IL (s) s 2 +1
Therefore, the power balancing approximation is not far from
reality. Despite the difference between the model and the C. EXTERNAL LINEARIZED PI CONTROL LOOP
real implementation being low, PI controllers were employed Equation (18) shows the transfer function for a fixed resis-
to ensure proper operation. Finally, to demonstrate that the tance. If a load change occurs the control may show inap-
assumptions are adequate, several experimental tests were propriate behavior for extreme impacts. However, the tests in
performed to validate the systems, as described in section VI. this work were carried out with a model without linearizing
(shown in Figure 7) the power balance equation, obtaining
B. EXTERNAL LOOP TRANSFER FUNCTION correct results for the R = 200  with which the plant
To find the transfer function that represents the Boost con- was tuned and also showing good result with similar load
verter, the input-output energy balance technique is used. values. To solve this problem the linearization of (16) is
With this transfer function the PI for the external loop of the implemented, in this way an equivalent plant is obtained equal
cascade control can be tuned. The inner loop is the fast loop, to:
which allows the system to establish the equality ILb∗ = ibL Vob (s) 1
(the reference current is equal to the inductor current L) in = . (19)
ILb∗ (s) sC
steady state.
The fundamental equation of the power balance is pre- Then, considering that the Boost converter output power is
sented in Eq. (12): equal to:

Pbi = Pbo . (12) Pbo = vbo ibo (20)

VOLUME 10, 2022 56929

 R. Aliaga et al.: Implementation of Exact Linearization Technique for Modeling and Control of DC/DC Converters

FIGURE 7. Boost converter control block, with internal loop linearized. a) equivalent, b) detailed.

FIGURE 8. Boost converter control block, external and internal loop linearized: a) equivalent, b) detailed.

and solving the derivative of the Eq. (16): calculation stability margins, as shown in Figure 9.
dub  
kibi
C o = 2 Vib ibL − vbo ibo (21) kpbi + s
dt Gb (s) = . (25)
sL
defining uecb as:
It can be seen that the proposed controller has an infinite
dub
uecb =C o (22) gain margin, which ensures the stability regardless of any
dt change in the transfer function gain. Due to the finite value of
replacing: the phase margin, the convergence of the system will only be
  compromised if more poles or delays are incorporated into
uecb = 2 vbi ibL − vbo ibo (23)
the closed loop. This analysis shows similar results for the
solving for ibL : SEPIC converter, considering the similarity in the transfer
functions obtained after linearization. It is important to note
uecb + 2vbo ibo that the aforementioned behavior is valid while the range of
ibL = . (24)
2vbi the system variables remain within the converter operating
region. This is achieved when the duty cycles of both SEPIC
The block for the control of the voltage base in the energy
and Boost converters are less than one. In order to determine
balance and the exact linearization is shown in Figure 8.
the valid operating range, it is necessary to evaluate the steady
state equations of both converters, from where (26) and (27)
V. STABILITY ANALYSIS
can be obtained for SEPIC and Boost, respectively.
With the selected kpbi and kibi parameters (Table 1), it is
possible to address a stability analysis of the proposed lin- vo
ds = (26)
earized system. The direct loop transfer function (25) allows vpv + vo
56930 VOLUME 10, 2022
R. Aliaga et al.: Implementation of Exact Linearization Technique for Modeling and Control of DC/DC Converters

FIGURE 11. Boost converter operating range.

TABLE 1. Setup parameters.

FIGURE 9. Phase margin analysis for the control of the current iLb .

FIGURE 10. SEPIC converter operating range.

vbo − vbi
db = . (27)
vbo
Both equations are plotted for different operating values
and the curves shown in Figure 10 and Figure 11 can be the SEPIC converter and the Boost converter is a function
obtained. It is possible to observe that the only restriction of the battery charge level, and therefore the regulation is
is that the output of the Boost converter must be greater due to the battery capacity. Once the battery is fully charged,
than the battery voltage, which can be ensured with a proper the SEPIC converter stops sending power to the batteries,
reference. As vpv will always be between zero and the open protecting them from overcharge.
circuit voltage, the SEPIC converter cannot reach the over In the case of the inverter and LCL filter, the technique
modulation region, ensuring linear behavior. used to control the semiconductors of the full bridge is not
controlled in a closed loop, being the application of a fixed
VI. EXPERIMENTAL RESULTS modulator. The Boost converter is in charge of the control
A. PROTOTYPE IMPLEMENTATION of the vbo DC voltage, and therefore, the inverter is fed with
The experimental setup of the overall microgrid system with a regulated voltage and the AC voltage can be defined as
the proposed linearization controllers is shown Figure 12. voAC = Gac mvbo . Nevertheless, a simple controller can be
Table 1 shows the system parameters adopted in the added to regulate the AC voltage more accurately under
implementation. The control is implemented in a STM32 different conditions.
F103C8T6 microcontroller; the frequency of the control All power converters have a limited operating region,
interrupt is 10 kHz in the control of the SEPIC and Boost which does not always allow overmodulation. Nonlinear
converter, while the switching frequency in both converters based control allows control of the power converter in the
is equal to 100 kHz (it is possible to design a smaller filter). entire operating region with a linear behavior, and therefore
Regarding the power involved, there are three PV modules the natural response of the power converters is improved by
of 300 Wp under STC, one PV module of 35 Wp under this control technique. The proposed system can work prop-
STC and three batteries of 12 V / 50 A, to carry out the erly under any event so long as the power converter remains
different tests of the system. The DC bus voltage between in the operating region. Naturally, if the event takes the power

VOLUME 10, 2022 56931

 R. Aliaga et al.: Implementation of Exact Linearization Technique for Modeling and Control of DC/DC Converters

B. SEPIC AND MPPT EXPERIMENTAL RESULTS

The implementation of the P&O algorithm was performed
with a frequency of 2 Hz, to achieve non-fluctuating dynam-
ics considering the limitations of the P&O algorithm, such as
the case presented in [39].
The current disturbance generated by the MPPT algorithm
was set equal to 0.2 A for the 300 Wp PV module and 0.05 A
for the 35 Wp PV module. The power ripple and the voltage
ripple are less than 3%.

1) CASE 1: MPPT TEST

In this case, the P&O algorithm extracts 240 Wp of power
from the PV panel. The PV voltage vpv oscillates around 34 V
and the current around 8 A values, as it can be seen in
Figure 13a. The perturbation of the current is equal to 0.2 A.
The time taken to reach the maximum power point is equal
to 20 s, considering that the MPPT starts with an extracted
power of 150 W.

2) CASE 2: MPPT TEST IN A LOW POWER PV PANEL

In case 2, a low PV power is connected, whose maximum
power is 35 Wp under STC. In this case the output voltage
is greater than the input, as the result of the connection of
two 12 V batteries, adding a total of 24 V. The voltage vpv
at the point of maximum power is around 17 V, as shown
in Figure 13b, in this image, the current ipv , voltage of
the battery bank vb and the maximum power are detailed.
In Figure 13c a zoom of Figure 13b is shown, detailing the
disturbance steps in the current isL1 that are reflected in the
FIGURE 12. Implemented topology: 1) SEPIC, 2) Boost, 3) inverter, 4) LCL
filter, 5) battery bank. current of the PV module, which corresponds to 0.05 A;
the frequency of the MPPT as presented in Table 1, is equal
to 2 Hz.
converter out of the valid region, there is not sufficient control
capable to manage the currents and/or voltages variables to
3) CASE 3: ELECTRICAL VARIABLES IN BATTERY CHARGING
track their respective references. Thus, the proposed control
is suitable in the event that the microgrid may fall, as far as the Figure 13d shows the voltage vpv = 15 V, the current
power converter remains in the operating region. Regarding ipv = 1.4 A and the power ppv = 21 W of the PV module.
the operating range, this is only limited by conditions of max- Additionally, the battery voltage vb = 26 V and the current
imum voltage, maximum current, and therefore maximum injected into the battery ib = 0.75 A are shown. With these
input power, which are: 50 V, 20 A and 1000 W respec- values, the power injected to the battery is equal to 19.5 W
tively. The maximum output voltage of the Boost converter which implies, for this case, that the efficiency is equal
was 200 V. It should be mentioned that the control guarantees to 92%.
stability in the full operating range.
The minimum working power of the system is 15 W, 4) CASE 4: CHANGE OF IRRADIANCE
therefore, if less than that power is generated, the system In this case (Figure 14), a change on the irradiance is
trips out, (below that power level, no tests were carried out). applied from 1000 W/m2 to 900 W/m2. A correct perfor-
In addition, the passive filter and the switching frequency mance is observed in following the maximum power point,
were designed to avoid non-CCM. The main purpose of this from 305 Wp to 275 Wp respectively. The frequency of the
work is to evaluate the control technique, it is for this reason P&O algorithm is equal to 10 Hz.
that the efficiency of the converters is not evaluated.
The filter parameters can be designed as a function of C. EXPERIMENTAL RESULTS FOR THE BOOST CONVERTER
the transferred power, the switching frequency and the rip- The voltage boost is carried out using a Boost converter.
ple specification on currents and voltages [35], [38]. There As presented in the analysis, through linearization an easy-
are other works that have worked on the filter designed to-control plant is obtained in the form of a simple inte-
[26], [34] where reference can be made for more details of grator. Therefore, the implementation validates the control
the approximate selection of the filter components. strategy proposed in this paper for the voltage boost stage of

56932 VOLUME 10, 2022

R. Aliaga et al.: Implementation of Exact Linearization Technique for Modeling and Control of DC/DC Converters

FIGURE 13. SEPIC converter and MPPT.

FIGURE 14. Step-change in irradiance from 1000 W /m2 to 900 W /m2 .

FIGURE 15. Boost converter output 100 V.

the topology. The different control cases are detailed below,

either with a load connected to the output of the Boost con- 3) CASE 3: AC SIGNAL
verter, as well as to the output of the LCL filter after being Figure 17 shows the output signal of the LCL inverter, with a
converted into an AC signal through the full-bridge inverter. peak voltage value of 100 V. Figure 18 shows a zoom of the
signal, where it can be seen that the control responds properly,
1) CASE 1: CONTROL OF vob , 200 V independently and very similarly in its response, either if the
In this first case, (Figure 15), the voltage of the Boost con- load is connected to the output of the Boost converter or to the
verter is controlled at a voltage of 200 V with a load of 200 . output of the LCL filter. The THD of the AC voltage signal
The inductor current iL is observed in the graph; this current is is equal to 1.9 % which represents a value lower than the 5 %
controlled by the internal control loop of the Boost converter established by the IEEE 519-2014 standard.
and the value of iL in steady state is equal to 5 A.
4) CASE 4: STEP CHANGE AT AC SIGNAL
2) CASE 2: VOLTAGE STEP CHANGE 50 V, 100 V, AND 150 V Finally, in Figure 19, there is a step change from 60 V to 120 V
In this case (Figure 16), step changes are made to the output at the output of the Boost converter, and, therefore, the AC
voltage of the Boost converter. The voltages are 50 V, 100 V, signal of the filter takes on these maximum magnitudes.
and 150 V. The correct performance in the control of the
references is observed with a time equal to 300 ms for reach- D. MICROGRID TESTS
ing the steady-state value. In the dynamic and the stationary The tests carried out are presented with all the power con-
part of operation, this is a great advantage in a microgrid verters working together. Figure 20 shows the voltage, cur-
where loads may require different voltage levels for their rent and power waveforms of the PV module. After 1.6 s
operation. the MPPT algorithm is activated, the full-bridge inverter is

VOLUME 10, 2022 56933

 R. Aliaga et al.: Implementation of Exact Linearization Technique for Modeling and Control of DC/DC Converters

FIGURE 16. Boost converter step ref. FIGURE 20. MPPT and inverter waveform.

FIGURE 17. AC output voltage. FIGURE 21. MPPT and inverter waveform zoom.

FIGURE 18. Zoom to the output voltage inverter with LCL filter. FIGURE 22. MPPT and inverter waveform.

voltage equal to 150 V, next to the graphs of the variables of

the PV module.

VII. CONTROL AND TOPOLOGY COMPARISON

A SEPIC converter and a Boost converter have nonlinear
behavior, which implies that a linear controller (for example,
a PI controller) will have an adequate performance only in
the proximity of the operating point where the system was
linearized. However, at other values of reference, such a task
FIGURE 19. Step to the output voltage inverter with LCL filter. will not occur. Therefore, strategies that use linear control
are not practical for systems that require different operating
points, (unless linearized point by point within the operating
switched-on. It is observed that there are no deficiencies in range, which is impractical). This becomes more impractical
the general operation of the topology, with power drawn from in PV systems where the power curve is highly nonlinear.
the PV module in this test of 240 W. Additionally, Figure 21, For this reason, the proposed control in this work comes
shows a zoom to the signal presented in Figure 20, where the to enhance the controllability of DC-DC converters, whose
oscillation of the MPPT (whose current disturbance was set intrinsic nature is nonlinear. From the point of view of control
at 0.2 A) and the dynamics of the AC signal at the output of simplicity, the advantage in the use of exact linearization
the LCL filter can both be observed. Finally, Figure 22 shows is clearly appreciated, allowing a simple integrator to be
the control around a reference voltage of the Boost converter obtained as an equivalent model, unlike the transfer function

56934 VOLUME 10, 2022

R. Aliaga et al.: Implementation of Exact Linearization Technique for Modeling and Control of DC/DC Converters

TABLE 2. Control strategies disadvantages. takes advantage of the research done in [40], immediately
using one of the converters that have greater advantages in
characteristics such as voltage polarity, input current, switch
driver and costs for this application.
In the case of the boost function, as it is only necessary
to raise the voltage to standard voltages, a simpler con-
verter such as the Boost converter can be used (Figure 16).
This choice allows demonstration that the exact linearization
technique is applicable to other topologies using the same
methodology for modeling and control.

VIII. CONCLUSION
In this paper, the implementation of an exact linearization
technique for control of SEPIC and Boost converters was
presented. Such converters are an essential part of a microgrid
when one of the energy sources is solar PV. It should be
noted that this technique can be applied in different energy
conversion systems, including wind and wave energy. The
simplification that comes with exact linearization allows the
development of simple controllers, with the ability to control
the desired variables over a wide range of values to verify the
proper performance of the linearization proposal, a prototype
of a SEPIC converter [40] and Boost converter [41] linearized was developed with the aforementioned converters, whose
around an operating point. In this way, the design process objective is the extraction of energy from the PV panels,
of a controller that meets the operating requirements of the storage of energy in the battery bank and DC/AC conversion
converters is highly efficient. the P&O algorithm and the current control are used, and are
In order to justify the superiority of the exact linearization able to follow that the reference generated by the MPPT in a
method, Table 2 shows the main disadvantages in other con- proper manner using the current control with exact lineariza-
trol strategies reported in the literature. In this context, the tion due to the ability of the SEPIC converter to Boost and
following comparative points are detailed from Table 2. In the Buck the input voltage. It was possible to connect PV panels
exact linearization technique, there is no extreme dependence with a voltage at the maximum power point lower and higher
on the system parameters, unlike other control strategies, and than the battery voltage. The application of exact linearization
it can be noted that in both converters in this work, the func- allows the Boost converter to be controlled even if the voltage
tion of transfer only makes use of the value of the inductance. rise. DC-type loads, for example some electronic equipment,
Regarding the computational effort, only basic mathematical can be fed directly from the DC link voltage, allowing the size
operations are required. For steady state error, it has been of the inverter operates to be reduced, along with a reduction
shown that this is zero in the exact linearization technique, in losses, increasing the efficiency of the use of energy in a
thanks to the integrator in the transfer function. The switching rural-microgrid application.
frequency is fixed, product of comparing a modulating signal
with respect to a fixed-frequency triangular. Finally, it is REFERENCES
convenient to mention that linearization does not present the [1] M. Nasir, H. A. Khan, A. Hussain, L. Mateen, and N. A. Zaffar, ‘‘Solar
disadvantages of fuzzy control, sliding and neural networks. PV-based scalable DC microgrid for rural electrification in develop-
ing regions,’’ IEEE Trans. Sustain. Energy, vol. 9, no. 1, pp. 390–399,
Regarding the selected topology, the use of a SEPIC con- Jan. 2018.
verter to extract PV energy has a great advantage over other [2] R. Viswadev, A. Mudlapur, V. V. Ramana, B. Venkatesaperumal, and
converters, within the group of non-isolated up and down S. Mishra, ‘‘A novel AC current sensorless hysteresis control for grid-
tie inverters,’’ IEEE Trans. Circuits Syst. II, Exp. Briefs, vol. 67, no. 11,
converters, highlighting that the output voltage is not inverted, pp. 2577–2581, Nov. 2020.
and the current input is constant. This last characteristic [3] J. Rocabert, A. Luna, F. Blaabjerg, and P. Rodríguez, ‘‘Control of power
being a fundamental requirement for the application of the converters in AC microgrids,’’ IEEE Trans. Power Electron., vol. 27,
no. 11, pp. 4734–4749, Nov. 2012.
exact linearization technique in converters where the control [4] M. Ahmed, L. Meegahapola, A. Vahidnia, and M. Datta, ‘‘Stability and
objective is the current [40]. control aspects of microgrid architectures–A comprehensive review,’’
The SEPIC converter works as a Buck-Boost, thus allow- IEEE Access, vol. 8, pp. 144730–144766, 2020.
[5] T. Dragicevic, S. Vazquez, and P. Wheeler, ‘‘Advanced control methods
ing adequate battery charging independent of their voltage for power converters in DG systems and microgrids,’’ IEEE Trans. Ind.
or of the voltage in the PV modules (Figure 13a). In addi- Electron., vol. 68, no. 7, pp. 5847–5862, Jul. 2021.
tion, [40] highlights the advantages and characteristics of the [6] M. Khosravi, M. Amirbande, D. A. Khaburi, M. Rivera, J. Riveros,
J. Rodriguez, A. Vahedi, and P. Wheeler, ‘‘Review of model predictive
SEPIC that make it suitable for charging batteries considering control strategies for matrix converters,’’ IET Power Electron., vol. 12,
the variable voltage of the PV modules. In this way, this work no. 12, pp. 3021–3032, Oct. 2019.

VOLUME 10, 2022 56935

 R. Aliaga et al.: Implementation of Exact Linearization Technique for Modeling and Control of DC/DC Converters

[7] L. Callegaro, M. Ciobotaru, D. J. Pagano, and J. E. Fletcher, ‘‘Feedback [26] F. Li, X. Zhang, H. Zhu, H. Li, and C. Yu, ‘‘An LCL-LC filter for grid-
linearization control in photovoltaic module integrated converters,’’ IEEE connected converter: Topology, parameter, and analysis,’’ IEEE Trans.
Trans. Power Electron., vol. 34, no. 7, pp. 6876–6889, Jul. 2019. Power Electron., vol. 30, no. 9, pp. 5067–5077, Sep. 2015.
[8] C. R. Baier, M. A. Torres, P. Acuna, J. A. Muñoz, P. E. Melín, C. Restrepo, [27] IEEE Recommended Practice and Requirements for Harmonic Control in
and J. I. Guzman, ‘‘Analysis and design of a control strategy for tracking Electric Power Systems, IEEE Standard 519-2014 (Revision IEEE Stan-
sinusoidal references in single-phase grid-connected current-source invert- dard 519-1992), 2014, pp. 1–29.
ers,’’ IEEE Trans. Power Electron., vol. 33, no. 1, pp. 819–832, Jan. 2018. [28] Y. Tang, W. Yao, P. C. Loh, and F. Blaabjerg, ‘‘Design of LCL filters
[9] J. Zhou and X. Lu, ‘‘Review of exact linearization method applied to with LCL resonance frequencies beyond the Nyquist frequency for grid-
power electronics system,’’ in Proc. Asia–Pacific Power Energy Eng. Conf., connected converters,’’ IEEE Trans. Emerg. Sel. Topics Power Electron.,
Mar. 2012, pp. 1–4. vol. 4, no. 1, pp. 3–14, Mar. 2016.
[10] W. E. Aouni and L.-A. Dessaint, ‘‘Real-time implementation of input-state [29] M. Lakka, E. Koutroulis, and A. Dollas, ‘‘Development of an FPGA-based
linearization and model predictive control for robust voltage regulation of a SPWM generator for high switching frequency DC/AC inverters,’’ IEEE
DC-DC boost converter,’’ IEEE Access, vol. 8, pp. 192101–192108, 2020. Trans. Power Electron., vol. 29, no. 1, pp. 356–365, Jan. 2014.
[11] X.-F. Cheng, C. Liu, D. Wang, and Y. Zhang, ‘‘State-of-the-Art review [30] D. Puyal, L. A. Barragan, J. Acero, J. M. Burdio, and I. Millan, ‘‘An FPGA-
on soft-switching technologies for non-isolated DC-DC converters,’’ IEEE based digital modulator for full- or half-bridge inverter control,’’ IEEE
Access, vol. 9, pp. 119235–119249, 2021. Trans. Power Electron., vol. 21, no. 5, pp. 1479–1483, Sep. 2006.
[12] X. Pan, H. Li, Y. Liu, T. Zhao, C. Ju, and A. K. Rathore, [31] W. J. Cha, J. M. Kwon, and B. H. Kwon, ‘‘Highly efficient asymmetrical
‘‘An overview and comprehensive comparative evaluation of current-fed- PWM full-bridge converter for renewable energy sources,’’ IEEE Trans.
isolated-bidirectional DC/DC converter,’’ IEEE Trans. Power Electron., Ind. Electron., vol. 63, no. 5, pp. 2945–2953, May 2016.
vol. 35, no. 3, pp. 2737–2763, Mar. 2020. [32] D. Wu, Y. Wu, J. Kan, Y. Tang, J. Chen, and L. Jiang, ‘‘Full-bridge current-
[13] F. Blaabjerg, ‘‘Control of Power Electronic Converters and Systems, vol. 1. fed PV microinverter with DLFCR reduction ability,’’ IEEE Trans. Power
New York, NY, USA: Academic, 2018. Electron., vol. 35, no. 9, pp. 9541–9552, Sep. 2020.
[14] D. Murillo-Yarce, J. Alarcón-Alarcón, M. Rivera, C. Restrepo, J. Muñoz, [33] M. Ciobotaru, R. Teodorescu, and F. Blaabjerg, ‘‘Control of single-stage
C. Baier, and P. Wheeler, ‘‘A review of control techniques in photovoltaic single-phase PV inverter,’’ in Proc. Eur. Conf. Power Electron. Appl., 2005,
systems,’’ Sustainability, vol. 12, no. 24, p. 10598, Dec. 2020. p. 10.
[15] R.-J. Wai and W.-H. Wang, ‘‘Grid-connected photovoltaic generation sys- [34] M. Sanatkar-Chayjani and M. Monfared, ‘‘Design of LCL and LLCL filters
tem,’’ IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 55, no. 3, pp. 953–964, for single-phase grid connected converters,’’ IET Power Electron., vol. 9,
Apr. 2008. no. 9, pp. 1971–1978, Jul. 2016.
[16] A. Merabet, L. Labib, A. M. Y. M. Ghias, C. Ghenai, and T. Salameh, [35] J. J. Jozwik and M. K. Kazimierczuk, ‘‘Dual SEPIC PWM switching-
‘‘Robust feedback linearizing control with sliding mode compensation mode DC/DC power converter,’’ IEEE Trans. Ind. Electron., vol. 36, no. 1,
for a grid-connected photovoltaic inverter system under unbalanced grid pp. 64–70, Feb. 1989.
voltages,’’ IEEE J. Photovolt., vol. 7, no. 3, pp. 828–838, May 2017. [36] E. Babaei and M. E. S. Mahmoodieh, ‘‘Calculation of output voltage
[17] M. B. Shadmand, R. S. Balog, and H. Abu-Rub, ‘‘Model predictive control ripple and design considerations of SEPIC converter,’’ IEEE Trans. Ind.
of PV sources in a smart DC distribution system: Maximum power point Electron., vol. 61, no. 3, pp. 1213–1222, Mar. 2014.
tracking and droop control,’’ IEEE Trans. Energy Convers., vol. 29, no. 4, [37] M. Forouzesh, P. Y. Siwakoti, A. S. Gorji, F. Blaabjerg, and B. Lehman,
pp. 913–921, Dec. 2014. ‘‘Step-up DC-DC Converters: A comprehensive review of voltage-
[18] J. Hong, J. Yin, Y. Liu, J. Peng, and H. Jiang, ‘‘Energy management and boosting techniques, topologies, and applications,’’ IEEE Trans. Power
control strategy of photovoltaic/battery hybrid distributed power genera- Electron., vol. 32, no. 12, pp. 9143–9178, Dec. 2017.
tion systems with an integrated three-port power converter,’’ IEEE Access, [38] S. A. Shirsavar, ‘‘Teaching practical design of switch-mode power sup-
vol. 7, pp. 82838–82847, 2019. plies,’’ IEEE Trans. Educ., vol. 47, no. 4, pp. 467–473, Nov. 2004.
[19] L. V. Bellinaso, H. H. Figueira, M. F. Basquera, R. P. Vieira, [39] C. González-Castaño, C. Restrepo, S. Kouro, and J. Rodriguez, ‘‘MPPT
H. A. Grundling, and L. Michels, ‘‘Cascade control with adaptive voltage algorithm based on artificial bee colony for PV system,’’ IEEE Access,
controller applied to photovoltaic boost converters,’’ IEEE Trans. Ind. vol. 9, pp. 43121–43133, 2021.
Appl., vol. 55, no. 2, pp. 1903–1912, Mar. 2019. [40] S. J. Chiang, H.-J. Shieh, and M.-C. Chen, ‘‘Modeling and control of PV
[20] R. B. Bollipo, S. Mikkili, and P. K. Bonthagorla, ‘‘Hybrid, optimal, intelli- charger system with SEPIC converter,’’ IEEE Trans. Ind. Electron., vol. 56,
gent and classical PV MPPT techniques: A review,’’ CSEE J. Power Energy no. 11, pp. 4344–4353, Nov. 2009.
Syst., vol. 7, no. 1, pp. 9–33, Jan. 2021. [41] R. H. G. Tan and L. Y. H. Hoo, ‘‘DC-DC converter modeling and simu-
[21] C. R. Baier, M. Torres, J. A. Muñoz, R. A. Marco, E. N. Eduardo, and lation using state space approach,’’ in Proc. IEEE Conf. Energy Convers.
P. Acuña, ‘‘Bidirectional power flow control of a single-phase current- (CENCON), Oct. 2015, pp. 42–47.
source grid-tie battery energy storage system,’’ in Proc. IEEE 24th Int. [42] M. Khalilzadeh, S. Vaez-Zadeh, J. Rodriguez, and R. Heydari, ‘‘Model-
Symp. Ind. Electron. (ISIE), Jun. 2015, pp. 1372–1377. free predictive control of motor drives and power converters: A review,’’
[22] D. Sera, L. Mathe, T. Kerekes, S. V. Spataru, and R. Teodorescu, ‘‘On the IEEE Access, vol. 9, pp. 105733–105747, 2021.
perturb-and-observe and incremental conductance MPPT methods for PV [43] P. Mattavelli, L. Rossetto, G. Spiazzi, and P. Tenti, ‘‘General-purpose fuzzy
systems,’’ IEEE J. Photovolt., vol. 3, no. 3, pp. 1070–1078, Jul. 2013. controller for DC-DC converters,’’ IEEE Trans. Power Electron., vol. 12,
[23] K. Jain, M. Gupta, and A. K. Bohre, ‘‘Implementation and comparative no. 1, pp. 79–86, Jan. 1997.
analysis of P&O and INC MPPT method for PV system,’’ in Proc. 8th [44] T. Xue and H. Cheng, ‘‘Research and improvement of current hysteresis
IEEE India Int. Conf. Power Electron. (IICPE), Dec. 2018, pp. 1–6. control strategy for boost DC-DC converter,’’ in Proc. Int. Conf. Sensor
[24] S. Bhattacharyya, D. S. Kumar P, S. Samanta, and S. Mishra, ‘‘Steady Netw. Signal Process. (SNSP), Oct. 2018, pp. 266–270.
output and fast tracking MPPT (SOFT-MPPT) for P&O and InC algo- [45] N. Chettibi, A. Mellit, G. Sulligoi, and A. M. Pavan, ‘‘Adaptive neural
rithms,’’ IEEE Trans. Sustain. Energy, vol. 12, no. 1, pp. 293–302, network-based control of a hybrid AC/DC microgrid,’’ IEEE Trans. Smart
Jan. 2021. Grid, vol. 9, no. 3, pp. 1667–1679, May 2018.
[25] M. Pokharel, N. Hildebrandt, C. N. M. Ho, and Y. He, ‘‘A fast-dynamic [46] R.-J. Wai and L.-C. Shih, ‘‘Design of voltage tracking control for DC–
unipolar switching control scheme for single-phase inverters in DC DC boost converter via total sliding-mode technique,’’ IEEE Trans. Ind.
microgrids,’’ IEEE Trans. Power Electron., vol. 34, no. 1, pp. 916–927, Electron., vol. 58, no. 6, pp. 2502–2511, Jun. 2011.
Jan. 2019.

56936 VOLUME 10, 2022