Abstract— This paper presents a sliding mode control connected PV system. It utilizes various control strategies:
method of a three-phase three-level neutral point clamped power controlled, [7], and current controlled grid connected
(NPC) inverter based grid connected photovoltaic system. The PV system, [8].
three-level NPC inverter is forced to deliver just active power
to the grid and regulates its self-sustaining DC bus voltages
using three sliding mode controllers. The three-level space Generally, multilevel topologies have been studied to
vector modulation is used in order to turn-on and turn-off the reduce the voltage stress on the power-switching devices and
inverter switches and to ensure also the balance of the DC to generate less harmonic components on both output
capacitor voltages. The three-level NPC inverter is connected to voltage and load current [9]. Varied multilevel inverter
the photovoltaic panel through a DC/DC converter. For the configurations have been introduced and studied in the
optimum use of the PV module, a maximum power point literature. The most known of these types are the neutral-
tracking (MPPT) controller is used which enables the
maximum power extraction under varying irradiation and clamped introduced by [10], flying capacitors [11], and
temperature conditions. Finally, the effectiveness of the Cascaded H-Bridge multilevel inverter [12].
proposed control system is demonstrated with Matlab/Simulink
simulations. Likewise, various modulation and control technics have
been introduced for multilevel-inverters, the most known
Keywords: Photovoltaic system, Space vector modulation, are: selective harmonic elimination [13], sinusoidal pulse
MPPT, NPC inverters, Sliding mode control. width modulation (SPWM) and space vector modulation
(SVM) [14]. However, there is as yet an extraordinary
I. INTRODUCTION enthusiasm on new modulation strategies to enhance
multilevel inverter performance.
Growing energy consumption, the finite nature of fossil
fuels and the worsening of the global environment generated
In this paper, the grid-connected Photovoltaic system
a growing interest in renewable energy production systems
distributes only active power generated by the Photovoltaic
[1]. Among these systems, photovoltaic (PV) energy seem panels to the grid. A control system based on a sliding mode
particularly promising. It is converted to the electrical power controller is proposed to enhance the system performance.
by photovoltaic cells that don't create any dangerous or toxic The proposed approach allows obtaining a fast and robust
substances that contaminate the nature [2]. Compared to response of the system. Simulation results are provided to
regular electrical power plants, these systems are validate the proposed photovoltaic systems and
inexhaustible, littler and can be installed near to load centers. corresponding controllers.
Photovoltaic systems are sometimes worked in stand-alone II. SYSTEM CONFIGURATION AND MODELING
mode and they feed fixed loads by stand-alone Photovoltaic
inverters [2], [3]. In addition to stand-alone applications, as A. System Configuration
of late, grid-connected photovoltaic systems have pulled in The structure of the grid-connected power system and its
significant consideration [4], [5]. control are shown in Figure 1. The whole system is
composed of a PV arrays, a DC/DC boost converter, a three-
Grid-connected photovoltaic systems extract maximum level inverter and sliding mode based controllers. The
photovoltaic panels catch the sunlight and produce DC
power from the photovoltaic panels by the MPPT technic
power. A power transformer is used for connecting the NPC
that is mostly related with a DC–DC converter to increase inverter with the AC grid. This permits adjusting the inverter
the output efficiency of PV arrays [6]. Thus, the inverter is voltage to the grid voltage and guaranteeing electrical
the key-component for successful operation of the grid protection between the PV system and the grid.
B. Photovoltaic Cell Modeling
A. Bouzidi is with Laboratoire de Génie Electrique, Université de The PV cell is described by its current–voltage
M’sila, Algérie (e-mail: bouzidi.ali@univ-msila.dz).
M. L. Bendaas is with Département d’Electrotechnique, Université de
characteristic function, as
Batna2, Algérie (e-mail: mlbendaas@yahoo.fr).
S. Barkat is with Laboratoire de Génie Electrique, Université de M’sila, V R s I s
Algérie (e-mail: sa_barkat@yahoo.fr). I s I Ph I 0 exp s 1 (1)
M. Bouzidi is with Département de Electronique et communication, nkT q
Université de Ouargla, Algérie (e-mail: bouzidi.m.28@gmail.com).
where Is is the output current of the solar cell (A), Vs is dimensionless deviation factor from the ideal p-n junction
the output voltage of the solar cell (V), IPh is the photonic diode, k=1.3807.10-23 J / K4 is Boltzmann’s constant, T is the
produced current (A), I0 is the cell reverse saturation current cell temperature (K), Rs is the series resistance (Ω).
(A), q= 1.6022.10-19 C is the electronic charge, n is the
vs_abc Rs Ls is_abc
vdc*
vp_abc +
Rl Ll il_abc PLL DC voltage
SMC
-
θ i *fd
vpq vpd
ifd +
if_abc Abc - Currents
SMC
Lf dq
ifq - +
i *fq
Rf 0
vf_abc
id 3-Level Sabc
ipv
Boost iceq v*f v*fd
C1 SVM with
Chopper αβ
67 vpv + balancing v *f dq v*fq
MPPT C2 strategy
Inverter
vc1,vc2
Figure 1. Sliding mode control of grid-connected photovoltaic system
III. THREE LEVEL INVERTER should be On (state 1). Table 1 briefs about the switching
states of three-level NPC inverter.
Three-level NPC inverters are an excellent compromise
solution between performance and cost in multilevel
converters for high and medium power systems. The circuit
diagram shown in figure 2 gives idea about the arrangement
of the power circuit required for the three level NPC inverter.
Two capacitors are used to split the DC voltage and provide a
neutral point o. The inverter leg a is composed of four active
switches Sa1, Sa2, Sa3 and Sa4 with antiparallel diodes and two
clamped diodes Da1 to Da2.
TABLE I. SWITCHING STATE OF THREE-LEVEL INVERTER
For one leg operation of phase-a, to have an output voltage As shown in Figure.3, all the 27 switching vectors can be
of vdc/2 the switches Sa1, Sa2 should be On (state 2) and to sorted into two centred hexagons. The diagram of space
have – vdc/2 voltage, the switches Sa3. Sa4 should be On (state vectors can be divided into six sectors with every sector
0) and to have output voltage as zero the switches Sa2, Sa3 further divided into four triangles.
now to model the system dynamics in the abc frame as
020 120 220
follows:
v8 v6 v5
v fa i fa i fa v pa
v R i L d i v (3)
fb fb pb
dt
f f fb
i fc v
121 221
v fc i fc
021 210
v9 010 v7 v4 110 v3
pc
v * v*
For the DC-side circuit, and based on the power balance
v 11
122
v 10 222 v1 211
v2 200 equation between the output and input of the inverter, it y
022 111
011 100 v *
000 v 0
1
i ceq i d v fa i fa v fb i fb v fc i fc (4)
v dc
v 12 v 13 v 16 v 18
012
112
001
212
101
201 where id is the boost output current.
The system mode given by (3) can be transformed in the
v 14 v 15 v 17 synchronous frame dq as fellow:
002 102 202
di fd
Figure 3. Space vector modulation three level NPC inverter
1
dt L R f i fd i fq v fd v pd
f
(5)
di fq 1 R i i v v
SVM is a discrete category of inverter modulation
technics in which the reference of a voltage vector v* dt Lf
f fq fd fq pq
(computed by sliding mode currents controllers) is studied by
and for the DC bus, the equation (4) becomes
the time average of an appropriate switching state vectors
[15]. When the vector v* is located in witch sector at any dv dc 1
sampling instant, the edge of the voltage vector lies in a i ceq (6)
dt C eq
triangle formed by the three switching vectors adjacent to it,
(see Figure. 3). The nearby vectors necessary to combine the
reference voltage vector are connected by: The equations (5) and (6) represent the model that can be
used to synthesize the system control.
T hv * t pv p t qv q t uv u
(2) V. SLIDING MODE CONTROL
t p t q tu T s
A. Background on Sliding Mode Control
where Th is the switching period, v p , v q and v u are the
Sliding mode control is able to constrain the system stats
switching vectors adjacent to the reference voltage vector, to follow trajectories which lie on a suitable sliding surface.
and t p , t q and t u are their on-duration time intervals The main steps for sliding mode controller design can be
respectively. summarized as follows [16]:
The algorithm of multi-level SVM is implemented in the - Selecting a switching surface S(x) (where x is the system’s
Three steps: Reference voltage vector location in witch sector state vector;
and witch triangle; Calculation of the switching times; - Verification of existence conditions to ensure that
Generating control pulses.
Lyapunov stability criterion S S 0 is fulfilled;
In the NPC inverter, the voltages of the two arrangement
associated DC-connect capacitors must be bound to the half - Obtaining the equivalent control ueq by applying the
of dc-voltage to take advantage of the inverter. The DC- invariance condition S (x ) 0 and S (x ) 0 .
voltage sliding mode controller regulates only the total DC
voltage. In that event, the DC-link voltages are maintaining - Finally, the main control is obtained by:
balanced using three-level SVM that uses advantages of
redundant switching states to eliminate the dc voltages drift u (t ) ueq (t ) u n (t ) (7)
phenomenon [15].
where un is the nonlinear control that take the function
IV. GRID CONNECTED SYSTEM MODELING smooth as follows:
The idea is to represent the inverter by three voltages
s x
sources connected to the PCC by a line filter (Rf, Lf). The DC Smooth s (x ) (8)
bus is represented by a current source connected to s x
equivalent condenser Ceq, where Ceq=C/2 [15]. It is possible
Where s x is the absolute value of s (x ) .
adding an integral action in its expression. This action can u K p B Ki Edc K p ( AI G I ) K s smooth(Sdq ) (15)
*
remove the static error; the new surface is takes the
following form [17]: D. DC Voltage Controller
The equation (6) is used to control the DC voltage vdc
r 1 where the current iceq is the control variable. The sliding
d
s (e y ) e y (t ) k i e y (t ) dt (9) surface chosen for this control is
dt
where ey is the error, λ is a positive constant, r is a relative
*
S dc k pc v dc
v dc k ic v dc
*
v dc dt (16)
degree and ki is an integral gain.
During the sliding phase, the following equation is satisfied
C. Currents Controllers
The equation (5) can be written as:
*
S dc k pc v dc *
-v dc k ic v dc
-v dc 0 (17)
I AI Bu G (10) The equivalent control can be given by
T T
with: I i fd i fq : state vector; u v fd vfq : control
k
T
vector; G 1 L f v fd vfq : perturbation vector; k pc
*
i ceq eq C eq ic v dc *
v dc v dc (18)
The nonlinear control can be defined as
R L 1
A f f ;B
Rf Lf Lf i ceq n k sc smooth (S dc ) (19)
The two sliding surfaces for the direct and quadrature
Finally, the capacitor voltage control low is given by
current components and that for the component are defined as
follow
k
S dq K p E dq K i E dq dt (11) k pc
*
i ceq C eq ic v dc *
v dc v dc k sc smooth (S dc ) (20)
T Using the power conservation principle, the d-axis
with: S dq S d Sq : sliding surface vector; reference current is calculated as
T
E dq i d* i d i q* i q : Error vector; *
Ppv v dc i ceq
*
i fd (21)
k pd 0 k id 0 v pd
Kp ; K i : matrix of positive
0 k pq 0 k iq
gains. VI. SIMULATION RESULTS AND DISCUSSION
The equivalent control low can be found by solving The proposed scheme is implemented using the
Sdq (x)=0. By differentiating Sdq(x) = 0, it yields parameters given in Table 2. The PV array in the proposed
scheme consists of solar PV array of 201panels (three
S dq K p I * K i E dq K p (AI Bu éq G ) 0 (12) cascaded arrangements connected in parallel each one consist
of 67 panels in series).
The equivalent control can be given by
TABLE II. SIMULATED SYSTEM PARAMETRS
uéqd
1
uéq K pB *
K E K p (AI G I ) (13)
i dc
uéqq Parameter Value
vseff 5.5 kV
If the system did not reach the sliding surface, the equivalent Ls, Rs 5 µH, 6 mΩ
control must be enforced by the nonlinear control define as Lf, Rf 30 mH, 0.3 mΩ
follows vdc, C 5 kV, 6.5 mF
fs, fh 50 Hz, 5 kHz
u nd kpd, kpq 12
un K s smooth (S dq ) (14) kid, kiq 100
u nq ksd, ksq 30
k sd 0 At the beginning of simulation, the system operates in the
Ks : matrix of positive gains.
k sq
standard conditions of temperature and radiation and feeds a
0 load of 30kW. At t = 0.25s, the solar radiation is decrease
Figure 8. Injected current and reduce PCC voltage.
Figures 6 and 7 show that the inverter injects only the
active power in the main grid and the reactive component of
the inverter current is always null. Consequently, it ensures
unit power-factor operation at the inverter side. That is more
Figure 4. DC voltage waveforms and its reference.
clearly shown in figure 8, where the waveform of the injected
current to the grid is sinusoidal and in phase with its
corresponding grid voltage.
Figure 5. DC capacitor voltages waveforms.
Figures 4 and 5 show that the DC voltage SMC maintains Figure 9. Powers consumed by the load, delivered by the PV and
a constant DC bus voltage across the capacitors with very by grid.
short response time with low overshot. Thanks to the Figure 9 shows the corresponding changes in active
balancing algorithm associated to the three-level SVM, the power, both of grid Ps and the photovoltaic system PPV to
DC-capacitor voltages are maintained balanced. meet the demand in active power of load Pl. At the beginning
the PV feeds the load alone and there is no need to the grid.
The decrease in radiation or the increase of temperature
results in a decrease in the active power of PV system as
shown in Figure 9. At the same time there is a rise of active
power provided by the grid to the load.
VII. CONCLUSION