ICCCCT-10

Solar Powered Single Stage Boost Inverter with

ANN Based MPPT Algorithm
M.Kaliamoorthy R.M.Sekar I. Gerald Christopher Raj
Assistant Professor/EEE Assistant Professor/EEE Lecturer/EEE
kalias_ifet@yahoo.com ssvedha08@gmail.com gerald.gera@gmail.com
PSNA College of Engineering and Technology, Dindigul, Tamil Nadu, India

Abstract-This paper presents the theoretical analysis, design voltage depending on the instantaneous duty cycle. In this
and simulation of a single phase single stage boost dc-ac paper a solar powered boost inverter is proposed. The closed
converter powered from PV array. The main attribute of the loop regulation of the boost inverter is necessary to achieve
boost inverter topology is the fact that it generates an ac output good dynamic response under varying plant parameter
voltage larger than the dc input one, depending on the
conditions [12]-[13]. Proportional and integral (PI) and
instantaneous duty cycle. This paper proposes an accurate
solar panel simulation to incorporate the temperature and sliding mode controller (SMC) are theoretically analyzed.
irradiation dependence of panel voltage and current along with II. PV PANEL MODELING
neural network based MPPT algorithm. This paper also The possibility of predicting a photovoltaic plant’s behavior
analyses the performance of proportional and integral (PI) and in variance irradiance and temperature is very important for
sliding mode controllers (SMC) for the closed loop controlling sizing the PV plant and converter. There are numerous
of PV fed boost inverter. methods presented in the literature, for extracting the panel
Index Terms—Boost inverter, PV array, PI controller, sliding parameters. In this paper a photovoltaic panel model based
mode control (SMC) Maximum Power Point Tracking on the manufacturer’s data sheet is presented. The
(MPPT), Neural Networks.
equivalent circuit of the single-diode model for the PV cells
is shown in fig 1 [11]. Series resistance (RS) and shunt
I. INTRODUCTION resistance (RSH) are parasitic resistances. In this model the
Recently, energy generated from clean, efficient and effect of RSH is neglected to simplify the model. The output
environmentally-friendly sources has become one of the voltage of the PV cell is given by:
major challenges for engineers and scientist. There are
various non-conventional energy resources. Among them, AkTc  I ph + I 0 − I c 
Vc = ln   − Rs I c (1)
the photovoltaic (PV) generation system has received great e  I0 
attention in research because it appears to be one of the
possible solutions to the environmental problems such as
global warming, air pollution, acid precipitation, ozone
RS Ic
depletion, forest destruction and radioactive substances ID
emission [1]-[4].In general the solar panel voltage and
current is affected by the variations in atmospheric
condition such as irradiation and temperature. An accurate
D
RSH Vc
panel modeling is needed to incorporate the parameter
variations in simulation. The panel modeling based on the
data sheet values is used in this paper [5]-[6].Voltage
Source Inverter (VSI) is probably the most important power
converter topology. It is used in many distinct industrial and Fig 1 Equivalent Circuit of PV Cell
commercial applications. Among these applications, Active compensation of series resistance, cell output voltage
uninterruptible power supply (UPS) and ac motors drives and photo current are taken into account in the modeling of
are the most important. solar PV panel. Rs, Vc, Iph are taken as a function of
The conventional VSI, normally referred as buck inverter temperature and irradiance. Thermal voltage for variable
is having the characteristics that the instantaneous average temperature Tx
output voltage is always lower than the input dc voltage. A kTnc
When an output voltage larger than the input is needed, a Vt = D (2)
boost dc-dc converter must be used between the dc source e
and and inverter. This solution can result in high volume, Diode ideality factor:
weight, cost and reduced efficiency depending on the power ( )
− I m pp − I sc * Vmpp
and voltage levels involved. A new VSI is proposed [7]- AD = (3)
[11], referred to as boost inverter, which naturally generates Vt * I m pp
an ac output voltage lower or larger than the input dc

978-1-4244-7770-8/10/$26.00 ©2010 IEEE 165

 ICCCCT-10

S1 + Load
- S3

L1 L2
Vc1 C1 C2 Vc 2

PV PANEL

S2 S4

Fig 2 Single Stage Boost Inverter

Series resistance calculation using datasheet values: other, which maximizes the voltage excursion over the load
− Impp+ Isc  [7]-[8].
Rs = −Vmpp− ln [
 Impp− Isc ]I
Vmpp
+ Voc (4) Steady state analysis
 Isc  mpp The analysis of the boost inverter under steady state is done
Open circuit voltage: by considering one converter as a voltage source (Fig. 4).
The gain of the converter vs. the duty cycle is obtained by
 I sc  [9].
Voc = ln  Vt (5)
Vo = 2Va − 2Vdc 1
 I0  (12)

Temperature dependence of voltage: Vo (

2 D − Gn 1 − D )
=
( )
(13)
VocT = Voc + K v Tx − Tc (6) Vin (1 − D )
The short circuit current and photo current were considered Where Gm is the maximum gain (Vop/ Vin), Vop is the peak
to be proportional to the value of irradiation: output voltage, Va is the capacitor voltage Vdc, is the dc
I scG = I sc S x (7) component of Va. The inductor current depends on the
demanded current and maximum gain; it is determined by
I phG = I ph S x (8) [9],
Dark saturation current: (
2 D − G m 1 − D VIN )
IL = (14)
I0 =
− I scT
(9)
(
1− D R
2
)
Voc I scT Rs
The gain characteristic of the boost inverter is shown in fig.
Vt Vt
−e +e − 0.001 5. With Gm=4. It is interesting to note that the feature of
The output voltage of the PV panel, compensated for zero output voltage is obtained for D=0.5. If the duty cycle
irradiance and temperature: is varied around this point, then there will an ac voltage at
the output terminals.
 Voc I scT Rs Voc 
 V
− I x R s + ln − I x e t − I xe
Vt
− I sc e
Vt

 
Vx =
  (10)
I sc
PV current compensated for irradiance and temperature:
I x = I scT S x (11)
III. THE BOOST INVERTER
The boost inverter achieves DC – AC conversion as follows:
the power stage consists of two current bi-directional boost
converters and the load is connected differentially across
them (Fig. 2). These converters produce a DC - biased
sinusoidal waveform (Fig. 3), so that each converter
Fig. 3. Output voltage for each DC-DC converter
produces a unipolar voltage. The modulation of each
converter is 180 degrees out of phase with respect to the

166
 ICCCCT-10

_
(
2 D max − G n 1 − D max Vin )
Load I L max = (18)
Va + Vb (1 − D max
2
R )
u = -1 Vin ' (
2 V dc − Vin )
L Where D max = 1 − , Gm =
C V op Vin
Vc2 V dc +
u=1 2
Vc1
PV PANEL The inductance and capacitance are calculated with a 20%
ad 1.5% of ripple respectively.
Fig. 4. Simplified circuit of boost inverter t on t on
L= Vin and C = I op (19)
0.2 I L max 0.015Vcmax
System modeling
Dmax
The boost inverter is modeled as two dc-dc boost converters, Where I op = Peak output current and t on =
but one of them is considered as an ideal sinusoidal voltage f max
source plus a dc component. The system model is given by
IV. TRAINING OF NEURAL NETWORK
[9].
The training of a neural network consists of solar irradiance
 − Wo   Wo 
 X.   0   X1   0   X 1  b  and cell temperature as the input patterns. The target pattern
 . 1 =  2 +
 X   − W
2
  X  u +  c  (15)
 2  o is given by measured Imp for training the neural network. The
 X 2   Wo − W1  0  2 
2   2  Imp is calculated for different values of irradiance and cell
temperature with respect to above modeled PV module. This
Where X1 = I L L , X 2 = Vc C , Wo = 1 LC , W1 = 1 RC , b = Vin L calculated Imp values are given as a training data to the neural

C = Vb R , u = { 1
−1
network. Fig (6) shows the structure of neural network used
for training and Fig (7) shows the convergence of error
during training process. During the training process, the
15 convergence error is taken as 0.01.
Input Hidden
10

G
5 Output

0 Tc O(k ) = Vmp

-5
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Oi (k )
Duty Cycle D
Bias
Fig. 5. Gain vs. Duty cycle for the boost inverter (Gm=4) j i
Design of boost inverter Wij
The inductor and capacitor values are calculated based on
the inductor current and capacitor voltage ripple. The
switching frequency should be greater than the output Fig 6 Neural Network Structure
voltage frequency. The DC component of the capacitor
voltage Vdc must be calculated as: The below graphs represent the Imp of the neural network
Vop and the actual value of the PV model. The both graphs are
Vdc ≥ + Vin (16) compared as shown in fig (10) we can conclude that the
2 neural network based MPPT gives the best results as close
as the actual values of the PV model. Once the value of Imp
The maximum capacitor voltage and inductor current is is calculated the value of maximum voltage point
determined by: corresponding to the value of Imp can be found out from look
Vop up tables.
Vc max = Vdc + (17)
2

167
 ICCCCT-10

voltage. The other state variable is inductor current. The

level of this current depends on the load. To avoid the
dependence of the current with load, the measured current is
filtered using a high pass band filter to obtain the ripple at
switching frequency that simulates the error. In the boost
inverter, since the current is varying in nature. The cut off
frequency of the high pass band filter must be chosen
carefully. It must be high enough to eliminate the 50Hz
frequency of the current, but not too high that distort the
current ripple that simulate the error. (At least 25% higher
than 50Hz). Sliding mode controller scheme is shown in fig.
11. The controller parameters are calculated based on the
boundary condition.
S 2 X1 p
S1 = (23)
X2p

Where X1p = I L max L and X 2p = Vc max C

V. PI CONTROLLER SIMULATION PV panel
The proportional and integral (PI) controller is very simple The inputs given to the models are current ramp Ic (0 to Isc
and easy to implement. It improves the system gain and to simulate the varying current of the panel), TX the variable
eliminates the steady state error. The output signal from the ambient temperature and SX the variable irradiance. The
PI controller is: results can be viewed in Figure 12 (a and b) and 13. It may
()
U t = K p e ( t ) + K p Ti ∫ e ( t ) dt (20)
be observed, that with rising temperature, the output voltage
of the panel, and implicitly the output power, decreases. The
The value of proportional gain (Kp) improves the system value of the current is not greatly affected by changes of
gain and thus the steady state tracking accuracy, disturbance temperature. Irradiance, lower than STC irradiance causes
signal rejection and the relative stability. The proportional the output voltage, current and power to fall, shown in
and integral gain (Ki) values are chosen based on the fig.12.
maximum overshoot and system stability. _
+ V0
VI. SLIDING MODE CONTROLLER
For the purpose of optimizing the boost inverter dynamics,
+
while ensuring correct operation in any working condition, a L1 +
sliding mode controller is a more feasible approach. The V2
_
main advantage over the classical control schemes is its V1 _
insusceptibility to plant parameter variations that leads to +
invariant dynamics and steady-state response in the ideal Vin
case. But compared PI controller the control theory involved _
is rather complex. The sliding surface is a linear
combination of the state variables. SMC forces the system to
be held in the surface and system is driven to the + I L1 + V1
I ref _ Vref
equilibrium point.
∑ K ∑ K ∑
σ = SX − SX r = S e X (21) _ Σ1 Σ2

Where S = [S1 S 2 ] ,
X is the state variable,
reference variables and eX = [eX 1 eX 2 ] .The condition for
T H
the existence of sliding mode is
.
σ σ < 0 and S1 X 2 − S 2 X 1 > 0 (22)
Since X2 is always positive, X1 must be positive.
The sliding mode uses two state variables to control the
boost inverter, capacitor voltage and the inductor current.
For capacitor voltage the reference is a sinusoidal voltage
plus a dc component. This reference is independent of
S (IL1, V1)
X r is the
S2
S1
Fig (11) Block Diagram of Sliding Mode Controller.
PV fed open loop boost inverter
The simulation of the boost inverter is done with circuit
parameters with Input voltage Vin= 42 V, Output voltage
Vo=100 Sin(ωt), Output frequency f0= 50Hz, Switching
frequency fs=30KHz, Load resistance R= 50Ω Inductance
L1 and L2=500µH, Capacitance C1 and C2=20µFThe

168
 ICCCCT-10

simulation results of a PV fed single stage boost inverter are considered to be varying due to the variation in the
shown in figure 14. The panel voltage is considered to be irradiance
varying due to the variation in the irradiance. CONCLUSION
250 4 A solar powered boost inverter has been proposed and
25 Deg 3.5 simulated. The accurate solar panel has been designed to
200 50
75 3 change the panel voltage and current with the changing
100
150 atmospheric condition. Due to the parameter variations

Current in Amps
2.5
Power in Watts

100 2
25 Deg
50
closed loop controlling has been employed to improve the
1.5
75
100
system performance. A simple PI controller has been
50
1
employed and found from the simulation that system is not
0
0.5
completely robust. To improve the dynamic stability and
-50 0
robustness sliding mode controller has been proposed and
0 10 20 30 40 50
Voltage in Volts
60 70 80 0 20 40 60
Voltage in Volts
80 100
simulated. It has been observed that the system is robust and
Fig 12 PV and VI Characteristics with Constant Irradiance insensitive to parameter variation when it is controlled by
of 1400 W/Sq.M sliding mode controller. A PV fed boost inverter is useful
for applications in which the instantaneous output ac voltage
400

350
6 should be larger than the input dc voltage. The total
1400 W/Sq.M

300
1600 W/Sq.M 5
harmonic distortion of the open loop boost inverter is below
1800 W/Sq.M
1900 W/Sq.M 5. And also the proposed inverter is very environmental
Current in Amps
Power in Watts

250 4

200
friendly as it is powered by solar panel. Reference Points for
150
3
1400 W/Sq.M
the controllers are fed from neural networks.
2 1600 W/Sq.M
100 1800 W/Sq.M 78 350
1900 W/Sq.M
1 76
50

Capacitor Voltage V c1 in Volts

300
74
Panel Voltage in Volts

0 0
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 72 250
Voltage in Volts Voltage in Volts
70

Fig 13 PV and VI Characteristics with Constant Cell 68

200

Temperature of 250C 66
150
64
78 200
62 100
1000W/sq.M 0 0.05 0.1 0.15 0.2 0.25 0.3 0.05 0.1 0.15 0.2 0.25 0.3
76 900W/sq.M 150
Boost Converter ouput Voltage in Volts

Time in Secs Time in secs

700W/sq.M 200 350
74 100
Panel Voltage in Volts

500W/sq.M 150

Capacitor Voltage Vc2 in Volts

72 50 300
Output Voltage in Volts

100
70 0
50 250
68 -50
0
66 -100 -50 200

64 -150 -100
50W/sq.M 50W/sq.M
150
62 -200 -150
0 0.05 0.1 0.15 0.2 0.25 0.3 0 0.05 0.1 0.15 0.2 0.25 0.3
Time in secs Time in secs -200 100
0.05 0.1 0.15 0.2 0.25 0.3 0.05 0.1 0.15 0.2 0.25 0.3
Fig.14. Simulation results of open loop PV fed boost Time in secs Time in secs

inverter Fig.16. Simulation results for sliding mode controlled

PV fed boost inverter with PI controller PV fed boost inverter
The simulation results of a PV fed single stage boost
inverter are shown in figure 15. The panel voltage is REFERENCES
considered to be varying due to the variation in the [1] S. R. Bull, “Renewable energy today and tomorrow,”
irradiance. From figure 15 it is clear that the dynamic Proc. IEEE, vol. 89, no. 8, pp. 216–1226, Aug. 2001.
response of the PI Controlled PV fed boost inverter is poor. [2] S. Rahman, “Green power: What is it and where can we
. find it?” IEEE Power Energy Mag., vol. 1, no. 1, pp. 30–37,
Jan./Feb. 2003.
[3] J. A. Gow and C. D. Manning, “Photovoltaic converter
system suitable for use in small scale stand-alone or grid
connected applications,” Proc. Inst. Electr. Eng.—Electr.
Power Appl., vol. 147, no. 6, pp. 535–543, Nov. 2000.
[4] T. J. Liang, Y. C. Kuo, and J. F. Chen, “Single-stage
photovoltaic energy conversion system,” Proc. Inst. Electr.
Fig.15. Simulation results of PI Controlled PV fed boost Eng.—Electr. Power Appl., vol. 148, no. 4, pp. 339–344,
inverter Jul. 2001.
PV fed boost inverter with SMC controller [5] Krisztina Leban, Ewen Ritchie, “Selecting the accurate
The simulation results of a PV fed single stage boost solar panel simulation model,” NORPIE/200, Nordic
inverter are shown in figure 16. The panel voltage is

169
 ICCCCT-10

Workshop on Power and Industrial Electronics., pp.1-7,

June 9-11 2008.
[6] Dezso Sera, Remus Teodorescu, Pedro Rodriguez, “PV
panel model based on datasheet values,” IEEE Transactions.
Industrial Electronics., pp 2392-2396, June 2007.
[7] R. O. Caceres, I. Barbi, “ A boost DC-AC converter:
Operation, analysis, control and experimentation”, In Proc.
Int. Conf. Industrial Electronics Conf. (IECON’95)” , pp
546-551, Nov 1995.
[8] R. O. Caceres, I. Barbi, “A boost dc-ac converter:
Analysis, design, and experimentation,” IEEE Trans. Power
Electron., vol. 14, no. 1, pp. 134–141, Jan. 1999.
[9] J.Fernando, Sonia S. Paulo, “Fixed frequency Sliding
Mode Modulator for Current Mode PWM inverters.”, IEEE
Power Electronics Specialists Conference-PSEC’93, pp.
623-629
[10] F. Barzegar, S. Cuk, “Solid-state drives for induction
motors: Early technology to current research,” in Proc.
IEEE Region 6 Conf., Anaheim, CA, Feb. 15–18, 1982.
[11]. M.Kaliamoorthy ,Nagaraju Reddy,D.V.Ashok Kumar,
“Control of grid connected PV cell distributed generation
system”, in Proc. IEEE TENCON Conf Hyderabad.

M.Kaliamoorthy received the M.Tech degree in

electrical drives and control from Pondicherry
University, India, in 2006. He is currently working
as an assistant professor in PSNA College of
Engineering and technology, Dindigul, Tamilnadu,
India in the Department of Electrical and
Electronics Engineering. His research interests
include alternative energy sources, fuel cells,
energy conversion, power system modeling,
analysis and control, power quality and active
harmonic analysis.

R.M.Sekar received the M.Tech degree in

Power Electronics from VIT University, India, in
2004. He is currently working as an assistant
professor in PSNA College of Engineering and
technology, Dindigul, Tamilnadu, India in the
Department of Electrical and Electronics
Engineering. His research interests include
multilevel inverters, alternative energy sources,
energy conversion, power quality, active
harmonic analysis and Facts devices analysis and
control.

I.Gerald Christopher Raj received the M.E

degree in Power Electronics and drives from
Anna University, India, in 2006. He is currently
working as an Lecturer in PSNA College of
Engineering and technology, Dindigul,
Tamilnadu, India in the Department of
Electrical and Electronics Engineering. His
research interests include CSI fed AC motor
drives, energy conversion, power system
modeling, analysis and control and power
quality

170