International Review of Electrical Engineering (I.R.E.E.), Vol. 6, N.

5
September-October 2011

NSGA-II Optimized Neural Network Controlled

Active Power Line Conditioner Under Non-Sinusoidal Conditions

D. Kavitha1, A. F. Zobaa2, P. Renuga1, V. Suresh Kumar1

Abstract – Optimization of total harmonic distortion (THD) and power factor subjected to power
quality constraints for the evaluation of proposed active power line conditioner (APLC) is carried
out in this paper. Non-dominated sorting genetic algorithm-II is used to obtain the reference
source current to optimize both power factor and THD. The proposed APLC is evaluated using
conventional fixed frequency variable slope (FFVS) technique and neural network, which is
trained from the samples obtained using FFVS. Computer simulations of the proposed APLC have
been performed using MATLAB and the results are encouraging. The results show that the
proposed APLC can reduce the total harmonic distortion of a specific nonlinear load from 25% to
about 2%, and improve the power factor close to unity under non-sinusoidal conditions. Copyright
© 2011 Praise Worthy Prize S.r.l. - All rights reserved.

Keywords: Power Systems Harmonics, Active Filters, Neural Networks

Nomenclature The use of power electronic circuits in a wide range of

applications has resulted in distorted current waveforms
Vs(t) Instantaneous voltage at point of common in the power system. This results in non-sinusoidal
coupling (PCC) voltage drops across the transformers and transmission
Vpi Peak value of voltage corresponding to ith line impedances, resulting in a non-sinusoidal voltage
harmonic at PCC supply at the point of common coupling. The
Ipi Peak value of desired source current requirements of harmonic-free current waveforms and
corresponding to ith harmonic good power factor, under non-sinusoidal voltage
S Apparent power conditions, contradict each other [6]-[8]. Under these
gi Conductance corresponding to ith harmonic conditions, an optimum performance is the best one can
P Active power achieve. This paper proposes a new scheme as a
SHLi Selective harmonic limit of ith harmonic compromise between the power factor and current
t time instant of sampling distortion under non-sinusoidal voltage conditions.
is*(t) Desired source current at tth instant Classically, a passive filter consists of tuned LC filters
icom*(t) Desired compensation current at tth instant and/or high pass filters are used to suppress harmonics,
IL(t) Instantaneous load current value at tth instant and power capacitors are employed to improve the power
Icom(t) Actual compensation current from APLC at tth factor of the utility/mains. But these conventional
instant methods have the limitations of fixed compensation, a
n Total number of harmonics large size and can also excite resonance conditions [10]-
T Sampling time [11]. Hence shunt active power filters are introduced as a
k Sample number viable alternative to compensate harmonics and improve
VAB(tk) Output voltage of APLC at tkth instant power factor [12]-[15]. Two fundamental approaches to
Lcom Compensation Inductor improving power quality with active power filters (APFs)
are correction in the time domain and correction in the
frequency domain. In early APF designs, PWM-based
I. Introduction methods, such as constant frequency control, sliding
The adverse effects of harmonics have been well- mode control, hysteresis control, and triangular
established in the literature for the past 3 decades [1]-[4]. waveform control, are used to control the switches in the
Hence harmonics have to be maintained within the limits APF using a time domain approach [16]-[19].
[5]. The purpose of an active power line conditioner is to The main shortcoming of this method is that, in order
compensate the utility line current waveform, so that it to obtain optimum results, relatively high switching
approximates a sine wave in phase with the line voltage frequencies are needed, which subsequently leads to high
when a nonlinear load is connected to the system. The switching losses. Frequency domain methods include
APLC not only works as a harmonic compensator, but predetermined harmonic injection, and PWM-based
also as a power factor corrector. techniques, such as optimized injection method [20] and

Manuscript received and revised September 2011, accepted October 2011 Copyright © 2011 Praise Worthy Prize S.r.l. - All rights reserved

2604
 D. Kavitha, A. F. Zobaa, P. Renuga, V. Suresh Kumar 

adaptive frequency control [21], are proposed as

alternatives to a time domain approach. The switching
frequencies for frequency domain methods can be much
lower than time domain schemes, resulting in much
lower switching losses. The main disadvantage of the
frequency domain method is the longer computational
time than time domain methods. Nowadays, high speed
processors are available to reduce computational time in
practice. Whether a time domain or frequency domain
approach, the conventional active power filters are too
complex and costly in practice, when the quantity to be
controlled varies over a wide range [22].
The shortcoming of these methods is that, in order to Fig. 1. Voltage type APLC
obtain optimum results, relatively high switching
frequencies are needed, which subsequently leads to A variable DC voltage source provides eight levels of
excessive switching losses. Neural network techniques DC voltage VD, and a DC/AC converter applies a positive
were successfully applied to frequency domain APLC or negative polarity to the variable DC voltage. Thus,
control [8]-[11]. voltage VAB has fifteen possible combinations of the four
Even in the artificial intelligence based APLC [26]- on-off control signals, sw, d1, d2, d3, as shown in Table I.
[28], a single objective function is considered to mitigate These four control signals control the ten switches in the
harmonics. The optimization of power factor while APLC. Thus if sw = 1 then (S1,S4) is ON & (S2,S3) is
reducing the harmonics is not included. Hence an attempt OFF. Similarly:
is made to compensate for both power factor and
harmonics. In this work, about 400 different nonlinear d1 = 1 Æ Sd1 = ON & Sd2 = OFF
conditions of a load are considered to train the neural d2 = 1 Æ Sd3 = ON & Sd4 = OFF
network, whose input is load current samples, and whose d3 = 1 Æ Sd5 = ON & Sd6 = OFF
output is the switching signals of APLC. For each
nonlinear condition, the load current harmonics and TABLE I
COMBINATIONS FOR ON-OFF CONTROL SIGNALS
voltage at the point of common coupling is estimated
using an adaptive neural network [29], and non-
dominated sorting genetic algorithm-II (NSGA-II) [30]-
[32] is used to obtain the reference compensation current.
FFVS technique is used to determine the switching
signals of APLC for the given switching frequency and Owing to the current voltage relationship at inductor
the given reference compensation current. The generated Lcom, the switches provide a positive or negative slope for
switching signals, along with the load current, are the compensation current icom. By permitting several
considered as a sample to train the neural network. levels of DC output voltage, the VDVS allows for more
NSGA-II is used to obtain the reference compensation precise tracking of an ideal or reference compensation
current so that it optimizes the power factor, and the total current by the APLC, than would a fixed DC output
harmonic distortion (THD) is subjected to power quality voltage. The value of this inductor is crucial in the
constraints. performance of the APLCs. If a small value of inductor is
The time taken by the FFVS method is considerably selected, then large switching ripples are injected into the
large, and is not suitable for online operations. Hence a supply currents, and a large value of inductance does not
neural network controller is proposed. The FFVS method allow proper tracking of the compensating currents close
is used to obtain the training samples for the neural to the desired values. An optimum selection of
network because of its accuracy. The trained neural inductance is essential to obtain satisfactory performance
network has a real-time advantage of high speed, and can of the APLC.
be implemented online for the particular load under The purpose of an APLC is to compensate the utility
study. source current waveform so that it approximates a sine
wave, with power factor close to unity when a nonlinear
load is connected to the system.
II. Active Power Line Conditioner
(APLC)
III. Estimation of Reference Current
II.1. System Overview
Under non-sinusoidal supply conditions, the unity
The APLC is illustrated in Figure 1 [27]. Basically, it power factor is achieved by making the source current
consists of two primary components: identical in phase angle, and of an identical shape as that
• Variable DC voltage source (VDVS) of voltage. When the source current is given the same
• DC/AC converter. shape as voltage, current THD may rise beyond the

Copyright © 2011 Praise Worthy Prize S.r.l. - All rights reserved International Review of Electrical Engineering, Vol. 6, N. 5

2605
 D. Kavitha, A. F. Zobaa, P. Renuga, V. Suresh Kumar 

acceptable limit. To obtain perfect harmonic n

∑ g iV pi 2
1
compensation, the current drawn from the source needs P= (6)
2 i =1
to be a perfect sine wave. However, in that case, the
unity power factor is not realized. Hence, there is a need
to optimize the power factor, satisfying power demand
and total and selective current harmonic limit.
NSGA-II is used to optimize the nonlinear equation
for apparent power and THD, subject to equality and
inequality constraints. Apparent power is considered as a
function to be minimized, since power factor is the ratio
of active to apparent power, and active power supplied
by a source to load is a fixed constant. A non-sinusoidal
voltage supply is considered at the point of common
coupling (PCC), owing to source impedance [6].
The first objective function is to minimize the THD in
the source current. The frequency components in the
supply current and supply voltage should be identical,
and the corresponding phase angles should be zero, to
make power factor unity. But in this case the harmonic
effect cannot be completely nullified. Hence, there is a
need for optimum reference current generation to
mitigate the harmonics, and to improve the power factor.
The desired source current should be calculated in such a
way that it should supply the constant active power
demanded by the load. As active power remains constant
the only way to improve the power factor is to minimize
the apparent power supplied from the source, and it is Fig. 2. Flowchart for NSGA-II algorithm
considered to be the second objective function.
For each ith harmonic: The other constraint of the problem is selective
harmonic elimination:
  I pi = g iV pi (1)
g iV pi
< SHLi (7)
Then: g1V p1

Objective function I: The optimization problem is solved using NSGA-II,

and the optimized values of conductance corresponding
to harmonics are obtained.
n n A brief outline of the NSGA-II flowchart is shown in
∑ I pi 2 ∑ g i 2V pi 2 Figure 2. A block diagram of NSGA-II algorithm is
i=2 i=2
Minimize THD 2 = = (2) given in Figure 3.
I p12 g12V p12 The desired or reference source current for the
fundamental frequency of 50Hz under non-sinusoidal
Objective function II: supply is given by:

n n
i * s (t) = ∑ I pi sin(100πit + φ i ) (8)
∑ ∑
1 i =1
Minimize S 2 = V pi 2 I pi 2 (3)
4 i =1 i =1
where φi is the phase angle of ith harmonic of estimated
n n voltage. The compensation current i*com(t) to be injected
∑V pi 2 ∑ g i 2V pi 2
1
S2 = (4) by APLC is:
4 i =1 i =1

subject to constraints (active power balance): i *com (t ) = I L (t ) − i *s (t) (9)

n The task of the APLC is to produce an actual

∑
1
P= V pi I pi (5) compensation current icom(t) that will closely track
2 i =1 i*com(t).

Copyright © 2011 Praise Worthy Prize S.r.l. - All rights reserved International Review of Electrical Engineering, Vol. 6, N. 5

2606
 D. Kavitha, A. F. Zobaa, P. Renuga, V. Suresh Kumar 

Fig. 3. Block diagram of NSGA-II algorithm

Fig. 4. A single hidden layer feed forward neural network

IV. Fixed Frequency with Variable
Slope Control In this problem, sampling time is selected as 0.625 ms,
and so for 50 Hz signal using rectangular window the
To obtain training samples for a neural network
number of input neurons n is 32. The switching
controller, FFVS technique is used. FFVS control
frequency is set as 1600 Hz. Hence the number of output
technique is associated with the mathematical model of
neurons m is 32. For the optimum performance of neural
the system. The estimation of the switching functions of
net, the neurons in the hidden layer are considered as 32.
APLC is obtained using the following equations:

icom ( tk +1 ) = {VAB ( tk ) − Vs ( tk )} ×
T
+ icom ( tk ) (10)
VI. Validation of Proposed Algorithm and
Lcom Creation of Samples
The test system used in this work comprises a two bus
Voltage across inductor Lcom is: single phase system with a nonlinear load. The voltage at
the point of common coupling is distorted by adding
{icom ( tk +1 ) − icom ( tk )} (11)
Lcom
VAB ( tk ) − Vs ( tk ) = third and fifth order harmonics to a fundamental
T frequency supply of 50 Hz. The load current IL is
estimated using anadaptive neural network up to 10th
For the APLC consisting of a 6-switch variable DC order, and NSGA-II is used to solve the above stated
source and a full bridge DC/AC inverter, 15 different optimization problem, and computes the reference source
combinations of VAB can be produced by 4 control current i*s(t) using (8). The genetic parameters of
signals. The reference compensation current is sampled NSGA-II are as follows: number of population = 50;
and held at a regular interval tk. Then at each sampling number of generations = 250; number of runs =15;
point, the above equation is solved fifteen times; once for simulated binary crossover with cross-over index, ζc=5,
each of the possible values of VAB. The value of VAB(tk) and mutation index, ζm=20; tournament selection and
and its associated signal set that minimize the error penalty less constraint handling method [32] is adopted.
between icom and icom* is then chosen for that interval.
-3
x 10
0.237 1.81

1.8

V. Neural Network based APLC 0.236

1.79

0.235
Objective function II
Objec tiv e func tion II

1.78

To obtain a faster response the neural network (NN) 0.234 1.77

controller is designed. The neural network controller for 1.76

0.233
the proposed APLC consists of a single hidden layer in a 1.75

feed forward structure, and controls the ON and OFF 0.232

1.74

states of the switches in the variable DC supply and the 0.231

0 0.002 0.004 0.006
Objective function I
0.008 0.01 0.012
1.73
0 0.002 0.004 0.006 0.008
Objective function I
0.01 0.012 0.014

DC/AC inverter.
Figure 4 illustrates a two-layer feed forward neural Fig. 5. Paretofront obtained from NSGA-II for 185 A and 20 A,
network with n inputs and m outputs. The hidden layer respectively
consists of p neurons. The fully connected weight
TABLE II
matrices W and V are initially non-zero. The neural RESULTS OBTAINED FROM PARETOFRONT
network is implemented in software and trained offline CASE 1: 185 A CASE 2: 20 A
using the FFVS control scheme. A back propagation THD (%) Power factor THD (%) Power factor
algorithm is used for training. The training data obtained 0 0.97 0 0.972
with the FFVS method are built up from a collection of 2 0.9836 2 0.98
5 0.9896 5 0.9897
400 typical nonlinear load current waveforms. Back 7 0.9915 7 0.992
propagation training produces optimum weight matrices 10 0.9951 10 0.9961
for the neural network to be used in online control.

Copyright © 2011 Praise Worthy Prize S.r.l. - All rights reserved International Review of Electrical Engineering, Vol. 6, N. 5

2607
 D. Kavitha, A. F. Zobaa, P. Renuga, V. Suresh Kumar 

The value of the coupling inductor plays a dominant After injecting the compensation current through the
role in the generation of actual compensation current APLC, the THD is reduced to about 3% from 26%, and
from APLC. The values selected for simulation are the power factor is improved from 0.73 to 0.9836. The
L=0.0075 H for low current load and L=1mH for high harmonic spectrum of source current before and after
current load. compensation is shown in Figure 7, for current value of
The filtering problem is solved using the FFVS 20 A.
method [28]. Table III shows the variation in the THD For online operation, the APLC monitors both the
and power factor level for non-sinusoidal supply source current and the load current. When it determines
conditions, if the reference current is evaluated with and that the source current THD is higher than the allowable
without optimization, using NSGA-II for the line current limit, the well-trained NN controller takes a number of
of 185A. From the results, it is seen that if the reference equally spaced samples from one 50Hz cycle of the load
current is obtained after optimizing the objective current, and produces control signals for the 10 switches
functions, the trade-off between THD and power factor is in the APLC.
achieved. Similarly, the results are given for 20A current The samples of load current are given as input to the
as shown in Table IV. NN controller, and the output is the AC side output
voltage of APLC. The output current of the APLC for the
TABLE III control signals are evaluated and it is compared with the
RESULTS WITH AND WITHOUT OPTIMIZATION USING NSGA-II FOR
185A LOAD
compensation current. The source current before and
Reference Current THD Power factor
after compensation is shown in Figure 6.
current (%)
Compensation

Compensation

Compensation

Compensation

40
before compensation
Before
Before

Source current (A)

After

After

after compensation
20

-20

Non optimized 12.3 1.1 0.8205 0.968 -40

Optimized 12.3 2.4 0.8205 0.981 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
time(s)

40
Compensation current (A)

reference compensation
TABLE IV
20 actual compensation
RESULTS WITH AND WITHOUT OPTIMIZATION USING NSGA-II FOR
20A LOAD 0
Reference Current THD Power factor
-20
current (%)
-40
Compensation

Compensation

Compensation

Compensation

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
time(s)
Before
Before

After

After

Fig. 6. Source current before and after compensation & comparison of

reference and actual compensation current
Non optimized 26.19 1.9 0.731 0.970
Optimized 26.19 3.14 0.731 0.981

As the time taken by the FFVS method is considerably

large, it is not suitable for online operations. Hence, a
high speed controller is required, and a neural network
controller is proposed in this case. The conventional
FFVS method is used to obtain the training samples for
the neural network because of its accuracy. For the
design of the neural network controller, the low current Fig. 7. Harmonic spectrum after compensation
load is considered. For creating samples to train neural
network, the APLC output (compensation current) is As harmonics have many adverse effects on the
generated using the FFVS control technique [28]. 400 system, and the power factor needs to be as high as
different nonlinear conditions with low current are possible, a proper control strategy must be provided to
simulated for the system. Initially, for each condition, mitigate the harmonics, and to improve power factor.
voltage at PCC and load current were estimated. Therefore, a flexible algorithm was proposed to limit the
Optimized conductance for each harmonics was obtained total and individual harmonic distortion, and to improve
by solving the optimization problem explained. From the the power factor under non-sinusoidal supply current
population of NSGA-II, the operating point is selected conditions. The proposed APLC was controlled using a
when THD is about 2%. The compensation current neural network. A neural network was used to estimate
i*com(t), to be injected by APLC, is calculated using (9), compensation current accurately online, and previously,
and the switching states are obtained using (10) and (11). the NSGA-II algorithm was used to optimize the
estimated line current offline. Exhaustive simulations

Copyright © 2011 Praise Worthy Prize S.r.l. - All rights reserved International Review of Electrical Engineering, Vol. 6, N. 5

2608
 D. Kavitha, A. F. Zobaa, P. Renuga, V. Suresh Kumar 

were carried out with different load conditions. Based on energy storage components,” IEEE Trans. Industry Applications,
vol. IA-20, pp. 625–630, May/June 1984.
the simulation results, it can be concluded that the [19] J.W. Dixon, J.J. Garcia, and L. Moran, “Control system for three-
proposed active power line conditioner performs phase active power filter which simultaneously compensates
satisfactorily for the compensation of harmonics and power factor and unbalanced loads,” IEEE Trans. Industrial
reactive power. It makes the line current waveform as Electronics, vol. 42, pp. 636–641, Dec. 1995.
[20] A. Cavallini and G.C. Montanari, “Compensation strategies for
close as possible to a sinusoid, and actually improves the shunt active-filter control,” IEEE Trans. Power Electronics, vol.
power factor by reducing the apparent power supplied by 9, no. 6, pp. 587–593, Nov. 1994.
the utility. [21] S.M. Williams and R.G. Hoft, “Adaptive frequency domain
control of PWM switched power line conditioner,” IEEE Trans.
Power Electronics, vol. 6, pp. 665-670, Oct. 1991.
References [22] W.R.A. Ibrahim and M.M. Morcos, “Artificial intelligence and
advanced mathematical tools for power quality applications: A
[1] E.L.O. Wen, "History of harmonics in power system," IEEE survey,” IEEE Trans. Power Delivery, vol. 7, pp. 668-673, Apr.
Industry Application Magazine, pp. 6-12, Jan/Feb 1998. 2002.
[2] W.M. Grady and S. Santoso, “Understanding power system [23] Koray Şener Parlak, Mehmet Özdemir, and M. Timur Aydemir,
harmonics,” IEEE Power Engineering Review, vol. 1, pp. 8-11, "Elimination of voltage harmonics caused by nonlinear loads in
Nov. 2001. distributed power systems consisting of inverters," International
[3] V.E. Wagner, “Effect of harmonics on equipment,” IEEE Trans. Review of Electrical Engineering (IREE), vol. 4, no. 2, pp. 228-
Power Delivery, vol. 8, no. 2, pp. 672–680, Apr. 1993. 234, April 2009.
[4] R. Barnes, "Harmonics in power system, "Power Engineering [24] J.R. Vazquez, and P. Salmeron, “Active power filter control using
Journal, pp. 11-15 Jan 1989. neural network technologies,” IEE Proc. Elect. power appl.,
[5] IEEE recommended practices and requirements for harmonic vol.150, no.2, pp. 139 – 145, March 2003.
control in electrical power systems, ANSI/IEEE Std.519, 1992. [25] A. Dinu, and M.N. Cirstea, “A Novel neural PWM Controller,” in
[6] Sincy George, and Vivek Agarwal, “Optimum control of selective proc. 1998 IEEE International Conference on Simulation, pp.
and total harmonic distortion in current and voltage under non- 375-379.
sinusoidal conditions,” IEEE Trans. Power Delivery, vol. 23, no. [26] Yen-ju Wang and R.M. O’connell, “Experimental evaluation of
2, Apr. 2008. an active power line conditioner based on a neural control
[7] H. Yahia, N. Liouane, and R. Dhifaoui, "Weighted differential scheme,” in proc. 2000 IEEE conf., pp 493-497.
evolution based PWM optimization for single phase voltage [27] Phumin Kirawanich and Robert M.O’connell, “Fuzzy logic
source inverter," International Review of Electrical Engineering control of an active power line conditioner,” IEEE Trans. Power
(IREE), vol. 5, no. 5, pp. 1956-1962, Oct. 2010. Electronics, vol.19, no.6, pp.1574-1585, Nov. 2004.
[8] Ghasemi, and S.S. Mortazavi, "Comparison of different control [28] V. Suresh Kumar, D. Kavitha, K. Kalaiselvi, and P.S. Kannan,
strategies for robust shunt active power filter," International “Harmonic mitigation and power factor improvement using fuzzy
Review of Electrical Engineering (IREE), vol. 4, no. 6, pp. 1147- logic and neural network controlled active power filter,” Journal
1155, Dec. 2009. of Electrical Engineering & Technology, vol.3, no.3, pp. 520-527,
[9] Mau Teng Au, and J.V. Milanovic, “Planning approaches for the 2008.
strategic placement of passive harmonic filters in radial [29] V. Suresh Kumar, D. Kavitha, K.Kalaiselvi, and P.S. Kannan,
distribution networks,” IEEE Trans. Power Delivery, vol. “Optimal estimation of harmonics in power system using
22, no.1, pp. 347 – 353, Jan. 2007. intelligent computing techniques,” in proc. 2007 IEEE Neural
[10] H. Akagi, “New trends in active filters for power conditioning,” Networks conference, Orlando, USA.
IEEE Trans. Industry Application, vol.32, pp.1312- 1322, Nov. [30] C.M. Fonseca and P J. Fleming, “Genetic algorithms for multi
1996. objective optimization: formulation, discussion and
[11] H. Akagi and H. Fujita, “A new power line conditioner for generalization,” in proc. 1993 5th International Genetic
harmonic compensation in power distribution systems,” IEEE Algorithms conf., San Mateo, CA, pp. 416–423.
Trans.Power Delivery, vol.10, no.3, pp.1570-1575, July 1995. [31] N. Srinivas and K. Deb, “Multi-objective optimization using non-
[12] M.S.A. Dahidah, and V.G. Agelidis, "Optimal SHE-PWM dominated sorting in genetic algorithms,” IEEE Trans.
technique for three-level voltage source converter control," Evolutionary Computing, vol. 2, no. 3, pp.221–248, 1994.
International Review of Electrical Engineering (IREE), vol. 3. [32] K. Deb, Multi-objective optimization using evolutionary
no.5, pp. 874-880, Oct. 2008. algorithms, New York: Wiley, 2001.
[13] M. Davari, A.R. Pourshoghi, M.R. Hajiahmadi, G.B.
Gharehpetian, and S.H. Fathi, "New Hybrid Controller based on
Sliding Mode Modulation for Inverter-Based DG with Existing Authors’ information
Fast Load Changes," International Review of Electrical
1
Engineering (IREE), vol. 5, no. 3, pp. 1078-1086, June 2010. Department of Electrical and Electronics Engineering,
[14] Fang Zeng Peng, “Application issues of active power filters,” Thiagarajar College of Engineering,
IEEE Industry Applications Magazine, vol.4, no. 5, pp. 21- 30, Madurai-625015 India.
Sep.-Oct. 1998.
2
[15] Po-Tai Cheng, S. Bhattacharya, and D. Divan, “Experimental School of Engineering and Design,
verification of dominant harmonic active filter for high-power Brunel University, United Kingdom.
applications,” IEEE Trans. Industry Applications, vol. 36, no. 2,
pp.567-577, March-April 2000. D. Kavitha graduated in Electrical and
[16] Masume Khodsuz, and Abdolreza Sheikholeslami, "Voltage Electronics Engineering in 2003 from Madurai
regulation and harmonic compensation in stand-alone inverter- Kamaraj University, India, and holds a post-
based distributed generation using non-sinusoidal unipolar graduate in Power systems from Anna
hysteresis voltage control," International Review of Electrical University, India in 2005. Currently, she works
Engineering (IREE), vol. 4, no. 5, pp. 792-798, Oct.2009. as Assistant Professor in Electrical and
[17] H.D. Battista and R.J. Mantz, “Harmonic series compensators in Electronics Engineering at Thiagarajar College
power systems: Their control via sliding mode,” IEEE Trans. of Engineering, Madurai, India. Her areas of
Control System, vol. 8, pp. 939–947, Nov. 2000. interest are power system and power quality.
[18] H. Akagi, Y. Kanazawa, and A. Nabae, “Instantaneous reactive
power compensators comprising switching devices without

Copyright © 2011 Praise Worthy Prize S.r.l. - All rights reserved International Review of Electrical Engineering, Vol. 6, N. 5

2609
 D. Kavitha, A. F. Zobaa, P. Renuga, V. Suresh Kumar 

Ahmed Faheem Zobaa holds BSc (hons), MSc Suresh Kumar V graduated in Electrical &
and PhD degrees in Electrical Power & Electronics Engineering from Madurai Kamaraj
Machines from the Faculty of Engineering at University, India in 1994, with a post-graduate
Cairo University, Giza, Egypt, from 1992, 1997 qualification in Power systems from National
and 2002. Institute of Technology, Tiruchi, India in 1995.
Currently, he is a Senior Lecturer in Power He then obtained his PhD in Power quality &
Systems at Brunel University, UK. Also, he has Harmonics from Madurai Kamaraj University in
been an Associate Professor at Cairo University, 2006. Since 1997, he has been faculty of
Egypt, since April 2008. He was a Senior Lecturer in Renewable Electrical and Electronics Engineering at Thiagarajar College of
Energy at the University of Exeter, UK, from 2007 to 2010, an Engineering, Madurai, India, and currently works as an Associate
Instructor in the Department of Electrical Power & Machines, with the Professor. He has published 30 papers for national and international
Faculty of Engineering at Cairo University, Giza, Egypt, from 1992 to journals and conferences. His fields of interest include power quality
1997, Teaching Assistant from 1997 to 2002, and Assistant Professor and power electronics.
from 2003 to April 2008. His main areas of expertise are power
quality, photovoltaic energy, wind energy, marine renewable energy,
grid integration, and energy management.
Dr. Zobaa is Editor-In-Chief for the International Journal of
Renewable Energy Technology. He is also an Editorial Board member,
Editor, Associate Editor, and Editorial Advisory Board member for
many international journals.
Dr. Zobaa is a registered Chartered Engineer and European Engineer,
and a registered member of the Engineering Council UK, Egypt
Syndicate of Engineers, and Egyptian Society of Engineers.

Dr. P. Renuga obtained her BE (EEE), ME

(Power system), and PhD (Power system
reliability) degrees from Thiagarajar College of
Engineering, Madurai Kamaraj University in the
years 1982, 1993 and 2006, respectively. Her
fields of interest are electrical machines, power
generation, and power system reliability and
planning. She has published more than ten
national and international journal papers, and given more than forty
national and international conference papers. She has done an AICTE
sponsored project for 7 lakhs under the title “Brushless DC motor
position control”. Now, she works as Associate professor in Electrical
and Electronics engineering department at Thiagarajar College of
Engineering. Currently ten candidates are pursuing a PhD under her
guidance.

Copyright © 2011 Praise Worthy Prize S.r.l. - All rights reserved International Review of Electrical Engineering, Vol. 6, N. 5

2610