Anda di halaman 1dari 25

Article

Sliding-Mode Controller for Maximum Power Point


Tracking in Grid-Connected Photovoltaic Systems
Paula Andrea Ortiz Valencia 1, * and Carlos Andres Ramos-Paja 2
Received: 11 September 2015 ; Accepted: 19 October 2015 ; Published: 2 November 2015
Academic Editor: Tapas Mallick
1 Instituto Tecnológico Metropolitano (ITM), Calle 73 No 76A-354 Vía al Volador,
Medellín 05001000, Colombia
2 Universidad Nacional de Colombia, Carrera 80 No 65-223-Block M8-113, Medellín 050041,
Colombia; caramosp@unal.edu.co
* Correspondence: paulaortiz@itm.edu.co; Tel.: +57-4581-3957 or +57-4460-0727 (ext. 5587)

Abstract: The maximum power point tracking (MPPT) of photovoltaic systems must be as fast and
accurate as possible to increase the power production, which eventually increases the PV system
profitability. This paper proposes and mathematically analyses a sliding-mode controller to provide
a fast and accurate maximum power point tracking in grid-connected photovoltaic systems using a
single control stage. This approach avoids the circular dependency in the design of classical cascade
controllers used to optimize the photovoltaic system operation, and at the same time, it reduces
the number of controllers and avoids the use of linearized models to provide global stability in all
the operation range. Such a compact solution also reduces the system cost and implementation
complexity. To ensure the stability of the proposed solution, detailed mathematical analyses
are performed to demonstrate the fulfillment of the transversality, reachability and equivalent
control conditions. Finally, the performance of the proposed solution is validated using detailed
simulations, executed in the power electronics simulator PSIM, accounting for both environmental
and load perturbations.

Keywords: DC/DC converter; maximum power point tracking (MPPT); photovoltaic systems;
sliding mode control; grid-connection

1. Introduction
The world energy demand is projected to more than double by 2050, and more than triple by the
end of the century [1]. Incremental improvements in existing energy networks will not be adequate
to supply this demand in a sustainable way. Hence, it is necessary to find sources of clean energy
with a wide distribution around the world. The energy generation with photovoltaic (PV) systems is
inexhaustible, hence it is a suitable candidate for a long-term, reliable and environmentally friendly
source of electricity. However, PV systems require specialized control algorithms to guarantee the
extraction of the maximum power available, otherwise the system could be unsustainable.
The PV generator, also known as PV array, produces DC power that depends on the
environmental conditions and operating point imposed by the load. To provide a high power
production, the PV system includes a DC/DC converter to isolate the operating point of the generator
(voltage and current) from the load, where such a power converter is regulated by an algorithm
that searches, online, the maximum power point (MPP, i.e., the optimal operation condition) known
as Maximum Power Point Tracking (MPPT) algorithm. The classical structure of a grid-connected
photovoltaic system is presented in Figure 1, in which the PV generator interacts with a DC/DC
converter controlled by a MPPT algorithm [2,3]. Such a structure enables the PV system to modify

Energies 2015, 8, 12363–12387; doi:10.3390/en81112318 www.mdpi.com/journal/energies


Tracking (MPPT) algorithm. The classical structure of a grid-connected photovoltaic system is presented
in Figure 1, in which the PV generator interacts with a DC/DC converter controlled by a MPPT
algorithm [2,3]. Such a structure enables the PV system to modify the operation conditions in
agreement with the environmental circumstances (mainly changed by the irradiance and temperature)
Energies 2015, 8, 12363–12387
so that a maximum power production is achieved [4,5]. Figure 1 also illustrates the gird-connection
side of the PV system, which is formed by a DC-link (capacitor ) and a DC/AC converter (inverter).
the operation conditions in agreement with the environmental circumstances (mainly changed by
The inverter is controlled
the irradiance to follow
and temperature) so athat
required
a maximum power factor,
power provideissynchronization
production and protect
achieved [4,5]. Figure 1
against
also illustrates the gird-connection side of the PV system, which is formed by a DC-link (capacitor bulk
islanding, among others. Moreover, the inverter must to regulate the DC-link voltage at the
capacitor
Cb ) and, where
a DC/AC twoconverter
cases are(inverter).
possible:The first, the inverter
inverter regulates
is controlled the DC
to follow component
a required power of factor,
voltage,
provide synchronization and protect against islanding, among others. Moreover, the inverter must
but due to the sinusoidal power injection into the grid, voltage experiments a sinusoidal perturbation at
to regulate the DC-link voltage at the bulk capacitor Cb , where two cases are possible: first, the
twice inverter
the grid regulates
frequencythe andDCwith a magnitude
component of Cbinversely proportional
voltage, but due to thetosinusoidal
the capacitance [3]. In the
power injection intosecond
case, the
the grid,
DC component of voltage
Cb voltage experiments is not properly
a sinusoidal regulated,
perturbation at twicewhich produces
the grid multiple
frequency and withharmonic
a
components with amplitude inversely proportional to the capacitance [3]. In both cases, the DC/DC
magnitude inversely proportional to the capacitance [3]. In the second case, the DC component of
Cb voltage is not properly regulated, which produces multiple harmonic components with amplitude
converter output terminals are exposed to voltage perturbations that could be transferred to the PV
inversely proportional to the capacitance [3]. In both cases, the DC/DC converter output terminals
generator terminals,
are exposed thus perturbations
to voltage degrading the thatMPP
could tracking process.
be transferred to theConcerning
PV generator the DC/DC
terminals, thuspower
converter, the boost
degrading the MPP topology
tracking is the most
process. widelythe
Concerning used
DC/DC due power
to theconverter,
low voltage levels
the boost exhibited
topology is by
commercial
the most PV modules
widely used[6].
due to the low voltage levels exhibited by commercial PV modules [6].

Figure 1. Typical structure of a photovoltaic (PV) system.


Figure 1. Typical structure of a photovoltaic (PV) system.

Multiple types of MPPT solutions are reported in the literature, which differ in complexity,
number of sensors needed for operation, convergence speed, cost-effective range, etc. [7]. Two of
the most commonly used MPPT techniques are Perturb and Observe (P & O) and Incremental
Conductance (IC); the reason for this popularity is its implementation simplicity and its relatively
good performance [4,8]. Other MPPT techniques are based on using fractional values of the open
circuit voltage and short circuit current, i.e., the Fractional Open Circuit Voltage and the Fractional
Short-Circuit Current. The main advantages of those solutions are the low cost and implementation
simplicity since they only require a single (voltage or current) sensor [9,10]; But their efficiency is
low compared with the P & O and IC algorithms. In contrasts, techniques based on computational
intelligence, such as neural networks and fuzzy logic, offer speed and efficiency in tracking the
MPP [11–13]; however its complexity and implementation costs are high compared with the P & O
and IC algorithms, which make them costly solutions.
The main problem of using traditional MPPT algorithms acting on the duty cycle of the DC/DC
converter associated to the PV generator, i.e., Figure 1, concerns the large disturbances caused by
irradiance transients in the system operating point, which generates a slow tracking of the MPP.
This condition is also present at the system start-up, in which the MPPT algorithm takes a large
amount of time to reach the MPP [3]. To mitigate such disturbances and speed-up the MPPT
procedure, a two-stage control structure is usually adopted: it generally involves an algorithmic
MPPT controller in cascade with a conventional voltage regulator (e.g., based on lineal or nonlinear
control) as depicted in Figure 2. Moreover, such a structure is also needed to increase the reliability
of double-stage grid-connected PV systems: the sinusoidal oscillation on the DC-link caused by the
inverter operation must be mitigated, otherwise the MPPT procedure could be inefficient as reported
in [3]. Such mitigation is traditionally performed by using large electrolytic capacitors for Cb , however
the electrolytic technology introduces reliability problems due to its high failure rate [14]. Then, the

12364
Figure 2. Moreover, such a structure is also needed to increase the reliability of double-stage
grid-connected PV systems: the sinusoidal oscillation on the DC-link caused by the inverter operation
must be mitigated, otherwise the MPPT procedure could be inefficient as reported in [3]. Such
mitigation
Energies is traditionally
2015, 8, 12363–12387 performed by using large electrolytic capacitors for Cb, however the

electrolytic technology introduces reliability problems due to its high failure rate [14]. Then, the
voltage regulator
voltage in Figure
regulator 2 enables
in Figure 2 enablestotomitigate
mitigate the voltageoscillations
the voltage oscillations in the
in the DC-link
DC-link produced
produced by by
small non-electrolytic capacitors, thus improving the system reliability [3,14].
small non-electrolytic capacitors, thus improving the system reliability [3,14].

Figure 2. Figure
Structure
2. of a PV system
Structure with
of a PV a classical
system with apulse width
classical modulation
pulse (PWM)-based
width modulation cascade control.
(PWM)-based
cascade control.

However, the design process of such a voltage regulator depends on the MPPT parameters, while
However, the design process of such a voltage regulator depends on the MPPT parameters, while
the MPPT controller stability also depends on the cascade controller performance [3]: for example, to
the MPPT controller stability also depends on the cascade controller performance [3]: for example, to
ensureensure
the stability of aofPa&
the stability P&OOalgorithm,
algorithm, its perturbationperiod
its perturbation periodmust
must be be larger
larger thanthan the settling
the settling time time
of theofPVthevoltage, which
PV voltage, depends
which depends ononthe
thesystem
system operating pointand
operating point andononthethe cascade
cascade voltage
voltage controller.
controller.
However, the voltage controller design requires performance criteria, e.g., the settling
However, the voltage controller design requires performance criteria, e.g., the settling time as time as in [3],in [3],
which are usually imposed in terms of the P & O parameters. Such a circular dependency in the two
controllers design makes difficult to guarantee both system stability and desired performance in all
the operating conditions.
When the voltage regulator is implemented using conventional linear control techniques it is
necessary to linearize the system model around a given operation point, which is usually the MPP at
some irradiance condition [3]. However, due to the nonlinear nature of the PV module and DC/DC
converter, the performance (and even stability) of the linear controller is limited to the neighborhood
around the MPP [15]. This constraint puts at risk the system performance since the operating
point changes with the unpredictable and unavoidable environmental perturbations. To address this
problem, the work in [16] uses a sliding mode controller (SMC) to regulate the inductor current of
a boost converter associated to the PV module, which enables to guarantee global system stability
at any operating point. The solution proposed in that work considers three controllers in cascade
as follows: the SMC that generates the activation signal for the MOSFET, a PI controller designed
to provide the SMC reference depending on the command provided by a P & O algorithm, which
is in charge of optimizing the power. However, the design of the PI controller requires a linearized
model of the system around the MPP, hence it cannot guarantee the same performance in all the range
of operation. In fact, a wrong design of such a PI controller could make the P & O unstable, hence
both PI and P & O controllers have a circular dependency on their parameters. Similarly, the work
in [17] uses a SMC to regulate the input capacitor current of the boost converter. This solution has
a major advantage over the work reported in [16]: the solution in [17] does not require a linearized
model since the transfer function between the capacitor current and voltage is linear and it does not
depend on the irradiance or temperature conditions. Therefore, such a solution is able to guarantee
the desired performance in all the operating range. However, as in the previous work, the three
controllers are designed separately, which makes difficult to perform the system design: again, a
wrong P & O perturbation period could lead to an unstable system operation.
To avoid the circular dependency between voltage controllers and MPPT algorithms, a single
controller in charge of both MPPT and voltage control operations is required. This problem has
been addressed in [18–21] by using the sliding mode control technique. However, those works

12365
Energies 2015, 8, 12363–12387

are based on two considerations difficult to apply to grid-connected PV systems: First, there is not
considered a capacitor linking the PV array and the DC/DC converter; second, the load impedance
is considered constant.
Regarding the first consideration, PV systems commonly consider a capacitor between the PV
source and the DC/DC converter to stabilize the PV voltage, which in turns stabilizes the power
produced, otherwise the current ripple at the DC/DC converter input will produce undesired
oscillations that will degrade the MPPT procedure. The worst case of that current ripple is exhibited
by DC/DC converters with discontinuous input current, e.g., buck or buck-boost topologies, while
DC/DC converters with continuous input current only inject the inductor current ripple, e.g., boost,
Sepic or Cuk topologies. In any case, a capacitor in parallel with the PV source is almost always
considered; in fact, the experimental scheme of [21] includes such a capacitor despite it is not taken
into account for the SMC analysis. It must be noted that this capacitor defines the dynamic behavior
of the PV voltage and power, therefore it must be considered in the PV controller design.
Regarding the second consideration, which assumes constant the load impedance, it is not
applicable to grid-connected inverters since, as reported in [22], such inverters are controlled to
provide a constant average value in the DC-link, with sinusoidal oscillations at twice the grid
frequency due to the injection of single-phase AC power. Therefore, the best representation of such a
load (from the DC/DC converter side) is a voltage source and not a constant impedance. This model
will be further justified in Section 2. Therefore, in the design of a PV controller for grid-connected
applications it is desirable to account for loads with non-constant impedance.
Other interesting sliding-mode controller for PV systems is presented in [23], it considering a
buck converter. Despite this work considers the capacitor linking the PV source and the power stage,
it is not applicable to grid-connected PV systems since the DC/DC converter provides an output
voltage lower than the one provided by the PV source. Such a condition is incompatible with the
high input-voltage required by classical buck inverters for grid-connection [22,24]. Moreover, this
work also assumes constant the load impedance, and it does not analyze the switching frequency,
which is an important parameter for implementation.
On the basis of those previous works and considerations, this paper presents the analysis and
design of a SMC aimed at performing a fast MPPT action on grid-connected PV systems using a single
control stage. This approach avoids the circular dependency among cascade controllers, reduces the
number of controllers and avoids the use of linearized models to provide global stability in all the
operation range. In such a way, even the use of a PWM is not needed. Such a compact design also
reduces the system cost and complexity.
The paper is organized as follows: Section 2 presents the non-linear mathematical model
representing the PV system, then Section 3 introduces the proposed sliding surface and the controller
structure. Section 4 presents the mathematical analysis of the transversality, reachability and
equivalent control conditions to demonstrate the global stability of the proposed controller. Finally,
Section 5 illustrates the performance of the proposed solution using detailed simulations executed in
a standard power electronics simulator. The conclusions close the paper.

2. System Model
A simplified circuital scheme of the PV system is presented in the Figure 3, which considers
a boost converter due to the widely adoption of such a step-up DC/DC structure in PV systems,
however the analysis presented in this paper can be extended to other DC/DC topologies.
The scheme includes a voltage source as the system load, this to model the DC-link of double-stage
structures in commercial PV inverters, in which the DC/AC stage regulates the DC-link voltage
(Cb capacitor voltage) [25]. This voltage source model is widely used to represent the closed-loop
grid-connected inverters due to its satisfactory relation between accuracy and simplicity, which is
confirmed in references [3,17,25–29].

12366
Energies 2015, 8, 12363–12387

The scheme considers the SMC acting directly on the MOSFET by means of the signal u, which
is a discontinuous variable. Hence, no linearization or PWM are needed. This condition reduces the
implementation cost, and circuit complexity, in comparison with SMC solutions implemented with
Energies
PWM 2015, 8 e.g., [21].
circuits, 6

3. Circuital scheme of the sliding mode controller (SMC) loop.


FigureFigure
3. Circuital scheme of the sliding mode controller (SMC) loop.
The dynamic behavior of the DC/DC converter is modeled by the switched Equations (1) and
The dynamic behavior of the DC/DC converter is modeled by the switched Equations (1) and (2) [30],
(2) [30], where i L represents the inductor current, v pv is the PV voltage, i pv represents the PV module
wherecurrent,
represents the inductor current, is the PV voltage, represents the PV module current,
v is the load voltage and L and C represent the inductor and capacitor values:
b in
is the load voltage and and represent the inductor and capacitor values:
di L v pv ´ vb p1 ´ uq
dt

L
1 (1)
(1)
dv pv i pv ´ i L
“ (2)
dt Cin
(2)
The current of the PV module is modeled with the simplified single diode model [17] given in
TheEquation
current(3).
of In
thesuch model isc
PV amodule is represented the short-circuit
modeled with current
the simplified thatdiode
single is almost proportional
model [17] given in
to the irradiance [17], B is the diode saturation current and A represents the inverse of the
Equation (3). In such a model represented the short-circuit current that is almost proportional to the
thermal voltage that depends on the temperature [25]. Such a model parameters are calculated
irradiance
as A [17],
“ Bistcis.exp
thep´B
diode saturation current
Bstc and A represents the
stc .voc q, B “ 1`a . T ´T and Bstc “ pvinverse
lnp1´i MPP {istc qof the thermal voltage that
, where istc and Tstc
v p pv stc q MPP ´voc q
depends
are on
the the temperature
short-circuit [25].and
current Such a model of
temperature parameters
the module areunder
calculated
standard . exp (STC),.
as test conditions ,
respectively. vand ln 1 ⁄
oc represents the open-circuit voltage, while ,v MPP whereand i MPPandcorrespond are the short-circuit
to the PV
.
voltage and current, respectively, at the MPP for the given operating conditions. Finally, av is the
current and temperature
voltage temperature of the module
coefficient [31]: under standard test conditions (STC), respectively. represents
the open-circuit voltage, while and correspond
´ to the
¯ PV voltage and current, respectively, at
i pv “ isc ´ B e Av pv ´ 1 (3)
the MPP for the given operating conditions. Finally, is the voltage temperature coefficient [31]:
In conclusion, the non-linear equations system formed 1
by Equations (1)–(3) describes the PV (3)
system dynamic behavior in any operation condition.
In conclusion, the non-linear equations system formed by Equations (1)–(3) describes the PV
3. Sliding Surface and Controller Structure
system dynamic behavior in any operation condition.
The non-linear relation between the PV current and voltage, given in Equation (3), produces
a non-linear
3. Sliding Surface relation between theStructure
and Controller PV power and voltage (or current), which exhibits a maximum as
depicted in Figure 4. In such a maximum, i.e., the MPP, the derivative of the power with respect to the
voltage (or current) is zero as given in Equation (4). Therefore, the sliding-mode controllers proposed
The non-linear relation between the PV current and voltage, given in Equation (3), produces a
in [18–21,23] are based on Equation (4). In a similar way, this paper proposes a sliding surface based
non-linear relation
on Equation (4):between the PV power and voltage (or current), which exhibits a maximum as
depicted in Figure 4. In such a maximum,dpi.e., pv the MPP,
“ 0 and
dp pvthe derivative of the power with respect to the
“0 (4)
dv pv
voltage (or current) is zero as given in Equation di pv
(4). Therefore, the sliding-mode controllers proposed
in [18–21,23] are based on Equation (4). In a similar way, this paper proposes a sliding surface based
on Equation (4): 12367

0 0 (4)
Energies 2015, 8, 12363–12387
Energies 2015, 8 7

90 90

80 80

70 70

60 (d ppv/d vpv) = 0 60
(d ppv/d ipv) = 0
ppv [W]

p [W]
50 50

40

pv
40

30 30

20 20

10 10

0 0
0 5 10 15 20 0 1 2
ipv [A] 3 4 5
vpv [V]

Figure
Figure4.4.Power
Power curves ofaaPV
curves of PVmodule.
module.

Taking into
Taking account
into account . . i pv
thatthat p pv = v pv , the
, thefollowing
followingrelation holds
relation at at
holds thethe
MPP:
MPP:

dp pv d ` . ˘ dv pv 0 (5)
“ v pv .i pv “ v pv ` i pv “0 (5)
di pv di pv di pv
Then, the following relation defines the maximum power condition, i.e., at the MPP:
Then, the following relation defines the maximum power condition, i.e., at the MPP:
0 for and (6)
dv pv v pv
` “ 0 for v pv “ v MPP and i pv “ i MPP (6)
di pv Equation
To correlate the MPP condition i pv (4) with the time-varying signals of the DC/DC converter,
the condition in Equation
To correlate the MPP(6) condition
is expressed in terms(4)
Equation of with
the time
the derivatives of both
time-varying the of
signals PVthe
voltage
DC/DCand
current, obtaining expression Equation (7):
converter, the condition in Equation (6) is expressed in terms of the time derivatives of both the PV
voltage and current, obtaining⁄ expression Equation (7):

0 for and (7)
dv pv {dt v pv
From such an expression, “ 0 for
the `proposed v pv “ v MPP
switching function i pv and
and Ψ “ i MPP
sliding surface Φ , given (7)
in
di pv {dt i pv
Equation (8), are defined:
From such an expression, the proposed switching function Ψ and sliding surface Φ, given in

Equation (8), are defined: Ψ ∧ Φ Ψ 0 (8)
dv pv⁄{dt v pv
Ψ“ ` ^ Φ “ tΨ “ 0u (8)
di pv {dt i pv
Then, the state-space model of both the PV system and SMC is formed by the simultaneous
Equations
Then,(1)–(3) and (8). The
the state-space modelrelations
of bothwithin such
the PV a state-space
system and SMCmodel are illustrated
is formed in the block
by the simultaneous
diagram of Figure 5, which put in evidence the signals processing and variables exchangedthe
Equations (1)–(3) and (8). The relations within such a state-space model are illustrated in block
between
diagram of Figure 5, which put in evidence the signals processing and variables exchanged between
the PV system and the SMC: the PV system disturbances are the short-circuit current (defined by the
the PV system and the SMC: the PV system disturbances are the short-circuit current (defined by the
irradiance and temperature) and the load voltage; the control signal is generated by the SMC on the
irradiance and temperature) and the load voltage; the control signal u is generated by the SMC on the
basis
basisof
of the
the PV current and
PV current and voltage;
voltage; and
and the
the SMC derives both
SMC derives both the
the PV
PVvoltage
voltageand
and current
current toto construct
construct
the
theswitching
switchingfunction.
function. Finally,
Finally, the
the SMC includes aa comparator
SMC includes comparator to to implement
implement thethe sign
sign function
function that
that
triggers
triggersthethechanges
changes on
on the
the signal
signal u.. The
The design
design of such aa comparator
of such comparator is is described
described inin the
the following
following
section in terms of the transversality and reachability conditions.
section in terms of the transversality and reachability conditions.

12368
Energies 2015, 8, 12363–12387
Energies 2015, 8 8

Figure 5. Block diagram of the SMC and PV system.


Figure 5. Block diagram of the SMC and PV system.
4. Analysis of the Sliding-Mode Controller
4. Analysis of the Sliding-Mode Controller
The SMC must fulfill three conditions to guarantee stability and a satisfactory performance:
The SMC mustreachability
transversality, fulfill three
and conditions to guarantee
equivalent control stability andanalyses
[32]. The transversality a satisfactory performance:
the controllability
of the system, the reachability analyses the ability of the closed-loop system to reach the surface, and
transversality, reachability and equivalent control [32]. The transversality analyses the controllability
the equivalent control grants local stability.
of the system,
Those the reachability
conditions analyses
grant the theofability
existence of the closed-loop
the sliding-mode, which alsosystem
imposesto the
reach the surface,
conditions
and the equivalent
defined control
in Equation (9)grants
[33]: local stability.

Those conditions grant the existence of the Ψ“ sliding-mode,
0 ^ “ 0 which also imposes the conditions (9) defined
dt
in Equation (9) [33]:
Such expressions provide information concerning the surface and its derivative. In that way, the
Ψ (10):
derivate of the switching function is given in Equation
Ψ 0Λ 0 (9)
d2 v d2 i
ˆ ˙

pv pv pv pv dv di
“ 2 i pv ` 2 v pv ` 2 ˚ (10)
Such expressions provide information
dt d t concerning
d t the surface
dt dtand its derivative. In that way,
the derivate of the switching function is given in Equation (10):
To analyze expression Equation (10) it is required to also derive expressions Equations (2) and
Ψ Equations (11)–(13), which are components of Equation (10):
(3), it leading to expressions
2 ∗ (10)
d2 v pv
ˆ ˙
1 disc dv pv Av pv di L
“ ´ BA e ´ (11)
To analyze expression Equation
d2 t(10) it
Cinis required
dt todtalso derivedtexpressions
Equations (2) and (3),
it leading to expressions Equations (11)–(13),
d i pv which
di
aredvcomponents
pv
of Equation (10):
sc
“ ´ BA e Av pv (12)
1dt dt dt
d2 i pv d2 v pv
˙2 (11)
d2 isc
ˆ
dv pv
2
“ 2 ´ BA 2 e Av pv ´ BA2 e Av pv (13)
d t d t d t dt

Then, the small-signal admittance of the PV module y “ ´BAe Av pv is introduced into the(12)
previous equations to provide more compact expressions:

(13)
d2 v pv
ˆ ˙
1 disc dv pv di L
“ `y ´ (14)
d2 t Cin dt dt dt
Then, the small-signal admittance of the PV module y is introduced into the previous
equations to provide more compact expressions:
12369
Energies 2015, 8, 12363–12387

d i pv disc dv pv
“ `y (15)
dt dt dt
˙2
d2 i pv d2 isc d2 v pv
ˆ
dv pv
“ ` y ` Ay (16)
d2 t d2 t d2 t dt
Replacing Equations (1) and (14)–(16) into Equation (10), and performing some mathematical
manipulations, the following expression for the switching function derivate is obtained:
´ ¯ ´ ¯
dΨ disc i pv `y v pv v ´v p1´uq i pv `y v pv
dt “ dt C ´ pv Lb Cin
´in ´ ¯ ¯
dv pv i pv `y v pv dv pv dv pv
` dt y Cin ` yAv pv dt ` 2 didtsc ` 2y dt
(17)
2
` v pv dd2isct

This new expression is used to analyze the transversality, reachability and equivalent control
conditions in the following subsections.

4.1. Transversality Condition


To ensure the ability of the controller to act on the system dynamics, the transversality condition
given in Equation (18) should be granted [33]:
ˆ ˙
d dΨ
‰0 (18)
du dt

Deriving Equation (17) with respect to the signal u leads to expression Equation (19):
ˆ ˙ ˆ ˙
d dΨ v i pv ` y v pv
“´ b (19)
du dt L Cin

The analysis of Equation (19) requires to review the condition in which the MPP occurs: taking
into account that p pv = v pv ¨ i pv , condition Equation (4) corresponds to expression Equation (20).
Therefore, expressions Equations (18) and (19) lead to the transversality condition Equation (21):

dp pv ´ ¯
“ v pv ¨ ´BAe Av pv ` i pv “ v pv .y ` i pv “ 0 (20)
dv pv
ˆ ˙
d dΨ
‰ 0 i f v pv ‰ v MPP (21)
du dt
Figure 6 shows the simulation of Equation (19) considering the following conditions: a BP585 PV
module with parameters B = 0.894 µF and A = 0.703 V´1 , and a DC/DC converter with L = 100 µH,
Cin = 44 µF and vb = 24 V. The simulation illustrates the transversality condition provided by
Equation (21):

‚ At the left of the MPP (voltages lower than v MPP ) the transversality condition is fulfilled, hence
the SMC is able to act on the PV system to drive it towards the MPP.
‚ At the right of the MPP (voltages higher than v MPP ) the transversality condition is also fulfilled.
‚ At the MPP (v pv = v MPP ) the transversality condition is not fulfilled. This is not a problem
because the PV system is already at the MPP. Moreover, if the system diverges from the MPP
the transversality condition is fulfilled.

12370
 At the right of the MPP (voltages higher than ) the transversality condition is also fulfilled.
 At the MPP ( ) the transversality condition is not fulfilled. This is not a problem
because
Energies 2015, the PV system
is already at the MPP. Moreover, if the system diverges from the MPP
8, 12363–12387
the transversality condition is fulfilled.

80

60

ppv [W]
MPP
40

20

0
0 2 4 6 8 10 12 14 16 18 20 22
40
Transversality

No transversality
20

0
Positive
Negative transversality transversality
-20
0 2 4 6 8 10 12 14 16 18 20 22
Voltage [V]

Figure 6. Simulation of the transversality condition.


Figure 6. Simulation of the transversality condition.
The previous conditions impose a behavior similar to the classical hysteresis implementation
Theofprevious conditions impose a behavior similar to the classical hysteresis implementation of the
the SMC for DC/DC converters: the system oscillates around the surface forming a hysteretic
SMC trajectory.
for DC/DC converters:
In the followingthe system oscillates
subsections around the surface
it will be demonstrated that suchforming a hysteretic
hysteresis trajectory.
band is imposed
In the following subsections it will be demonstrated that such hysteresis band is imposed by the
by the peak-to-peak amplitude of the voltage ripple at the PV module terminals.
peak-to-peak amplitude
4.2. Equivalent ofCondition
Control the voltage ripple at the PV module terminals.

The equivalent
4.2. Equivalent control condition analyzes the ability of the system to remain trapped inside
Control Condition
the surface [31,32]. This condition is analyzed in terms of the equivalent analog value ueq of the
discontinuous control signal u: if such an equivalent value is constrained within the operational
The equivalent control condition analyzes the ability of the system to remain trapped inside the
limits the system will not be saturated and its operation inside the surface is possible. Since the
surface [31,32].
control signalThis condition tois the
u corresponds analyzed
MOSFETinactivation
terms ofsignal,
the equivalent
its operational analog
limits value
are 0 and 1.of the
discontinuous
Moreover,control signal
since the : ifconsiders
analysis such an the
equivalent valuethe
system inside is constrained within thedtoperational
surface, the condition dΨ
“ 0 holds.limits
Then, matching expression Equation (17) to zero, and replacing u = u
the system will not be saturated and its operation inside the surface is possible. eq , enables to obtain the ueqsignal
Since the control
value given in Equation (22):
u corresponds to the MOSFET activation signal, its operational limits are 0 and 1. Moreover, since the
C
analysis considers the system inside
ueq “ 1the
v
surface,
´ vpv ` v1 didtscthe
` vcondition
y dv pv
dt `
in 0 holds.
dv pv
dt
Then, matching expression
b b b vb pi pv
¯`y v pv q
Equation (17) to zero, and replacing ˚ yAv pv, enables tosc obtain the value given in Equation (22):
´
dv pv di dv pv
dt ` 2 dt ` 2y dt ˙ (22)
1 `
Cin v pv d2 isc
1 vb pi pv `yv pv q d2 t

Such an expression can be rewritten as Equation (23), where u1 and u2 are given in Equations(22)

(24) and (25), respectively: 2 2
ueq “ u1 ` u2 (23)
v pv
Such an expression can be rewritten as Equation (23),
u1 “ 1 ´ where and are given in Equations
(24) (24)
vb
and (25), respectively:
Cin v pv d2 isc
„ ˆ ˙ 
1 disc dv pv Cin dv pv dv pv disc dv pv
u2 “ `y ` yAv pv `2 ` 2y ` (25)
vb dt dt i pv ` y v pv dt dt dt dt i pv ` y v pv d2 t (23)

1 (24)

12371
Energies 2015, 8, 12363–12387

The analysis of Equation (23) must be addressed by writing the derivative of Equation (8) in a
different way as given in Equation (26):
ˆ ˙ ¯2 ´
d2 v pv di pv d2 i pv dv pv
´ ¯
2 di pv dv pv di pv
i pv d2 t dt ´ d2 t dt
` dt dt i pv ´ dt .v pv

“ ¯2 (26)
dt
´
di pv
i pv 2 ˚ dt

Since the solar irradiance does not exhibit fast changes in comparison with the switching
frequency of the converter, it is assumed constant ( didtsc “ 0) to simplify the analysis of Equations
(17), (25) and (26), obtaining the simplified expressions Equations (27) and (28), respectively:
˜ ` ˘ ¸
dΨ dv pv i pv y ´ i pv A ´ y2 v pv
“ (27)
dt dt y ˚ i pv
ˆ ˙
y dv pv Cin dv pv ` ˘
u2 “ 1` Av pv ` 2 (28)
vb dt i pv ` y v pv dt
Taking into account that i pv ­“ 0 is a realistic condition since i pv = 0 does not occur at any MPP,
dΨ dv pv
the sliding-mode condition dt “ 0 is reached in Equation (27) for dt “ 0. Then, replacing those
dv pv
conditions dt“ 0 and i pv ­“ 0 into Equation (28) leads to u2 = 0.
Therefore, at low irradiance derivatives, expression Equation (23) for ueq is approximately
equal to u1 . To evaluate the validity of such an approximation, ueq from Equation (23) was
simulated at different irradiance derivatives and contrasted with u1 from Equation (24): the first test
considers a change from the highest irradiance possible on earth to a complete shade in 1 second,
i.e., 1 sun per second or dS/dt = 1 kW/(m2 ¨ s). The second test considers a derivative of
10 suns per s (dS/dt = 10 kW/(m2 ¨ s)), and the third test considers a derivative of 100 suns per s
(dS/dt = 100 kW/(m2 ¨ s)), which is very large. The tests results are presented in the Table 1, where it
is observed that the error generated by assuming ueq « u1 is less than 1% for the case with the largest
derivative, while the errors for the other cases are less than 0.1% and 0.01%.

Table 1. Error presented at different irradiance changes.

dS/dt Error (%) = ((ueq ´ u1 )/ueq ˆ 100%


1 kW /(m2 . s) 0.0075% < e < 0.008%
10 kW /(m2 . s) ´0.0912% < e < 0.04%
100 kW /(m2 . s) 0.989% < e < 0.6883%

Thus, the equivalent control analysis is based on expression Equation (29), which does not
introduce a significant error:
v pv
ueq « 1 ´ (29)
vb
Finally, since the equivalent control condition for a DC/DC converter application is 0 < ueq < 1,
such an expression is analyzed using the ueq value given in Equation (29), which leads to expression
Equation (30):
v pv
0 ă 1´ ă 1 Ñ 0 ă v pv ă vb (30)
vb
The condition in Equation (30) is always fulfilled since the analyzed PV system considers a boost
converter. Therefore, the equivalent control is fulfilled.

12372
Energies 2015, 8, 12363–12387

4.3. Reachability Conditions


The reachability conditions analyze the ability of the system to reach the desired condition Ψ = 0.
The work in [32] demonstrated that a system that fulfills the equivalent control condition also fulfills
the reachability conditions. That work also shows that the sign of the transversality condition imposes
the value of u for each reachability condition. A positive value of the transversality condition imposes
the following reachability conditions:
ˇ ˇ
dΨ ˇˇ dΨ ˇˇ
lim “ ą0 (31)
ΨÑ0´ dt u“1 dt ˇu“1,Ψ“0
ˇ
ˇ ˇ
dΨ ˇˇ dΨ ˇˇ
lim “ ă0 (32)
Ψ Ñ0` dt ˇu“0 dt ˇu“0,Ψ“0
Instead, a negative value of the transversality condition imposes the following
reachability conditions: ˇ ˇ
dΨ ˇˇ dΨ ˇˇ
lim “ ą0 (33)
ΨÑ0´ dt u“0 dt ˇu“0,Ψ“0
ˇ
ˇ ˇ
dΨ ˇˇ dΨ ˇˇ
lim “ ă0 (34)
ΨÑ0` dt u“1 dt ˇu“1,Ψ“0
ˇ

Moreover, the reachability conditions define the implementation of the switching law [32].
However, Section 4.1 shows that the transversality condition in this SMC is both positive and negative
depending on the operation condition. In such a way, from Figure 6 it is observed that the voltage
range for negative transversality (v pv < v MPP ) is much larger than the voltage range for positive
transversality (v pv > v MPP ); therefore the implementation and analysis of the proposed SMC is
performed in such a negative transversality condition, which imposes the following control law:
{Ψ < 0 Ñ u = 0 ^ Ψ > 0 Ñ u = 1}.
Then, the reachability of the surface is fulfilled in the following conditions: dΨ
dt ă 0 with u = 1
and dt ą 0 with u = 0. Moreover, from Equation (27) it is observed that the sign of dΨ

dt is the same
dv pv
sign of dt . Therefore, the analyzed reachability conditions are fulfilled in the following states:
dv pv
‚ The MOSFET is ON (u = 1) and the voltage is decreasing, i.e., dt ă 0.
dv pv
‚ The MOSFET is OFF (u = 0) and the voltage is increasing, i.e., dt ą 0.

It must be noted that a DC/DC converter always exhibits a periodic voltage ripple around an
average value [34]. In fact, the PV voltage corresponds to the voltage at the input capacitor Cin in
Figure 3, which voltage ripple is defined by the second order filter formed by Cin and the inductor L.
From the first differential equation of the DC/DC converter, i.e., Equation (1), it is observed that
the inductor current exhibits an almost triangular waveform as in any other second-order filter;
hence in steady-state the capacitor current has a triangular waveform centered in zero, again as in
any other second order filter [30]. Therefore, the capacitor current increases with the MOSFET OFF
(u = 0). This behavior is observed in the simulation of Figure 7, which includes the proposed SMC
with the control action implemented as u = 1 for Ψ < 0 and u = 0 for Ψ > 0, and adopting the same
electrical parameters previously described in Section 4.1. The implementation of the SMC is described
in Section 5. The simulation also confirms that the capacitor current decreases with the MOSFET ON
(u = 1). Moreover, it is noted that with a constant value for u (u = 0 or u = 1) the capacitor current
exhibits positive, negative and zero values.

12373
This behavior is observed in the simulation of Figure 7, which includes the proposed SMC with the
control action implemented as 1 for Ψ 0 and 0 for Ψ 0, and adopting the same electrical
parameters previously described in Subsection 4.1. The implementation of the SMC is described in
Section 5. The simulation also confirms that the capacitor current decreases with the MOSFET ON
1).2015,
(Energies Moreover, it is noted that with a constant value for (
8, 12363–12387 0 or 1) the capacitor current
exhibits positive, negative and zero values.
18.6
d vpv /dt > 0 d vpv /dt = 0 d vpv /dt < 0

vpv [V]
18.4

18.2
10 0.0203 0.0204 0.0204 0.0204 0.0204 0.0204

iL [A] 5

0
2 0.0203 0.0204 0.0204 0.0204 0.0204 0.0204
iC [A]

-2
0.0203 0.0204 0.0204 0.0204 0.0204 0.0204
1
u [-]

0.5
0
0.0203 0.0204 0.0204 0.0204 0.0204 0.0204
Time [s]

Figure
Figure7.7.PV
PV system dynamicbehavior.
system dynamic behavior.

This
Thismeans
means that,
that,asasreported
reportedbybythe thesecond
seconddifferential
differential equation
equation of of the DC/DC converter,
the DC/DC i.e.,
converter, i.e.,
Equation (2), the voltage ripple is a series of quadratic waveforms generated from the integral of the
Equation (2), the voltage ripple is a series of quadratic waveforms generated from the integral of
the capacitor
capacitor current
current iCin =iCin
since since = iipv
ipv − L. In i L . Ina way,
´ such such the
a way,PV the PV voltage
voltage exhibits exhibits
positive positive and
and negative
negative derivatives for both states of u as reported in Figure 7. Therefore, since
derivatives for both states of as reported in Figure 7. Therefore, since the switching function derivative the switching
function derivative has the same sign of the PV voltage derivative, as reported in Equation (27), the
has the same sign of the PV voltage derivative, as reported in Equation (27), the reachability conditions
reachability conditions are fulfilled in a fraction of the time intervals in which the MOSFET is in ON
are fulfilled in a fraction of the time intervals in which the MOSFET is in ON and OFF states, which
and OFF states, which means that in the other fractions of the time intervals the system diverges from
means that inSuch
the surface. the conditions
other fractions
createofa the time behavior
periodic intervals in
thewhich
systemthediverges from the surface.
system converges Such
and diverges
conditions
to/from the create a periodic behavior in which the system converges and diverges to/from the surface.
surface.
ToToillustrate
illustratethe the
periodic behavior
periodic of theofreachability
behavior conditions,
the reachability and theand
conditions, resulting periodic periodic
the resulting behavior
of the SMC,
behavior Figure
of the SMC, 8 shows
Figuretwo diagrams:
8 shows A state flow
two diagrams: describing
A state the convergence
flow describing and divergence
the convergence and
divergence states in Figure 8(a), and the associated time-depending signal of the
states in Figure 8(a), and the associated time-depending signal of the switching function in Figure 8(b). switching function
in Figure 8(b). For the sake of illustration, the system is considered at the beginning of the analysis
For the sake of illustration, the system is considered at the beginning of the analysis underdΨ the surface, i.e.,
under the surface, i.e., Ψ < 0, and with a negative switching function derivative, i.e., dt ă 0, which
Ψ 0 , and with a negative switching function dvderivative,
pv
i.e., 0 , which corresponds to a
corresponds to a negative PV voltage derivative dt ă 0 as reported in Equation (27). Due to the
dv
differential equation governing the capacitor “ iCin , the condition dΨ
Cin dtpv dt ă 0 also corresponds
to a negative input capacitor current, i.e., iCin ă 0. Those initial conditions are observed at the top of
Figure 8(a). Then, due to the control action implemented for the SMC, the control signal is set to u = 0
and the system enters in the first state of operation:

1. Diverge from the MPP: u = 0, Ψ < 0, dΨdt ă 0 (iCin ă 0). In this first stage (block 1 in Figure 8(a))
the reachability conditions are not fulfilled, hence the system diverges from the MPP, which
is observed in Figure 8(b). Such a condition is caused by the negative value of the switching
function derivative, which reduces even more the value of Ψ, hence the desired condition Ψ = 0
is not achievable. However, the control action u = 0 forces the increment of iCin , it driving the
system to iCin “ 0 ( dΨ dΨ
dt “ 0) and eventually to iCin ą 0 ( dt ą 0), which corresponds to the
second stage of the SMC operation.
2. Converge to the MPP: u = 0, Ψ < 0, dΨ dt ą 0 (iCin ą 0). In this second stage the reachability
condition Equation (33) is fulfilled, hence the system converges to the MPP, which is also
observed in Figure 8(b). In this case, the positive value of the switching function derivative
increases the value of Ψ, it driving the system towards the desired condition Ψ = 0. However,
when the condition Ψ = 0 is reached the switching function derivative is still positive, i.e.,
dt ą 0 (iCin ą 0), hence Ψ becomes positive. Moreover, when Ψ ą 0 the SMC imposes u “ 1,

which corresponds to the third stage of the SMC operation.

12374
Energies 2015, 8, 12363–12387

3. Diverge from the MPP: u = 1, Ψ ą 0, dΨ dt ą 0 (iCin ą 0). In this third stage the reachability
conditions are not fulfilled, hence the system diverges from the MPP. Such a condition is caused
by the positive value of the switching function derivative, which increases even more the value
of Ψ, hence the desired condition Ψ “ 0 is not achievable. However, the control action u “ 1
forces the decrement of iCin , it driving the system to iCin “ 0 ( dΨ
dt “ 0) and eventually to iCin ă 0

( dt ă 0), which corresponds to the fourth stage of the SMC operation.
4. Converge to the MPP: u “ 1, Ψ ą 0, dΨ dt ă 0 iCin ă 0). In this fourth stage the reachability
condition Equation (34) is fulfilled, hence the system converges to the MPP. In this case, the
negative value of the switching function derivative decreases the value of Ψ, it driving the
system towards the desired condition Ψ “ 0. However, when the condition Ψ “ 0 is reached
the switching function derivative is still negative, i.e., dΨ dt ă 0 (iCin ă 0), hence Ψ becomes
negative. Moreover, when Ψ ă 0 the SMC imposes u = 0, which corresponds to the first stage
of the SMC operation.
Energies 2015, 8 15
Limit cycle: Repeated periodically generating the hysterical behavior

(a)

(b)

Figure 8. Figure 8. behavior


Periodic Periodic behavior of the reachability
of the reachability conditions.
conditions. (a) Convergence
(a) Convergence and and divergence
divergence conditions;
conditions;
(b) Hysteretic (b) Hysteretic behavior.
behavior.

The previous four stages of the system operation are continuously repeated, they form a
limit-cycle [35] that imposes a hysteretic behavior to the SMC switching function around the
surface Ψ 0. Moreover, from Figure 8 it is noted
12375that the hysteresis band is defined by the condition
0, which corresponds to 0 and 0 as reported in Equation (27). This dependency is
Energies 2015, 8, 12363–12387

The previous four stages of the system operation are continuously repeated, they form a
Energies 2015, 8[35] that imposes a hysteretic behavior to the SMC switching function around the
limit-cycle 16
surface Ψ “ 0. Moreover, from Figure 8 it is noted that the hysteresis band is defined by the condition
dv pv

depicted in0,Figure
dt “ which 7, and confirmed
corresponds to dt by
“ 0Figure
and icin9,“which presentsinΨ,
0 as reported and(27).
Equation u generated by the same
This dependency
is depicted in Figure 7, and confirmed by Figure 9, which presents Ψ, dΨ
and u generated by the
simulation producing producing
same simulation the signalsthe
presented in Figurein7.Figure
signals presented In addition, Figure dt9Figure
7. In addition, also confirms the limit-cycle
9 also confirms the
of thelimit-cycle
switchingof function.
the switching function. Finally, it is noted that the hysteresis band of the SMC is definedby the
Finally, it is noted that the hysteresis band of the SMC is defined
voltage
by ripple at theripple
the voltage inputatcapacitor, which can
the input capacitor, be modified
which by changing
can be modified the input
by changing capacitance.
the input capacitance.

0.2 Divergence Condition


 [-] 0
Reachabality
-0.2
Condition
-0.4

0.0203 0.0203 0.0203 0.0203 0.0203 0.0204


4
x 10
5

d /dt [-] 0

-5
0.0203 0.0203 0.0203 0.0203 0.0203 0.0204
1
u [-] 0.5
0
0.0203 0.0203 0.0203 0.0203 0.0203 0.0204
Time [s]

Figure 9. SMC hysteretic behavior.


Figure 9. SMC hysteretic behavior.
4.4. Switching Frequency
4.4. Switching Frequency
The switching frequency is an important parameter in the operation of DC/DC converters, hence
it must be calculated
The switching frequencyin order
is antoimportant
provide practical guidelines
parameter in thefor the SMCof
operation implementation.
DC/DC converters, hence it
The switching frequency is calculated from the DC/DC converter differential equations and
must be calculated in order to provide practical guidelines for the SMC implementation.
ripple magnitudes as reported in [30]. In such a way, in a boost converter the ripple magnitude for
Thetheswitching frequency
inductor current is calculated
and capacitor from
voltage the DC/DC
are given converter
in Equations differential
(35) and equations
(36), respectively, andTsripple
where
magnitudes as the
represents reported in period,
switching [30]. Ind represents
such a way, in acycle,
the duty boostandconverter the
Fsw =1/Ts is theripple magnitude
switching frequency:for the
inductor current and capacitor voltage are given
v pv in Equations v pv (35) and (36), respectively, where Ts
∆i L “ dTs , d “ 1 ´ (35)
represents the switching period, d represents the
2L duty cycle, andvb Fsw =1/Ts is the switching frequency:

∆ ∆i (35)
∆v pvd“ , L Ts1 (36)
2 8Cin
∆ SMC system highlighting the time intervals (36)
Figure 10 presents a simulation of the closed-loop
in which the MOSFET is turned on (Ton ) ∆and turned 8 off (Toff ). In such a figure it is observed that
the magnitude of the voltage ripple ∆v pv is mesured from the average PV voltage v pv to the peak
Figure 10 presents
(maximum) voltagea vsimulation ofpvthe
pv´ max as ∆v “ vclosed-loop
pv´max ´ v pv . SMC
Then, system highlighting
approximating the time
the average intervals in
PV voltage
to the MPP voltage, the voltage ripple magnitude becomes ∆v
which the MOSFET is turned on (Ton) and turned off (Toff). In such a figure it is observed
pv « v pv ´ max ´ v MPP , where the that the
MPP
LambertW pexpp1qp B`isc qq´1
magnitude
voltageofis the voltage
calculated ripple
from ∆ (20)
Equation is mesured
as v MPP “from the average
A PV using
voltage to the peak
the LambertW
function [36].
(maximum) voltage as ∆ . Then, approximating the average PV voltage
to the MPP voltage, the voltage ripple magnitude becomes ∆ , where the MPP
voltage is calculated from Equation (20) as using the
LambertW function [36].
12376
Energies 2015, 8, 12363–12387
Energies 2015, 8 17

vpv-max
18.5

18.4 v
vpv [V] vpv
18.3

18.2

0.0197 0.0197 0.0197 0.0197 0.0197 0.0198 0.0198 0.0198 0.0198

1
u [-]

0.5 Ton Toff

0
0.0197 0.0197 0.0197 0.0197 0.0197 0.0198 0.0198 0.0198 0.0198
Time [s]

Figure 10. Principle for calculating the switching frequency.


Figure 10. Principle for calculating the switching frequency.
dv
The maximum PV voltage v pv´ max is achieved when dtpv “ 0, which also corresponds to dΨ “0
Theand
maximum PV voltage is achieved when 0, which also corresponds to dt
u = 1 as demonstrated in the previous subsection. Then, replacing those values in Equation (17),
0 and
u = 1and
as assuming
demonstrated in theEquation
isc constant, previous(37)subsection.
is obtained: Then, replacing those values in Equation (17),
and assuming constant, Equation (37) is
v pvobtained:
´max
ˆ
i pv ` y v pv´max
˙
0“´ (37)
L Cin
0 (37)
Finally, replacing expressions Equations (35) and (37) into Equation (36) with v pv « v MPP and
∆v pv «replacing
Finally, v pv´max ´ expressions
v MPP enables Equations
to calculate Ton as:and (37) into Equation (36) with
(35) and
∆ enables to calculate
g
f ˆ as: ˇ ˙
i pv ˇ
f 16LCin ´ v MPP
f
´ y ˇv “ MPP
e pv
Ton “ (38)
16 v MPP d
(38)
dv pv
Similarly, taking into account that the minimum PV voltage v pv´ min is achieved when “ 0, dt

dt “ 0 and u = 0, the voltage ripple measured from v pv´min to the average PV voltage can be
Similarly, takingas into
approximated ∆v pv account
« v pv´maxthat the . minimum
´ v MPP PV
Then, To f f is voltageas in Equation
calculated is achieved when
(39), where d1 = 1´d 0,
represents the complementary duty cycle:
0 and u = 0, the voltage ripple measured from to the average PV voltage can be
approximated as ∆
g
f . Then,ˆ
i ˇ is calculated as in Equation (39), where d′ = 1−d
ˇ ˙
f 16LCin ´ pv v
f
y ˇv “ MPP ´ MPP
represents the complementary duty
To f fcycle:

e pv
(39)
v MPP d1

Finally, the switching frequency16 imposed by the SMC is given in Equation (40). In such an
expression the duty cycle is evaluated as reported in Equation (35): (39)

1 1
Fsw “ “d (40)
Finally, the switchingTonfrequency
` To f f imposed the SMC ´ by
´v MPP is given in Equation
¯ (40). In such an
ˆ ˇ ˙
i pv ˇ
16LCin y ˇ
´b b
v pv “MPP vb vb
expression the duty cycle is evaluated as reportedv MPP
in Equation (35):vb ´v MPP ` v MPP
The accuracy 1 1 switching frequency of the PV system
of expression is illustrated by calculating the
simulated in Figure 10: Equation (40) predicts a switching frequency equal to 27.4 kHz, while the
simulation reports a switching16 frequency equal to 28.41 kHz, which correspond to an acceptable (40)
prediction error of 3.5%.

12377
Energies 2015, 8, 12363–12387

Moreover, from expression Equation (40) it is evident that the switching frequency depends
on the inductance and capacitance values, PV module characteristics and system operating point.
Therefore, if Cin or L change due to aging or by other effects such as inductor saturation, the switching
frequency will also change. However, the previous subsections demonstrate the robustness of the
proposed SMC to changes in Cin and L: the system will remain stable for any positive values of Cin
and L. In conclusion, tolerances in the converter parameters due to aging or practical saturations will
not compromise the system stability.

4.5. Sliding-Mode Dynamics


The sliding-mode dynamics provide information concerning the averaged voltage and current
behavior, which is useful to predict important time-based performance criteria such as settling
time. The sliding-mode dynamics consider the system within the sliding surface, hence expressions
Equation (7) hold, which leads to Equation (41):

dv pv v pv di pv
“´ . (41)
dt i pv dt

Similarly, from the second differential equation of the DC/DC converter, i.e., Equation (2), the
dv pv
relation i pv “ Cin dt ` i L is obtained, which derivative is given in Equation (42):

di pv d2 v pv di
“ Cin 2 ` L (42)
dt d t dt
Then, substituting Equations (1) and (41) into Equation (42), and naming the instantaneous PV
v pv
module impedance R pv “ i pv , the relation Equation (43) is obtained:

d2 v pv dv pv
R pv Cin L `L ` R pv v pv “ R pv vb d1 (43)
d2 t dt
Expressing Equation (43) in Laplace domain leads to the model reported in Equation (44):

v psq R pv vb
“ (44)
d1 psq R pv Cin L s2 ` L s ` R pv

Such a second-order model describes the closed-loop system a behavior around a given operation
condition, in a which the natural frequency is wn “ 1{ pCin Lq and the damping ratio is
ξ “ 1{p2R pv C 1{ pCin Lqq. From the classical second order system analysis [37], the settling time of
the PV voltage is, approximately, ts “ 6.4R pv Cin . To test such an estimation, the PV system controlled
by the SMC, considering the parameters previously described in Section 4.1, was simulated in both
step-up and step-down irradiance transients. The first test considers a step change in the irradiance
from 600 W/m2 to 1000 W/m2 , which corresponds to a change in isc from 3 A to 5 A, obtaining
the results presented in Figure 11(a): in such conditions the average PV module impedance is 5.1 Ω,
which leads to ts = 1.4 ms. To provide a more clear calculation of the settling time from the DC/DC
converter waveforms, the PV voltage is also filtered to remove the switching ripple, which enables to
verify the accuracy of the proposed approximation. Similarly, the second test considers a step change
in the irradiance from 1000 W/m2 to 600 W/m2 , obtaining the results presented in Figure 11b, where
the settling time estimation is the same, i.e., ts = 1.4 ms. Both simulations confirm the validity of the
approximation. It must be noted that the settling time will change if Cin or L change, e.g., due to
aging or practical saturations.

12378
enables to verify the accuracy of the proposed approximation. Similarly, the second test considers a
step change in the irradiance from 1000 W/m2 to 600 W/m2, obtaining the results presented in Figure 11b,
where the settling time estimation is the same, i.e., ts = 1.4 ms. Both simulations confirm the validity of
the approximation. It must be noted that the settling time will change if or change, e.g., due to aging
Energies 2015, 8, 12363–12387
or practical saturations.
19
Unfiltered signal
18.5 Filtered signal

vpv [V]
18
Predicted ts
17.5 ts = 1.4 ms
17
0.028 0.0285 0.029 0.0295 0.03 0.0305 0.031 0.0315 0.032
2 2
(a) Transient from 600 W/m to 1000 W/m
19

18
vpv [V]

17
Predicted ts
16 ts = 1.4 ms
0.028 0.0285 0.029 0.0295 0.03 0.0305 0.031 0.0315 0.032
Time [s]
2 2
(b) Transient from 1000 W/m to 600 W/m

Figure 11.11.Settling-time
Figure estimation.
Settling-time estimation.

5. Numerical Results and Performance Evaluation


5. Numerical Results and Performance Evaluation
The proposed SMC was implemented in the power electronics simulator PSIM using C language.
TheThis
proposed
procedureSMCenables
was implemented in the power
to tests, realistically, electronics simulator
the performance of the SMC PSIM using C language.
by emulating the
implementation of the controller in a Digital Signal Processor (DSP). In addition,
This procedure enables to tests, realistically, the performance of the SMC by emulating the implementation the ANSI C code
used in the PSIM simulation (within a C block) can be directly executed in any DSP without
of the major
controller in a Digital Signal Processor (DSP). In addition, the ANSI C code used in the PSIM
modifications.
simulation Figure
(within12a presents
C block)the cansimulation
be directly executed
scheme and in anyimplementation,
SMC DSP without major which modifications.
uses four devices to
Figure 12 presents
implement the simulation
the controller: a C block scheme
emulating andtheSMC implementation,
DSP, two classical comparators whichand uses four
a S-R devices to
flip-flop
to store the u signal. Such a structure is widely adopted in literature
implement the controller: a C block emulating the DSP, two classical comparators and a S-R flip-flop to implement the switching
function due to its simplicity and reliability [32]. However, since in this work the SMC performs
to store the u signal. Such a structure is widely adopted in literature to implement the switching
both MPPT and voltage regulation actions, there is no need of cascade (or any other) controllers or
function due to itsFigure
modulators. simplicity
12 alsoand reliability
presents the ANSI[32].C However,
code used to since in this
calculate thework the SMC
switching performs
function, which both
MPPTis and voltage
provided regulation
by using an emulatedactions, there is no Converter
Digital-to-Analog need of (DAC) cascade (orswitching
to the any other) controllers
circuit driving or
the MOSFET. To test the SMC in the same conditions used to illustrate the mathematical analysis, this
modulators. Figure 12 also presents the ANSI C code used to calculate the switching function, which is
simulation scheme considers a DC-link voltage near to 30 V imposed by the standard voltage source
provided
modelby representing
using an emulated Digital-to-Analog
the inverter. Converterscheme
An additional simulation (DAC)will to be
theused
switching
afterwardscircuit driving
to test the the
MOSFET.SMC, To testhigh
under theboosting
SMC inconditions,
the sameand conditions
interacting used
withtoa illustrate the mathematical
detailed grid-connected analysis, this
inverter.
simulationFigure
scheme 13 considers
shows twoatests DC-link voltage
performed near
in the to simulation
first 30 V imposed scheme: by Figure
the standard voltagethe
13(a) presents source
dynamic response of the system for a change in the irradiance from 600 W/m2 to 1000 W/m2 ,
model representing the inverter. An additional simulation scheme will be used afterwards to test the
while Figure 13(b) presents the dynamic response of the system for a change in the irradiance from
SMC, 1000
underW/m high2 to
boosting
600 W/m conditions,
2 . In bothand casesinteracting with a the
the SMC drives detailed grid-connected
PV system to the optimalinverter.
operation
Figure 13 shows
condition within twothetests performed
estimated in time.
settling the first simulation
Moreover, suchscheme:
simulations Figurealso13(a)
show presents
the pointsthein dynamic
the
2 2 condition; then
response of the system for a change in the irradiance from 600 W/m to 1000 W/m , while Figure 13(b)
I-V curve in which the PV system operates: (1) corresponds to the initial steady-state
condition (2) is triggered by an irradiance perturbation; and (3) is the new MPP detected by2 the SMC. 2
presents the dynamic response of the system for a change in the irradiance from 1000 W/m to 600 W/m .

12379
In
In both
both cases
cases the
the SMCSMC drives
drives the
the PV
PV system
system to
to the
the optimal
optimal operation
operation condition
condition within
within the
the estimated
estimated
settling
settling time.
time. Moreover,
Moreover, such such simulations
simulations also
also show
show the
the points
points in
in the
the I-V
I-V curve
curve in
in which
which the
the PV
PV system
system
operates:
operates: (1)
(1) corresponds
corresponds to to the
the initial
initial steady-state
steady-state condition;
condition; then
then condition
condition (2)
(2) is
is triggered
triggered by
by an
an
Energies 2015, 8, 12363–12387
irradiance perturbation; and (3) is the new MPP detected
irradiance perturbation; and (3) is the new MPP detected by the SMC. by the SMC.

Figure 12.
12. Simulation
Figure
Figure 12. Simulationscheme
Simulation scheme and
scheme and SMC
andSMC
SMC implementation.
implementation.
implementation.
19
19
18
18 (1)
(1) (3)
[V]

(3)
Voltage[V]

17
17 18.5 (2)
18.5 (2)
Voltage

18
16
16
18
17.5
17.5
15
17
17
tts=1.4ms
=1.4ms
15 16.5
16.50.0595
s
0.06 0.0605 0.061 0.0615
14 0.0595 0.06 0.0605 0.061 0.0615
14
0.02 0.03 0.04 0.05 0.06 0.07 0.08
0.02 0.03 0.04 0.05 0.06 0.07 0.08
100
100
[W]
Power[W]
Power

50
50
0.02
0.02 0.03
0.03 0.04
0.04 0.05
0.05 0.06
0.06 0.07
0.07 0.08
0.08
Time
Time [s]
[s]
100
100
[W]
Power[W]

(2)
(2) (3)
(3)
50
50
Power

(1)
(1)
00
22 44 66 88 10
10 12
12 14
14 16
16 18
18 20
20 22
22
Voltage
Voltage [V]
[V]
(a)
(a)
Figure 13. Cont.
Figure
Figure 13.
13. Cont.
Cont.

12380
Energies 2015, 8 21
Energies 2015, 8, 12363–12387
Energies 2015, 8 21
19
18

Voltage[V]
19
19
18
(1)
17 (3)
18 17
(2)

Voltage[V]
19
16 16
18
(1)
17 (3)
15 15
17 ts = 1.4 ms
16 (2)
16 0.06 0.0605 0.061 0.0615
14
150.02 15 0.03 ts = 1.4 0.04
ms 0.05 0.06 0.07 0.08
100 0.06 0.0605 0.061 0.0615
[W] [W]

14
0.02 0.03 0.04 0.05 0.06 0.07 0.08
50
100
PowerPower

500
0.02 0.03 0.04 0.05 0.06 0.07 0.08
0 Time [s]
100
0.02 0.03 0.04 0.05 0.06 (1)
0.07 0.08
[W] [W]

Time [s] (3)


100
50 (1)
PowerPower

(2)
50
(3)
0
0 2 4 6 8 10 12 14 16 (2)
18 20 22
0
Voltage[V]
0 2 4 6 8 10 12 14 16 18 20 22
(b)
Voltage[V]
(b)
Figure 13. System Response to irradiance perturbations. (a) Transient from 600 W/m2 to
2 2 2
1000Figure
FigureW/m ; (b)
13.13.SystemTransient
System from
Response
Response to1000 W/m
irradiance
to irradiance to 600 (a)
W/m
perturbations.
perturbations. . (a) from
Transient Transient
2 to 1000 W/m2 ; 2
from
600 W/m 600 W/m to
(b) Transient from 1000 W/m2 to 600 W/m2 .
1000 W/m2; (b) Transient from 1000 W/m2 to 600 W/m2.
Taking into account that grid-connected PV systems exhibit perturbations at the DC-link, the SMC
must be also
Taking intotested
Taking into in
account presence
account
that of sinusoidal
that grid-connected
grid-connected voltage
PV systems
PV systems oscillations at Cb. Such
exhibit perturbations
exhibit perturbations
atathe
at the test is presented
DC-link,
DC-link,
the
the SMC in
SMC must be also tested in presence of sinusoidal voltage oscillations at Cb . Such a test is presented
Figurebe14,
must where
also testeda large 20 V peak-to-peak
in apresence sinusoidalvoltage
voltageoscillation hasatbeen superimposed topresented
a 29 V DC
in Figure 14, where large 20 Vofpeak-to-peak voltage oscillations
oscillation has beenCsuperimposed
b. Such a testtois a 29 V DC in
component, this
Figurecomponent, corresponding
14, where this
a large to a 69%
20 V peak-to-peak
corresponding perturbation. The
voltage oscillation
to a 69% perturbation. simulation
has been
The simulation alsoalso considers
superimposed an irradiance
to a 29 V DC
considers an irradiance
2 2 22
perturbation from
perturbation 1000
from W/m
1000 W/m to 600
to 600 W/m
W/m atattt =
= 0.03
0.03 s,s,
andand even
even under
under
component, this corresponding to a 69% perturbation. The simulation also considers an the the perturbation
perturbation in thein the load
load
irradiance
voltage
voltage the SMCthe SMC is able to impose the expected behavior to reach the new MPP.
perturbation from is1000
ableW/mto impose
2
to 600theW/m
expected
2
at t behavior
= 0.03 s,toand reach
eventheunder
new MPP.
the perturbation in the load
voltage the SMC is able
20to impose the expected behavior to reach the new MPP.
vpv [V]

20
18
vpv [V]

18
16

16 0 0.01 0.02 0.03 0.04 0.05 0.06


90
p0.05 ppv
800 0.01 0.02 0.03 0.04 max 0.06
90 49.5
ppv [W]

70 49 pmax ppv
80 85.5
60 85 49.5
48.5
ppv [W]

70 85.5
84.5 49
0.04 0.045 0.05 0.055
50 0.01 0.015 0.02
60 85 48.5
40 84.5 0.04 0.045 0.05 0.055
50 0 0.01
0.01 0.015 0.02
0.02 0.03 0.04 0.05 0.06
40
40
0 0.01 0.02 0.03 0.04 0.05 0.06
vb [V]v [V]

40
30
b

30
20
0 0.01 0.02 0.03 0.04 0.05 0.06
20 Time [s]
0 0.01 0.02 0.03 0.04 0.05 0.06
Time [s]
Figure 14. System response with load perturbation.
Figure 14. System response with load perturbation.

Figure 14. System response with load perturbation.

12381
Energies 2015,
Energies 2015, 88 22
22

As anticipated
As anticipated before,
before, aa secondsecond simulation
simulation scheme scheme is is used
used to to test
test the
the SMC
SMC underunder high high boosting
boosting
Energies 2015, 8, 12363–12387
conditions and
conditions and interacting
interacting with with aa detailed
detailed grid-connected
grid-connected inverter. inverter. Such Such aa scheme
scheme is is presented
presented in in
Figure 15
Figure 15 [24],
[24], which
which includes
includes aa closed-loop
closed-loop inverterinverter withwith two control
control objectives: provide provide a givengiven power
power
As anticipated before, a second simulation scheme istwo used to test objectives:
the SMC under highaboosting
factor and
factor and regulate
regulate the the DC-link
DC-link voltage.
voltage. The The inverter
inverter isis designed
designed to to operate
operate at at 110
110 VAC@60
VAC@60 Hz Hz with an an
conditions and interacting with a detailed grid-connected inverter. Such a scheme is presented with in
average
average DC-link
DC-link
Figure 15 [24],voltage
voltage equal
whichequal
includes to 220
to 220 V. The
V. The DC-link
a closed-loop DC-link
inverter is is formed
formed
with by
two control 14 F,
by objectives:
14 F, which
which
provideat ata1000
1000 W/m22
givenW/m
experiment aa large
power factor
experiment large
and 120
regulate
120 V V peak-to-peak
peak-to-peak
the DC-link voltage. voltage
voltage
The oscillation superimposed
inverter is designed
oscillation superimposed
to operatetoat the DC component,
110 VAC@60
to the DC component,
Hz
with an
this corresponding
this
average
corresponding2 to DC-link
to aa 55% voltage
55% perturbation. equal
perturbation. Moreover,to 220 V.
Moreover, the The DC-link
the input is formed
input capacitor was by C
was changed
changed
b “ 14 µF,
to which150 FF
150
at 1000 W/m experiment a large 120 V peak-to-peak voltagecapacitor
oscillation superimposed to to the DC
to avoid
to avoid large PV
large
component,
PVthis
voltage
voltage oscillations
oscillations
corresponding
due perturbation.
due
to a 55%
to the
to the high
high voltage
voltage
Moreover,
conversion
conversion ratio imposed
the input ratio
imposed
capacitor
to the
to
was changed
the DC/DC
DC/DC
to
converter.
converter. Therefore,
Cin “Therefore,
150 µF to avoid the SMC
the SMCPV
large switching
switching frequency,due
frequency,
voltage oscillations calculated
calculated
to the high from
from Equation
Equation
voltage (40),ratio
(40),
conversion is approximately
is approximately
imposed
100 kHz.
to theIn addition,
DC/DC the
converter. simulation
Therefore, the also SMCconsiders
switching an irradiance
frequency, perturbation
calculated from
from Equation 1000 is 22 to
(40),W/m
100 kHz. In addition, the simulation also considers an irradiance perturbation from 1000 W/m to
300 approximately
W/m 2 100 kHz. In addition, the simulation also considers an irradiance perturbation from
2 at t = 0.65 s to illustrate the MPPT operation of the SMC. The results are presented in
300 W/m 1000 W/m at t 2=to0.65 s to illustrate
300 W/m 2 at t = 0.65 thes toMPPT operation
illustrate the MPPT of operation
the SMC.of The results
the SMC. Theare presented
results are in
Figure 16,
Figurepresented which
16, which confirms
confirms
in Figure
that even
that even
16, which
under
under
confirms that
high
high DC-link
evenDC-link
under high
perturbations
perturbations the SMC
the SMC the
DC-link perturbations
is able
is able
SMCto
to reach
is reach
able the
the
new MPP.
to reach It must
the new be
MPP.point
It out
must that
be the
point outinverter
that the also imposes
new MPP. It must be point out that the inverter also imposes a transitory behavior in the the
inverter also a transitory
imposes a behavior
transitory in
behavior the
in DC-link
DC-link
voltage
voltage due to
DC-link
due tovoltage
the change
the change
due toin the
in PV PV power
change
power
in at
at tt == 0.65
PV power 0.65ats,s, which
t which
= 0.65 s,iswhich
also mitigated
is also mitigated by the
is also mitigated
by the bySMC.
SMC.
the SMC.

FigureFigure
Figure 15. Circuital
15. Circuital scheme
15. Circuital schemeof
scheme
of the detailed
the detailed
of the detailed grid-connected
grid-connected PV system.
system.
PV system.
grid-connected PV
20
20
[V]

18
v [V]

18
pv
vpv

16
16
0.4 0.5 0.6 0.7 0.8 0.9 1
0.4 0.5 0.6 0.7 0.8 0.9 1
100
100 Pmax Ppv
Pmax Ppv
[W]
p [W]

50
50
pv
ppv

0
00.4 0.5 0.6 0.7 0.8 0.9 1
0.4 0.5 0.6 0.7 0.8 0.9 1
300
300
[V]
vbv [V]

200
b

200

0.4 0.5 0.6 0.7 0.8 0.9 1


0.4 0.5 0.6 0.7 0.8 0.9 1
Time [s]
Time [s]

FigureFigure
Figure 16. SMC
16. SMC behavior
16. SMC behaviorinteracting
behavior with
interacting with
interacting with aa grid-connected
grid-connected
a grid-connected inverter.
inverter.
inverter.

12382
Energies 2015, 8 23
Energies 2015, 8 23
Energies 2015, 8, 12363–12387
With the aim of further evaluate the proposed SMC, the performance of this controller is also
contrasted
With thewith aimtwo of classical solutionsthe
further evaluate based on the SMC,
proposed P & O thealgorithm. The P of
performance & this
O method is based
controller on
is also
contrasted
perturb With
thewith the
inputtwo aim of further
classical
variable evaluate
solutions
of the the
system,based proposed
e.g., on
thethe SMC,
P&
duty the performance
O algorithm.
cycle of this controller
The P &observe
of the converter, O method is also
is based on
the variation in
contrasted with two classical solutions based on the P & O algorithm. The P & O method is
the output
perturb thepower, and increase
input variable of theorsystem,
decrease thethe
e.g., perturbed variable
duty cycle to converter,
of the increase the power.the
observe
based on perturb the input variable of the system, e.g., the duty cycle of the converter, observe the
The flowchart
variation in
of
thethe P & power,
output O algorithm
variation in theand is shown
increase
output power,orin
andFigure
decrease 17.
increasetheorperturbed
decrease thevariable to increase
perturbed theincrease
variable to power.theThe flowchart
power.
of the P & O algorithm is shown in Figure 17.
The flowchart of the P & O algorithm is shown in Figure 17.

Figure 17. Flowchart of the Perturb and Observe (P & O) algorithm.


FigureFigure
17. Flowchart
17. Flowchart ofofthe
thePerturb andObserve
Perturb and Observe (P &(P
O)& O) algorithm.
algorithm.
The structure of the two classical solutions based on the P & O algorithm are presented in Figure 18:
(a) The &The
a P structurestructure ofdefining
of the two
O algorithm the two the
classical
classical duty solutions
solutions of based
based
cycle on theonP the
a PWM &(P O&P algorithm
&O O+ algorithm
PWM) are presented
are presented
driving in 18:
in MOSFET;
the Figure
Figure 18: (a) a P & O algorithm defining the duty cycle of a PWM (P & O + PWM) driving the
and
(a) a(b)P a&P O& algorithm
O algorithm defining
defining thetheduty
voltage
cyclereference
of a PWM of a (PPID&controller,
O + PWM) which in turns
driving the defines
MOSFET;the
MOSFET; and (b) a P & O algorithm defining the voltage reference of a PID controller, which in turns
duty
and (b)cycle
a P of
defines a duty
&theO PWM (P of&
algorithm
cycle aO + PID
defining
PWM the&+voltage
(P OPWM). Such aSuch
+ PID reference
+ PWM). figure
of a aPID put in
putevidence
controller,
figure thethe
which
in evidence inadvantages
turns of the
defines
advantages
duty cycle
proposed of a PWM
of thesolution:
proposed the (P & the
SMC
solution: O +SMC
only PIDonly
+ PWM).
requires a singleSuch
requires a control
a control
single figure put
block,block, ininstead
instead evidence the advantages
the classical solutions
the classical of the
require
solutions
two require
and three
proposed two and
control
solution: three control
theblocks. blocks.
SMC only requires a single control block, instead the classical solutions require
two and three control blocks.
Closed-loop
Cin Cb grid-
PV dc/dc Closed-loop
connected
Cin Cb grid-
inverter
PV dc/dc connected
inverter

(a) It requires the use of


P&O PWM
cascade control
(a) It requires the use of
P&O PWM
(b) cascade control
(a) P&O + PWM
(b) PID
Control (a) P&O
(b) P&O ++ PID
PWM + PWM
PID
Control (b) P&O + PID + PWM

Figure 18. Structure of a conventional Maximum Power Point Tracking (MPPT) architecture.
Figure 18. Structure of a conventional Maximum Power Point Tracking (MPPT) architecture.
Figure 18. Structure of a conventional Maximum Power Point Tracking (MPPT) architecture.
The P The
& OP algorithms
& O algorithmswerewere designed
designed followingthe
following the guidelines
guidelines given
giveninin
[3,38], which
[3,38], are aimed
which to
are aimed to
guarantee a perturbation period larger than the PV voltage settling time. Concerning the stand-alone
guarantee
The P &a Operturbation
algorithmsperiod larger than
were designed the PV the
following voltage settlinggiven
guidelines time.inConcerning
[3,38], whichtheare
stand-alone
aimed to
guarantee a perturbation period larger than the PV voltage settling time. Concerning the stand-alone
12383
Energies 2015, 8 24

P & OEnergies
controller (a), the parameters are: perturbation period Ta = 0.5 ms and perturbation amplitude
2015, 8, 12363–12387
Δd = 0.2.
TheP design of the(a),
& O controller PIDthecontroller
parametersfor
are:the cascade period
perturbation P & OTasolution (b) perturbation
= 0.5 ms and was based on the system
amplitude
2
Equations
∆d =(1)–(3)
0.2. linearized at the lower irradiance condition (100 W/m ) as suggested in [3]. Then,
the transferThe
function
design between
of the PIDthe PV voltage
controller and
for the the duty
cascade P &cycle is (b)
O solution calculated as on
was based in the
Equation
system (45).
Equations (1)–(3) linearized at the lower irradiance condition (100 W/m2 ) as suggested in [3]. Then,
It is important to remark that changes with the solar irradiance and v temperature, hence the settling
the transfer function between the PV voltage and the duty cycle dpv is calculated as in Equation
time ts(45).
of the
It isPV voltage to
important constantly changes
remark that [39]: with the solar irradiance and temperature, hence the
R pv changes
settling time ts of the PV voltage constantly changes [39]:
(45)
v pv psq R pv vb
“´ (45)
d psq L Cin R pv s2 ` L s ` R pv
Therefore, conventional linear controllers for PV systems must be designed at the longest settling
2
time value,Therefore,
which is conventional
obtained forlinear
the larger valuefor
controllers ofPV systems
(lowermust
irradiance, 100atW/m
be designed in thissettling
the longest example).
time value, which is obtained for the larger value of R (lower irradiance, 100 W/m2 in this example).
Then, using the root-locus placement technique, the pv
controller in Equation (46) was designed.
Then, using the root-locus placement technique, the controller in Equation (46) was designed.
The perturbation period
The perturbation of the
period of Pthe&PO& algorithm
O algorithm is is
designed
designedfor
forthe
theworst
worst case scenario(longest
case scenario (longest )
following the guidelines
ts ) following given given
the guidelines in [3],in it[3],leading to to
it leading TaTa== 0.5 ms and
0.5 ms and∆v
Δv
refref= = 0.25
0.25 V (perturbation
V (perturbation
amplitude applied to the PID reference):
amplitude applied to the PID reference):

10.1302
10.1302 ps `2.25 4ps ` 1423q
2.25e4q 1423
Gc psq “ ´ (46) (46)
s ps `8.453
8.453e5q5

Finally, Finally,
to provideto provide a fair comparison,
a fair comparison, both P &both O +PPWM & O + PWM
and P &andO +PPID & O + PID were
+ PWM + PWM were
implemented
implemented with a switching frequency equal to 100 kHz, which is the same switching frequency
with a switching frequency equal to 100 kHz, which is the same switching frequency exhibited by the SMC.
exhibited by the SMC.
A first test was
A first testperformed considering
was performed a much
considering a much larger
largerDC-link capacitanceCb “ 200200
DC-link capacitance µF to F to reduce
reduce
the voltage oscillations
the voltage generated
oscillations by the
generated by inverter. This This
the inverter. test also includes
test also the the
includes start-up of the
start-up PVPV
of the system
and ansystem and an
irradiance irradiance perturbation
perturbation at t = 0.65ats.t =The
0.65simulation
s. The simulation
resultsresults are presented
are presented in Figure
in Figure 19,19,where
where the fast tracking of the MPP provided by the proposed SMC is observed. The simulation also
the fastreports
tracking of the MPP provided by the proposed SMC is observed. The simulation also reports
the small voltage oscillations exhibited due to the large value of Cb by the three systems,
the small
which voltage oscillations
enables the correctexhibited
operation due to Pthe
of the & large
O + PWM valuesolution.
of byFinally,
the three
the systems, which
fast response enables
of the
the correct
SMCoperation of the
is translated intoPa& O + PWM
higher energysolution.
harvestedFinally,
from the thePVfast response
array, whichof the SMCincreases
eventually is translated
the into
system profitability.
a higher energy harvested from the PV array, which eventually increases the system profitability.

SMC P&O+PWM P&O+PID+PWM MPP


20
vpv[ V]

18.5 17
10 18 16
17.5 15
0.4024 0.4026 0.4028 0.403 0.4032 0.7 0.702 0.704 0.706 0.708 0.71
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
100
18
85.5
ppv [W]

16
85 14
50 12
84.5 0.65 0.652 0.654 0.656
0.41 0.411 0.412 0.413 0.414 0.415
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

200
vb [V]

150
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time [s]

Figure 19. Performance comparison between the SMC and conventional P & O-based solutions
Figure 19. Performance comparison between the SMC and conventional P & O-based
considering small DC-link voltage oscillations.
solutions considering small DC-link voltage oscillations.

12384
Energies 2015, 8
Energies 2015, 8, 12363–12387
25

A second test was performed accounting for large voltage oscillations at the DC-link, i.e., using a
A second test was performed accounting for large voltage oscillations at the DC-link, i.e., using
smallera smaller14
Cb F.
“ The simulation
14 µF. results are
The simulation reported
results in Figure
are reported 20, where
in Figure again again
20, where the SMC exhibits a
the SMC
faster exhibits
MPPT aprocedure
faster MPPTwith higherwith
procedure energy
higherproduction. Moreover,
energy production. the simplest
Moreover, P & PO&solution
the simplest O
solution (P & O + PWM) is unstable since the DC-link voltage oscillations are transferred
(P & O + PWM) is unstable since the DC-link voltage oscillations are transferred to the PV array to the PV
array terminals, which confuses the P & O algorithm. In contrast, the more complex P & O solution
terminals, which confuses the P & O algorithm. In contrast, the more complex P & O solution
(P & O + PID + PWM) is stable, but it requires a long time to reach the MPP in comparison with
(P & O + SMC.
the PID + PWM) is stable, but it requires a long time to reach the MPP in comparison with the SMC.
25
vpv [V]

20
15
SMC P&O+PWM P&O+PID+PWM MPP
10
0.635 0.64 0.645 0.65 0.655 0.66

80 14.92
ppv [W]

60 14.91
40
20 0.6591 0.6592 0.6593 0.6594 0.6595
0
0.635 0.64 0.645 0.65 0.655 0.66
300
vb [V]

200

0.635 0.64 0.645 0.65 0.655 0.66


Time [s]

Figure 20. Performance comparison between the SMC and conventional P & O-based solutions
Figure 20. Performance comparison between the SMC and conventional P & O-based
considering large DC-link voltage oscillations.
solutions considering large DC-link voltage oscillations.
6. Conclusions
6. Conclusions
A detailed analysis and design of a MPPT solution based on sliding-mode control theory has
been introduced in this paper. This solution is aimed at performing a fast MPPT action on PV systems
A detailed analysis and design of a MPPT solution based on sliding-mode control theory has been
using a single control stage. Moreover, since it is based on a non-linear model, no linearization process
introduced in thisTherefore,
is needed. paper. This thissolution
approachis aimed
avoids at theperforming a fast MPPT
circular dependency action
among on PVcontrollers,
cascade systems using
a single control
reduces the stage.
number Moreover,
of controlsince it isand
devices based on a global
provides non-linear model,
stability in allnothelinearization process is
operation range.
needed. Therefore, this approach avoids the circular dependency among cascade controllers,and
Such characteristics are major improvements over classical MPPT solutions based on cascade reduces
linear controllers.
the numberThe of performance
control devices and provides global stability in all the operation range. Such characteristics
of the proposed solution was tested using realistic simulations accounting for
are major improvements
perturbations in bothover classical MPPT
the irradiance and thesolutions based
load voltage, on cascade
obtaining and linear
satisfactory controllers.
results. Moreover, the
The performance of the proposed solution was tested using realistic simulations
sliding-mode controller was contrasted with classical MPPT solutions based on the P & accounting for
O algorithm,
where the SMC provides a much faster tracking of the MPP, it increasing the produced energy.
perturbations in both the irradiance and the load voltage, obtaining satisfactory results. Moreover, the
Therefore, the proposed SMC enables to increment the profitability of PV installations; hence the
sliding-mode controller was
return-of-investment timecontrasted
is reduced. with classical MPPT solutions based on the P & O algorithm,
where theFinally,
SMC aprovides a much faster
further improvement to thetracking
SMC could ofbethe MPP, itin increasing
performed a future work—a the produced energy.
different, but
Therefore, the proposed
more complex, switchingSMC enables
circuit could tobe increment
designed to the
avoidprofitability of PV
the sign change of the installations; hence the
transversality—to
remove the divergence stages in the SMC operation. This new development will enable to constraint
return-of-investment time is reduced.
the hysteresis band of the SMC, which will eventually enable to reduce, even more, the settling time
Finally, a further
of the PV power improvement
to achieve a faster to the
MPPT SMC could be performed in a future work—a different, but
procedure.
more complex, switching circuit could be designed to avoid the sign change of the transversality—to
Acknowledgments: This work was supported by the Instituto Tecnológico Metropolitano, the Universidad
remove the divergence
Nacional de Colombiastages in the SMC
and Colciencias operation.
(Fondo nacional This new development
de financiamiento will laenable
para ciencia, to constraint
tecnología y la
the hysteresis
innovaciónband of the SMC, which will eventually enable to reduce, even more, the settlingand
Francisco José de Caldas) under the projects MicroRENIZ-25439 (Code 1118-669-46197), P14215
P14220, and under the study commission-agreement number 25 from Instituto Tecnológico Metropolitano.
time of
the PVAuthor
power to achieve a faster MPPT procedure.
Contributions: All the authors contributed equally to the work.

12385
Energies 2015, 8, 12363–12387

Conflicts of Interest: The authors declare no conflict of interest.

References

1. Lewis, N.S.; Crabtree, G.; Nozik, A.J.; Wasielewski, M.R.; Alivisatos, P. Basic Research Needs for Solar
Energy Utilization. In Basic Research Needs for Solar Energy Utilization; Department of Energy: Washington,
DC, USA, 2005.
2. Kish, G.J.; Lee, J.J.; Lehn, P.W. Modelling and control of photovoltaic panels utilising the incremental
conductance method for maximum power point tracking. IET Renew. Power Gener. 2012, 6, 259–266.
[CrossRef]
3. Femia, N.; Petrone, G.; Spagnuolo, G.; Vitelli, M. A Technique for Improving P & O MPPT Performances of
Double-Stage Grid-Connected Photovoltaic Systems. IEEE Trans. Ind. Electron. 2009, 56, 4473–4482.
4. Esram, T.; Chapman, P.L. Comparison of Photovoltaic Array Maximum Power Point Tracking Techniques.
IEEE Trans. Energy Convers. 2007, 22, 439–449. [CrossRef]
5. Gonzalez, D.; Ramos-Paja, C.A.; Petrone, G. Automated Procedure for Calculating the Controller
Parameters in Photovoltaic DC/DC Converters. Int. Rev. Electr. Eng. 2011, 6, 3027–3040.
6. Adinolfi, G.; Femia, N.; Petrone, G.; Spagnuolo, G.; Vitelli, M. Design of DC/DC Converters for DMPPT
PV Applications Based on the Concept of Energetic Efficiency. J. Sol. Energy Eng. Trans. ASME 2010, 132.
[CrossRef]
7. Qi, J.; Zhang, Y.; Chen, Y. Modelling and maximum power point tracking (MPPT) method for PV array
under partial shade conditions. Renew. Energy 2014, 66, 337–345. [CrossRef]
8. Petrone, G.; Ramos-Paja, C.A.; Spagnuolo, G.; Vitelli, M. Granular control of photovoltaic arrays by means
of a multi-output Maximum Power Point Tracking algorithm. Prog. Photovolt. Appl. 2012, 21, 918–932.
[CrossRef]
9. Ahmad, J. A fractional open circuit voltage based maximum power point tracker for photovoltaic arrays.
In Proceedings of 2nd International Conference on Software Technology and Engineering (ICSTE), San
Juan, Puerto Rico, 3–5 October 2010; pp. 247–250.
10. Ahmed, J.; Salam, Z. A Maximum Power Point Tracking (MPPT) for PV system using Cuckoo Search with
partial shading capability. Appl. Energy 2014, 119, 118–130. [CrossRef]
11. Khaehintung, N.; Kunakorn, A.; Sirisuk, P. A novel fuzzy logic control technique tuned by particle swarm
optimization for maximum power point tracking for a photovoltaic system using a current-mode boost
converter with bifurcation control. Int. J. Control Autom. Syst. 2010, 8, 289–300. [CrossRef]
12. Lin, W.-M.; Hong, C.-M.; Chen, C.-H. Neural-Network-Based MPPT Control of a Stand-Alone Hybrid
Power Generation System. IEEE Trans. Power Electron. 2011, 26, 3571–3581. [CrossRef]
13. Chiu, C.-S.; Ouyang, Y.-L. Robust Maximum Power Tracking Control of Uncertain Photovoltaic Systems:
A Unified T-S Fuzzy Model-Based Approach. IEEE Trans. Control Syst. Technol. 2011, 19, 1516–1526.
[CrossRef]
14. Petrone, G.; Spagnuolo, G.; Teodorescu, R.; Veerachary, M.; Vitelli, M. Reliability Issues in Photovoltaic
Power Processing Systems. IEEE Trans. Ind. Electron. 2008, 55, 2569–2580. [CrossRef]
15. Zitouni, N.; Khiari, B.; Andoulsi, R.; Sellami, A.; Mami, A.; Hssen, A. Modelling and non linear control of a
photovoltaic system with storage batteries: A bond graph approach. IJCSNS Int. J. Comput. Sci. Netw. Secur.
2001, 11, 105–114.
16. Bianconi, E.; Calvente, J.; Giral, R.; Mamarelis, E.; Petrone, G.; Ramos-Paja, C.A.; Spagnuolo, G.; Vitelli, M.
Perturb and Observe MPPT algorithm with a current controller based on the sliding mode. Int. J. Electr.
Power Energy Syst. 2013, 44, 346–356. [CrossRef]
17. Bianconi, E.; Calvente, J.; Giral, R.; Mamarelis, E.; Petrone, G.; Ramos-paja, C.A.; Spagnuolo, G.; Vitelli, M.
A Fast Current-Based MPPT Technique Employing Sliding Mode Control. IEEE Trans. Ind. Electron. 2013,
60, 1168–1178. [CrossRef]
18. Ghazanfari, J.; Farsangi, M.M. Maximum Power Point Tracking using Sliding Mode Control for
Photovoltaic Array. Iran. J. Electr. Electron. Eng. 2013, 9, 189–196.
19. Khemiri, N.; Khedher, A.; Mimouni, M.F. A sliding mode control approach applied to a photovoltaic
system operated in MPPT. In Proceedings of Systems, Signals & Devices (SSD), 2013 10th International
Multi-Conference, Hammamet, Tunisia, 18–21 March 2013; pp. 1–6.

12386
Energies 2015, 8, 12363–12387

20. Afghoul, H.; Chikouche, D.; Krim, F.; Beddar, A. A novel implementation of MPPT sliding mode controller
for PV generation systems. In Proceedings of EUROCON, 2013 IEEE, Zagreb, Croatia, 1–4 July 2013;
pp. 789–794.
21. Chu, C.-C.; Chen, H.-L. Robust maximum power point tracking method for photovoltaic cells: A sliding
mode control approach. Sol. Energy 2009, 83, 1370–1378. [CrossRef]
22. Romero-Cadaval, E.; Spagnuolo, G.; Garcia Franquelo, L.; Ramos-Paja, C.A.; Suntio, T.; Xiao, W.M.
Grid-Connected Photovoltaic Generation Plants: Components and Operation. IEEE Ind. Electron. Mag.
2013, 7, 6–20. [CrossRef]
23. Zhang, F.; Maddy, J.; Premier, G.; Guwy, A. Novel current sensing photovoltaic maximum power point
tracking based on sliding mode control strategy. Sol. Energy 2014, 118, 80–86. [CrossRef]
24. Petrone, G.; Spagnuolo, G.; Vitelli, M. A Multivariable Perturb-and-Observe Maximum Power Point
Tracking Technique Applied to a Single-Stage Photovoltaic Inverter. IEEE Trans. Ind. Electron. 2011, 58,
76–84. [CrossRef]
25. Trejos, A.; Gonzalez, D.; Ramos-Paja, C.A. Modeling of Step-up Grid-Connected Photovoltaic Systems for
Control Purposes. Energies 2012, 5, 1900–1926. [CrossRef]
26. Manganiello, P.; Ricco, M.; Petrone, G.; Monmasson, E.; Spagnuolo, G. Optimization of Perturbative PV
MPPT Methods Through Online System Identification. IEEE Trans. Ind. Electron. 2014, 61, 6812–6821.
[CrossRef]
27. Renaudineau, H.; Donatantonio, F.; Fontchastagner, J.; Petrone, G.; Spagnuolo, G.; Martin, J.-P.;
Pierfederici, S. A PSO-Based Global MPPT Technique for Distributed PV Power Generation. IEEE Trans.
Ind. Electron. 2015, 62, 1047–1058. [CrossRef]
28. Mamarelis, E.; Petrone, G.; Spagnuolo, G. An Hybrid Digital-Analog Sliding Mode Controller for
Photovoltaic Applications. IEEE Trans. Ind. Inf. 2103, 9, 1094–1103. [CrossRef]
29. Haroun, R.; Aroudi, A.E.; Cid-Pastor, A.; Garcia, G.; Olalla, C.; Martinez-Salamero, L. Impedance Matching
in Photovoltaic Systems Using Cascaded Boost Converters and Sliding-Mode Control. IEEE Trans.
Power Electron. 2015, 30, 3185–3199. [CrossRef]
30. Erickson, R.W.; Maksimovic, D. Fundamentals of Power Electronics, 2nd ed.; Springer: Berlin, Germany, 2007.
31. Sira-Ramirez, H.T. Sliding motions in bilinear switched networks. IEEE Trans. Circuits Syst. 1987, 38,
919–933. [CrossRef]
32. Gonzalez Montoya, D.; Ramos-Paja, C.A.; Giral, R. Improved design of sliding mode controllers based on
the requirements of MPPT techniques. IEEE Trans. Power Electron. 2015, 31, 235–247. [CrossRef]
33. Tan, S.-C.; Lai, Y.M.; Tse, C.K. General Design Issues of Sliding-Mode Controllers in DC/DC Converters.
IEEE Trans. Ind. Electron. 2008, 55, 1160–1174.
34. Park, J.-H.; Jeong, H.-G.; Lee, K.-B. Output Current Ripple Reduction Algorithms for Home Energy Storage
Systems. Energies 2013, 6, 5552–5569. [CrossRef]
35. Venturini, R.P.; Scarpa, V.V.R.; Spiazzi, G.; Buso, S. Analysis of limit cycle oscillations in maximum power
point tracking algorithms. In Proceedings of Power Electronics Specialists Conference (PESC 2008), Rhodes,
Greece, 15–19 June 2008; pp. 378–384.
36. Veberič, D. Lambert W function for applications in physics. Comput. Phys. Commun. 2012, 183, 2622–2628.
[CrossRef]
37. Kuo, B.; Golnaraghi, F. Automatic Control Systems, 7th ed.; Prentice-Hall: New Jersy, NJ, USA, 2002;
pp. 398–401.
38. Femia, N.; Petrone, G.; Spagnuolo, G.; Vitelli, M. Optimization of Perturb and Observe Maximum Power
Point Tracking Method. IEEE Trans. Power Electron. 2005, 20, 963–973. [CrossRef]
39. Ortiz-Valencia, P.A.; Trejos-Grisales, L.A.; Ramos-Paja, C.A. Maximum power point tracking in PV systems
based on adaptive control and sliding mode control. Rev. Fac. Ing. Univ. Antioq. 2015, 75, 67–79.

© 2015 by the authors; licensee MDPI, Basel, Switzerland. This article is an open
access article distributed under the terms and conditions of the Creative Commons by
Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).

12387