2018 International Conference on Control, Power, Communication and Computing Technologies (ICCPCCT)

Residential Solar Photovoltaic System With

Artificial Neural Network Based Controller
M.R Sindhu
Jyothy lakshmi P N
Electrical and Electronics Department
Electrical and Electronics Department Amrita School of Engineering, Coimbatore
Amrita School of Engineering, Coimbatore Amrita Vishwavidyapeetham
Amrita Vishwavidyapeetham India
India

Abstract—The application of renewable energy with power maximum power point tracking and inverter control for AC
system is shooting up day by day. In that residential solar load voltage regulation.
photovoltaic system has an important role. Two types of
residential solar photovoltaic systems are standalone system and
grid connected system.In order to intensify output power of a
solar photovoltaic arrangement, it is imperative to find the II. PROPOSED SYSTEM
maximum possible energy harvest from photovoltaic panel. Also
another issue that concerns solar photovoltaic system is inverter Figure 1 shows the block diagram of an ANN controlled
control. In this paper residential solar photovoltaic system using residential solar PV system with AC loads.
Artificial Neural Network (ANN) based controller for both
Maximum Power Point Tracking (MPPT) and inverter control
for load voltage regulation is implemented. A bidirectional DC-
DC converter is used to control battery charging and discharging
such that dc bus voltage fluctuations get reduced. Simulations are
done by using MATLAB/SIMULINK to analyze results.

Keywords—Artificial Neural Networks, Solar Photovoltaic System,

Bidirectional Converter, Maximum Power Point Tracking

I. INTRODUCTION
Due to the increase of world’s energy demand and
environment pollution, development of renewable energy is an
urgent needof the day. Among all renewable energy sources
solar energy is commonly using one [1]. Solar photovoltaic
system is mainly used for both industrial and domestic
applications.
Residential solar photovoltaic system mainly includes solar
PV array, power electronics converters and battery.Solar Figure 1 Block diagram of the residential solar PV system
photovoltaic array converts solar energy obtained from sun to
electrical energy. The efficiency of solar PV system is low DC-DC boost converters help to extract maximum power
while comparing with other renewable energy sources but it from PV array. Conventional MPP tracking methods are
would be more beneficial if it can track maximum power perturb and observe method, incremental conductance method,
continuously for any change in environmental conditions like
fractional open circuit voltage method etc.A battery is used for
solar emanation and temperature [2]. Power electronic
converters are for energy extraction from solar PV array. A constant dc bus voltage, Bidirectional DC-DC converter is for
bidirectional DC-DC converter is used to control battery battery charging and discharging control to reduce dc bus
charging and discharging and Inverter is to convert DC to AC voltage regulation. Single phase inverter with LC filter is
voltage. Two major issues associated with the application of implemented to feed AC loads. ANN based MPP tracking and
residential solar photovoltaic system is maximum power point load voltage regulator is developed for the residential solar PV
tracking and inverter control. system.
The main objective of this paper is to develop a residential A. Solar PV array
solar photovoltaic system with artificial neural network based Solar photovoltaic array converts solar energy obtained
from sun to electrical energy. Input to PV panel is irradiance

978-1-5386-0796-1/18/$31.00 ©2018 IEEE 416

 2018 International Conference on Control, Power, Communication and Computing Technologies (ICCPCCT)

and temperature. For every temperature and irradiance

maximum power point varies. The specification of PV array
used is given in Table 1.

Table 1: Solar PV panel details

PARAMETERS VALUE
Maximum output power 70W
No. of Parallel strings 2
No. of Series connected 1
modules per string Figure 2. Structure of an ANN in MPPT
Cells per module 60
Open circuit voltage (Voc ) 22.1 V Input to neural network is solar temperature and irradiance.
Short circuit current (Isc ) 3.99 A Number of hidden layers is calculated by trial and error
Voltage at maximum power 18.1 V method. Output is duty ratio to the boost converter. It is
trained by using incremental conductance algorithm. The
point (VMPP )
training points are obtained by varying the irradiance and
Current at maximum power 3.88 A temperature inputs to the PV array and collecting values of
point (IMPP ) duty ratios to boost converter to track maximum power from
solar panel for different temperature and irradiance conditions.
B. Design of boost converter Some training points are kept as test points to test the neural
network after training. Neural network is trained by using
Boost power converter is employed to transfer the power “nntool” in MATLAB m-FILE. Figure 8 shows the neural
available from solar PV to load. Since two panels are placed in network training with MATLAB toolbox. The network is
parallel and the peak power output is 140W as per data sheet. obtained through training by using the Levenberg-Marquardt
So the maximum operating power of the converter is 140 W. algorithm. The performance function of ANN is mean squared
Output voltage of the panel is 18.1 V at its MPP.Therefore the error (MSE). To train neural network 2400 data are taken.
input current of the converter i.e.Temperature range is from (20℃ to 45℃) and irradiance
P
iL = and ∆iL = 10% of average value. range is from (600 W/m2 to 1000 W/m2 ). The activation
Vs
In boost mode 𝑉𝑜 =
𝑉𝑠
(1) function of input layer and the output layer is “tansig” and
1−𝑑 purelin” respectively.
Vs ∗d
Inductor value is given by Lmin = (2)  Input : Irradiance and Temperature
f∗∆iL
Where,Vo is the load voltage of 24V. From the relation, the  Output :Duty ratio
duty ratio is, d=0.2467.  Number of samples:2400
Where, f is the switching frequency (20 kHz).  Number of hidden neurons:5
d∗Io
Output capacitance for boost converter is C= (3)  No of epochs:100
f∗∆Vo
Where,∆Vo is taken as 1%  Method : Feed forward back propagation method
Figure 3 shows the generated ANN block and layers inside the
Designed value of boost converter is presented in Table 2. block.
Table 2: Component values
Inductance 300mH
Capacitance 1000μF
Output voltage 24V
Switching frequency 20kHz

C. ANN based Maximum power tracker

Conventional methods of solar power extraction show
slow responses for sudden change in solar temperature and
irradiance conditions and sometimes they may fail to track
maximum power point also. An ANN based MPPT tracker is
used in this work. Figure 2gives the overview of the ANN.

978-1-5386-0796-1/18/$31.00 ©2018 IEEE 417

 2018 International Conference on Control, Power, Communication and Computing Technologies (ICCPCCT)

Figure 3 ANN block and layers inside block bidirectional DC-DC converter by allowing the proper battery
charging and discharging and AC loads are controlled by
using ANN based inverter. ANN controller is using for AC
load voltage regulation i.e. for any change in loads output
D. Bidirectional DC-DC converter
voltage value will be same. An LC filter is used for current
and voltage ripple attenuation. The inductor and capacitor
values of LC filter is shown in Table 4.

Table 4. LC filter component values and system

parameters
Parameter Specification
L 3.56mH
C 5000𝜇F
DC link voltage 24V
Output voltage 16V (RMS)

Figure 4 Bidirectional DC-DC converter with buck and boost

mode operation
Input to neural network is the difference between reference
Figure 4 shows the circuit diagram of bidirectional DC-DC
voltage and AC load voltage. Output of neural network is
converter with both buck and boost operation. Bidirectional
reference AC voltage to be generated by the inverter. The
DC-DC converter is used for the purpose of stepping up or
training of ANN is based on the structure is done by using PI
stepping down the voltage level along with the flow of power
controller based load voltage regulator, in which the DC link
in both directions [5]. Here bidirectional DC-DC converter
voltage is kept constant for various load conditions. ANN is
with battery is used to reduce the power mismatch between the
trained to get the constant value of output voltage.ANN is
output power from solar photovoltaic panel and load demand.
trained to track reference output voltage under variety of input
It operates in both buck and boost mode for battery charging
conditions. Figure 5 shows the ANN controller used for
and discharging. When the DC bus voltage is greater than the
generating switching pulses for single phase inverter.
reference DC bus voltage bidirectional converter acts as buck
converter, battery gets charged. When the DC bus voltage is
less than the reference DC bus voltage bidirectional converter
acts as boost mode, battery gets discharged.
Two PI based current controllersare used to develop switching
pulses for MOSFET switches of converter - one for buck
mode and the other for boost mode. The battery specification
and bidirectional DC-DC converter component values are
given in Table 3.

Table 3. Battery and DC-DC converter component

specifications
Component Specification
Figure 5ANN Based Inverter Controller
Battery Voltage 24V
 Input= deviation of reference voltage and actual
Ampere 18Ah voltage
hour  Output= modulation index
Bidirectional Inductor 800mH  Hidden layers= 5
DC-DC  No of epochs:100
converter Capacitor 1000𝜇F  Method : Feed forward back propagation method

The training of neural network means adjusting the weights

of layers to get the target values. Here the training of neural
E. ANN based single phase inverter control network is done by using a PI controller based output voltage
In this paper residential solar photovoltaic system is regulator. The training points are obtained by varying load
developed for AC loads. DC bus voltage is controlled by conditions and PI controller output are collected. Neural

978-1-5386-0796-1/18/$31.00 ©2018 IEEE 418

 2018 International Conference on Control, Power, Communication and Computing Technologies (ICCPCCT)

network is trained by using the Levenberg-Marquardt

algorithm. The performance function of ANN is mean squared
error (MSE).
III. SIMULATION RESULTS
Simulation is carried out on a residential solar PV system at
standard test condition of solar PV panel 25℃ temperature and
1000 W/m2 irradiance.
Maximum power point tracking of solar PV panel is done
by using artificial neural network based MPPT algorithm.
Figure 6 shows the performance of ANN based MPPT
algorithm. From performance of trained ANN it is clear that
mean squared error (MSE) decreasesas epochs increases. The
low mean squared error means that the desired output and
neural network’s output are close to each other.
ANN based MPPT tracking is more reliable while
comparing with all other conventional control methods. As
shown in the Figure 7 and figure 8 it requires less settling Figure 8. Simulation result of ANN based MPPT tracking
time, no steady state oscillations and more output power. at 600 W/m2 25℃.
Figure 9 shows solar PV voltage and current waveform under
standard test condition. Therefore, ANN based MPPT
algorithm gives better result while comparing with other
MPPT methods.

Figure 9. Solar PV voltage and current waveform

A bidirectional DC-DC converter is used to reduce voltage
fluctuations of dc bus voltage. DC bus voltage waveform is
Figure 6 Performance of ANN based MPPT algorithm given in Figure 10. Bidirectional converter helps to maintain
constant dc bus voltage 24V constant all time.

Figure 10 DC bus voltage waveform

ANN controller is used to maintain constant voltage at load
side under varying load conditions. ANN controller
givesswitching pulses to the inverter. AC load voltage and load
current are shown in figure 11.
Figure 7. Simulation result of ANN based MPPT tracking
under standard test condition.

978-1-5386-0796-1/18/$31.00 ©2018 IEEE 419

 2018 International Conference on Control, Power, Communication and Computing Technologies (ICCPCCT)

constant for any variations in load. Thus the proposed system

is used for AC load control for residential solar photovoltaic
system. From the result it is clear that ANN based MPPT and
ANN based inverter control gives better performance in terms
of more output power, less settling time, and well regulated
AC side voltage.
REFERENCES

[1] R. Mahalakshmi, Aswin K.A, and Kumar A, “ Design of fuzzy logic

based maximum power point tracking controller for solar array for
cloudy weather conditions”, Proceedings of Power and Engy System
Conference, 2014
[2] N. Femia, G. Petrone, G. Spagnuolo and M. Vitelli, “Perturb and
Observe MPPT Technique Robustness Improved”, Proceedings of IEEE
International Symposium on Industrial Electronics, Vol. 2, pp. 845-850,
2004.
[3] P.K Vineeth Kumar and C.A Asha,”An efficient solar power converter
with high MPP tracking accuracy for rural electrification”, Proceedings
of International Conference on Computation of Power, Energy,
Figure 11. AC side Load voltage and current waveform using Information and Communication, ICCPEIC 2014.
ANN controller [4] Bendib, B.; Krim, F.; Belmili, H.; Almi, M.F.; Bolouma, S., "An
CONCLUSION intelligent MPPT approach based on neural-network voltage estimator
and fuzzy controller, applied to a stand-alone PV system," Industrial
In this paper, a single phase residential solar photovoltaic Electronics (ISIE), Proceedings of IEEE 23rd International Symposium
on , vol.no., pp.404,409, 1-4 June 2014.
system using ANN based controllers for both MPPT and
[5] Jaya Aniruddha Mane, “Design, Modelling and Control of Bidirectional
single phase inverter control with bidirectional DC-DC DC-DC Converter”, Proceedings of International Conference on
converter is developed. ANN is trained by using “nntool” in Emerging Research in Electronics, Computer Science and Technology –
MATLAB/SIMULINK model. ANN based MPPT algorithm 2015.
gives better performance to get maximum possible output [6] Yang Sun and Shuhui Li “Artificial neural network for control and grid
power from solar PV array along with less settling time. integration of residential solar PV system.”IEEE transactions on
sustainable energy, vol. 8, no. 4, october 2017.
Bidirectional controlled battery is used for reduce fluctuation
in dc link voltage by allowing proper charging and discharging
of battery. ANN is used to maintain the output voltage

978-1-5386-0796-1/18/$31.00 ©2018 IEEE 420