Expert Systems with Applications 38 (2011) 692–699

Contents lists available at ScienceDirect

Expert Systems with Applications

journal homepage: www.elsevier.com/locate/eswa

Elimination of lower order harmonics in Voltage Source Inverter feeding an

induction motor drive using Evolutionary Algorithms
V. Jegathesan a,*, Jovitha Jerome b
a
Department of Electrical and Electronics Engineering, Karunya University, Coimbatore 641 114, India
b
Department of Instrumentation and Control Systems Engineering, P.S.G College of Technology, Coimbatore 641 004, India

a r t i c l e i n f o a b s t r a c t

Keywords: Selective Harmonic Elimination technique is one of the control methods applied in Voltage Source Invert-
Inverter ers to eliminate the harmonics. However, finding the solutions for the harmonic reduction is a difficult
Harmonics problem to be solved. This paper presents an efficient and reliable Evolutionary Algorithms based solu-
Genetic Algorithm (GA) tion for Selective Harmonic Elimination (SHE) switching pattern to eliminate the lower order harmonics
Evolutionary Programming (EP)
in Pulse Width Modulation (PWM) inverter. Determination of pulse pattern for the elimination of lower
Pulse Width Modulation
Selective Harmonic Elimination
order harmonics of a PWM inverter necessitates solving a system of nonlinear transcendental equations.
Evolutionary Algorithms are used to solve nonlinear transcendental equations for PWM–SHE. In this pro-
posed method, harmonics up to 19th are eliminated using Evolutionary Algorithms without using dual
transformer. The experimental results are obtained and are validated with simulations using PSIM 6.1
and MATLAB 7.0.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction processor based implementation because of various sinusoidal

voltages and frequencies are required in the system.
Elimination of harmonics has been the subject of intensive re- Eliminating lower order harmonics using Programmed PWM
search in power electronic equipments. The usage of power elec- (Sun & Grotsotollen, 1992) generates high quality output spectra,
tronic equipments has increased in recent years in industrial and which in turn results in minimum current ripples, thereby satisfy-
consumer applications. Such loads draw the nonlinear sinusoidal ing several performance criteria and contribute to overall im-
current and voltage from the source (IEEE Recommended Practices proved performance. Performance characteristics of a rectifier/
& Requirements for harmonic control in Electrical Power Systems, inverter power conversion scheme largely depend on the choice
1992; Wagner, 1993). These nonlinear loads change the sinusoidal of the particular Pulse Width Modulation strategy employed. Pro-
nature of the AC current, thereby resulting in the flow of harmonic grammed PWM techniques optimize a particular objective func-
currents in the AC power system. These harmonics flow through tion, such as selective elimination of harmonics and therefore are
the power system, where they can distort the supply voltage, over the most effective means of obtaining high performance.
load electrical distribution equipments and resonate with power The programming PWM schemes are applied for single and
factor correction capacitors. three phase inverters (Enjeti, Ziogas, & Lindsay, 1990), thereby pro-
The power electronic equipments use Pulse Width Modulation viding the framework and guidelines for the selection of the appro-
(PWM) and in recent years, most of the applications are dominated priate technique for each application. Also provide the guidelines
by PWM inverters due to their ability of providing both voltage and for solving nonlinear equations associated with each one of the
frequency control at one stage and generating output voltage and harmonic elimination PWM’s for small and very large degree of
current waveform with low harmonic distortion. Different types freedom. Third harmonic current injection technique (Eltamaly,
of feed forward and feedback Pulse Width Modulation schemes 2008) is used to reduce the line current harmonic content and
having relevance for industrial application have been widely dis- increasing the power factor of the controlled and uncontrolled
cussed (Holtz, 1992). The Pulse Width Modulated inverter is most converters.
favored one for industrial application. The control scheme of PWM An optimized PWM technique was proposed in Shi and Li
inverters are broadly classified as programmed PWM inverter and (2005), to reduce harmonic distortion and to spread harmonic en-
sinusoidal PWM inverter. Sinusoidal PWM is not suited for micro- ergy for high switching frequency inverters. Optimization algo-
rithms are becoming increasingly popular in engineering design
activities, where the emphasis is on the maximizing or minimizing
* Corresponding author.
certain goal primarily because of the availability and affordability
E-mail address: [email protected] (V. Jegathesan).

0957-4174/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.eswa.2010.07.021
 V. Jegathesan, J. Jerome / Expert Systems with Applications 38 (2011) 692–699 693

of high speed computers. They are extensively used in engineering The power circuit for Voltage Source Converter drive system is
design problems. The 5th and 7th harmonics are eliminated in the given in Fig. 1. DC voltage is obtained using six pulse Voltage
five level SHE PWM converters by optimal switching transition Source Rectifier. The rectifier is connected to Voltage Source Inver-
through the Fourier theory (Dohidah, Agelidis, & Rao, 2006). ter through DC link capacitor and DC link Inductor.
A direct minimization of nonlinear transcendental trigonomet- This paper presents an Evolutionary Algorithms method for
ric Fourier function in combination with a random search was reduction of line voltage harmonics in PWM inverter without dual
illustrated to obtain multiple sets of solution for uni-polar PWM transformer. The use of dual transformer and 12 pulse rectifier are
circuits (Agelidis, Balouktsis, & Cossar, 2008). Generalized methods avoided to eliminate the characteristic harmonics 5th, 7th, 11th,
are developed for eliminating a fixed number of harmonics in the 13th, 17th and 19th for six pulse rectifier. The objective is achieved
half bridge and full bridge inverter (Chiasson, Tolbert, McKenzie, by determining the switching pattern for the three phase Inverter
& Du, 2004; Patel & Hoft, 1973, 1974; Wells, Nee, Chapman, & using Genetic Algorithm and Evolutionary Programming methods.
Krein, 2005) output waveforms and solutions are presented for Experimental results and simulation were carried out and vali-
eliminating up to 5th harmonics. dated using MATLAB 7.0 and PSIM 6.1.
The minimization of objective function used for SHE was done
using traditional mathematical techniques such as Conjugate Gra- 2. Problem formulation
dient Descent method (CGD) (Maswood, Wei, & Rahman, 2001)
and Newton Raphson method (NR) (Sun, Beineke, & Grotsllen, The Fourier coefficients of the PWM–SHE switching pattern for
1994). These methods need initial values to obtain the objective a three phase line to neutral voltage are given by Eq. (1).
function and are based on differential information, so they may " #
produce local minimum solution which leads to undesirable 4 XN
kþ1
an ¼ ð1Þ cosðnak Þ
pattern. np k¼1 ð1Þ
GA is one of the non-traditional programming which provides
bn ¼ 0
solution to nonlinear mathematical problems. GA is inspired by
the mechanism of natural selection, in which stronger individuals Eq. (1) has N variables (a1 to aN) and a set of solution are ob-
are likely to survive in a competing environment. GA uses a direct tained by equating N  1 harmonics to zero and assigning a specific
analogy of such selection. EP is a technique in the field of evolu- value of the fundamental amplitude a1, through Eq. (2).
tionary computation. It is a powerful and general global optimiza- " #
tion method which does not depend on the first and second 4 XN
kþ1
f1 ðaÞ ¼ ð1Þ cosðak Þ  M ¼ e1
differentials of the objective function of the problem to be opti- p k¼1
mized. The EP (Sinha, Chakrabarti, & Chattopadhyay, 2003) tech- " #
4 XN
kþ1
nique is based on the mechanism of natural selection. f2 ðaÞ ¼ ð1Þ cosð5ak Þ ¼ e2
As demonstrated in Maswood and Wei (2005), GA is applied to 5p k¼1 ð2Þ
eliminate the lower order harmonics in power converter with ...... ...... ......
dual transformer and 12 pulse rectifier. The 3rd and other triplen " #
4 X N
harmonics can be ignored if the machine has an isolated neutral. fN ðaÞ ¼ ð1Þkþ1 cosðnak Þ ¼ eN
A Genetic Algorithm optimization technique is applied to deter- np k¼1
mine the switching angles for a cascaded multilevel inverter
where the variables e1  eN are the normalized amplitude of the
(Ozpineci, Tolbert, & Chiasson, 2005) which eliminates specified
harmonics to be eliminated. The objective function of PWM–SHE
higher order harmonics while maintaining the required funda-
technique is to minimize the harmonic content in the inverter line
mental voltage. A seven level inverter is considered and optimum
current and it is given in Eq. (3).
switching angles are calculated offline to eliminate 5th and 7th
harmonics. Fða1 ; a2 ; a3 ; . . . ; aN Þ ¼ e21 þ e22 þ    þ e2N ð3Þ
Genetic Algorithm based approach is used for the optimal de-
sign of passive shunt compensator (Zacharia, Menti, & Zacharias, Subjected to the constraint equation ð4Þ; 0 < a1 < a2 < a3
2008) when non-ideal, non-sinusoidal voltage source supplies a p
linear load. No preliminary calculations are necessary to identify < a4 < a5    aN < ð4Þ
2
compensator parameter values which would create resonant con-
ditions at the harmonic frequencies of the source. GA and CGD for Quarter-wave symmetric pulse pattern. In the proposed method
methods are used to find the switching pattern for SHE to elimi- a1, a2, a3, a4 and a5 solutions are expected with elimination of 5th,
nate rectifier low input current harmonics without having any ini- 7th, 11th, 13th, 17th and 19th harmonics.
tial guess for the switching pattern. GA is used to provide the initial
values (Shen & Maswood, 2000). A five level PWM–SHE technique 3. Harmonic elimination using Genetic Algorithm method
for Voltage Source Converters (Maswood, Neo, & Rahman, 2001)
has been proposed by finding switching transitions without quar- Genetic Algorithms are numerical optimization algorithms
ter and half wave symmetry for the input waveform to remove har- based on the principles inspired from the genetic and evolution
monics up to 7th with the help of GA. mechanisms observed in natural systems and population of living
NR and GA are adopted to reduce the lower order line current beings. Binary encoding GA is dealing with binary strings, where
harmonics by developing ‘N’ number of pulses per half cycle (Sund- the number of bits of each string simulates the genes of an individ-
areswaran & Chandra, 2002). Genetic Algorithm and Particle ual chromosome, and the number of individuals constitutes a pop-
Swarm Optimization are applied to compute the switching angles ulation. Each parameter set is encoded into a series of a fixed
in a three phase seven level inverters (Barkati, Baghali, Berkouk, length of string symbols usually from the binary bits, which are
& Boucherit, 2008) to produce the required fundamental voltage, then concatenated into a complete string called chromosome. Sub-
while at the same time, specific harmonics are eliminated. 5th, strings of specified length are extracted successively from the con-
7th and multiples of 3rd harmonics are suppressed in line-to-line catenated string and are then decoded and mapped into the value
voltage. The line voltage total harmonic distortion increases in the corresponding search space. Generally, GA implementation
slightly when the modulation index decreases. comprises the procedures of initial population generation, fitness
694 V. Jegathesan, J. Jerome / Expert Systems with Applications 38 (2011) 692–699

Fig. 1. Power circuit of Voltage Source Converter drive system.

evaluation and genetic operations of selection, crossover and selected for fixed length chromosomes, the cut point between the
mutation. first and the last gene of the present chromosomes. For variable
In this paper, an attempt has been made to determine the most length chromosomes, the cut point is between the first and the last
optimal switching pattern to eliminate the lower order line voltage gene of the present chromosome with minimum length.
harmonics in the Voltage Source Inverter. The implementation of
GA algorithm is given below. 3.5. Step 5: Mutation

3.1. Step 1: Initialization In accordance with the mutation rate, some chromosomes from
the population produced after crossover will undergo mutation.
The initial population (Pi) of N chromosomes is generated with Random genes are selected and altered from 0 to 1 or vice versa.
randomly selected initial individual switching angles. The gener- In this way, the old population is replaced with the improved pop-
ated switching angles are distributed uniformly between their ulation generated through steps 2–5.
minimum and maximum limits by satisfying Eq. (4). Each chromo-
some is coded as binary string which is mapped into a real number.
3.6. Step 6: Elitism

3.2. Step 2: Fitness of the candidate solutions

The crossover and mutation for the two chromosomes are re-
peated until all of the chromosomes of the parent generation are
The Fitness function Value (FV) in this case attempts to mini-
replaced by the newly formed chromosomes. The best chromo-
mize the objective function using the given Eq. (5). Fitness function
some of the parent generation and the best chromosome found
Value can be calculated for each chromosome in the population.
in all of the previous generations are copied intact to the next gen-
1 eration, so that the possibility of their destruction through a genet-
FV ¼ ð5Þ ic operator is eliminated.
1 þ f ðaÞ
where f(a) can be calculated using Eq. (2).
3.7. Step 7: Termination criterion for GA
When the switching angles violate the minimum and maximum
value, the penalty factor is introduced to avoid violation. The alpha
The above said procedure from steps 2 to 6 is repeated until the
limit violation can be dealt with the violation coefficient value
maximum iteration count is reached.
using Eq. (6).
Vio coeff ¼ ð1 þ ½aði1Þ  aðiÞ Þ  q ð6Þ
4. Harmonic elimination using Evolutionary Programming
where q is the penalty parameter, aðiÞ is the ith value of a. method
In such cases, the objective function is calculated by multiplying
with the Vio_coeff value. After computing the Fitness function Va- Evolutionary Programming is powerful global optimization
lue of each individual chromosome, the parents then undergo ge- technique, has proved itself effective to handle complex optimiza-
netic operation of selection, crossover and mutation. After the tion problem. EP starts with a population of randomly generated
evaluation of the initial randomly generated population, GA begins candidate solution which evolves towards the better solution over
to create new generation. The process of selection and mating of a number of iterations. It uses the probabilistic rules to explore the
individuals continues until a new generation is reproduced. complex search space.
The primary objective of the PWM–SHE technique is to deter-
3.3. Step 3: Selection mine the most optimal switching pattern to eliminate the harmon-
ics in the Voltage Source Inverter. Evolutionary Programming is a
Chromosomes from the parent population are selected in pairs probabilistic search technique, which generates the initial parent
with a probability proportional to their fitness to replicate and vectors distributed uniformly in intervals within the limits and ob-
form offspring chromosomes. This selection scheme is known as tains global optimum solution over number of iterations. The main
Roulette Wheel selection. Each chromosome selects a percentage stages of this technique are initialization, creation of offspring vec-
of Roulette Wheel equal to its normalized fitness value. The chro- tors by mutation, competition and selection of best vectors to eval-
mosomes that will be copied are selected with rates proportional uate best fitness solution. The implementation of EP algorithm is
to their fitness. given below.

3.4. Step 4: Crossover 4.1. Step 1: Initialization

In accordance with the crossover rate, randomly chosen pairs of The initial population comprises of combination of all the
parent chromosome from the population produced after reproduc- switching angles generated after satisfying Eq. (4). The elements
tion undergo crossover to produce offspring. A random cut point is of parent vectors (Pi) are the various alpha values of generated
 V. Jegathesan, J. Jerome / Expert Systems with Applications 38 (2011) 692–699 695

population distributed uniformly between their minimum and Trajectory of Switching Angles
maximum limits. 100

Electrical Degrees
80 Alpha 5
Alpha 4
4.2. Step 2: Fitness of the candidate solutions 60
Alpha 3
40 Alpha 2
Alpha 1
The Fitness function Value (FV) in this case attempts to mini- 20
mize the error between the actual angles to the exact values of 0
the same, which is assumed to be the alpha values of the corre- -20 0 0.2 0.4 0.6 0.8 1 1.2 1.4
Modulation Index
sponding minimum objective function.
Fig. 2a. Switching angles versus the modulation index (using GA).
4.3. Step 3: Creation of offspring vector by mutation

An offspring vector P0i is created from each parent vector by add-

Trajectory of Switching Angles
ing Gaussian random variable with zero mean and standard devi-
  100
ation ri, denoted as N 0; r2i .

Electrical Degrees
80 alpha 1
 
P0i ¼ Pi þ N 0; r2i for i ¼ 1; 2; . . . ; Np  1 ð7Þ 60
alpha 2
alpha 3
fi
where ri ¼ b: f ðPimax  Pimin Þ. Np is the number of population; b 40 alpha 4
min
is the scaling factor; fi is the fitness of the ith individual; fmin is the 20
alpha 5
minimum fitness of the entire population.
0
The created offspring vector must satisfy the minimum and 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2
maximum generation limits of the units. After adding a Gaussian Modulation Index
random number to the parents, the element of offspring may vio-
late the constraint given by (4). Fig. 2b. Switching angles versus the modulation index (using EP).
These violations are dealt as follows:
(
Pimin if P0i < P0imin
P0i ¼ ð8Þ
Pimax if P0i > P0imax

4.4. Step 4: Selection and competition

In the competition stage, a selection mechanism is used to pro-

duce a new population from the existing population and the pop-
ulation is created by mutation. The selection technique used is the
Stochastic Tournament method described in the following.
The parent solution Pi along with their corresponding offsprings
formed by mutation P 0i , each under goes a series of Nt tournaments
with randomly selected opponents. Each individual is assigned a
score ws according to:

X
Nt
ws ¼ wi
i¼1 Fig. 3a. Simulated inverter output voltage waveform for induction motor drive
(
(using GA).
1 if u1 > DfsDþfsDfr
wt ¼ ð9Þ
0 otherwise

where

Dfs ¼ fs  fmin
Dfr ¼ fr  fmin

fs is the objective function of the individual under consideration, fr is

the objective function of randomly selected opponent individual
and fmin is the minimum objective function of an individual within
the two population. The opponent is chosen at random from 2  Np
individuals based on r = [2  Np  u2 + 1]. u1, u2 are uniform random
numbers in the interval [0, 1]. Individuals are ranked in descending
order of their corresponding ws score. The first Np individuals are se-
lected and transcribed along with their corresponding fitness values
to be the parents of the next generation.

4.5. Step 5: Termination criterion for EP

The above procedure from steps 2 to 3 are repeated until max- Fig. 3b. Simulated inverter output voltage waveform for induction motor drive
imum iteration count is reached. (using EP).
696 V. Jegathesan, J. Jerome / Expert Systems with Applications 38 (2011) 692–699

5. Optimization results 6. Simulation and experimental results

After solving the five nonlinear functions of Eq. (2) simulta- After obtaining the switching angles through the MATLAB using
neously using MATLAB 7.0 optimization toolbox, five angles are ob- Evolutionary Algorithms, the proposed system is developed using
tained. This process is repeated for the various modulation indices PSIM 6.1. The circuit uses three 230 V single phase AC supply
from 0.1 to 1.3. The trajectory of calculated switching angles of source which are connected to the star connected primary winding
proposed PWM–SHE switching pattern using GA is shown in of the three phase star/delta transformer. The six pulse Voltage
Fig. 2a.
The trajectories of the switching angles calculated using genetic
algorithm are almost smooth for a1, a2 and a5 over the whole range
of possible modulation indices. There is an abrupt rise of 25° for a4
and fall of about 25° for a3 in the modulation index 0.7. All five an-
gles are smooth after M = 0.8.
The trajectory of calculated switching angles of proposed
PWM–SHE switching pattern using Evolutionary Programming is
shown in Fig. 2b. In this approach, the trajectory of the angles
are smooth from M = 0.1 to M = 0.6 and from M = 0.8 to M = 1.3.
We can find in the trajectory that there is a sudden change in the
path of trajectory at M = 0.7.
All these characteristics bring unpredictability to traditional
algorithms that require precise initial values to guarantee conver-
gence. In this paper both Genetic Algorithm and Evolutionary Pro-
Fig. 5a. Harmonics spectrum for inverter output voltage for induction motor drive
gramming are used because of discrete nature of harmonics to be (after eliminating 5th, 7th, 11th, 13th, 17th and 19th harmonics using GA) at
eliminated. M = 0.9.

Fig. 4a. Experimental inverter output voltage waveform for induction motor drive (using GA).

Fig. 4b. Experimental inverter output voltage waveform for induction motor drive (using EP).
 V. Jegathesan, J. Jerome / Expert Systems with Applications 38 (2011) 692–699 697

were carried out on a Pentium IV 2.4 GHz, 512-MB RAM processor.

The coding was written using MATLAB 7.0.
The harmonics are to be observed in the output line-to-line
voltage. With the five switching angles calculated, the whole
switching pattern is constructed using quarter wave symmetry
method. The output line-to-line voltage waveforms for the modu-
lation index, M = 0.9 with induction motor drive are shown in
Figs. 3a and 3b respectively.
A low power laboratory prototype Voltage Source Inverter,
based on MOSFET Voltage Source Inverter was developed and
tested to verify the feasibility and validity of theoretical and simu-
Fig. 5b. Harmonics spectrum for inverter output voltage for induction motor drive
lation findings. The pre-calculated PWM signals are implemented
(after eliminating 5th, 7th, 11th, 13th, 17th and 19th harmonics using EP) at by a low cost, high speed ATMEL microcontroller. A digital real-
M = 0.9. time power quality analyzer (ALM 30) was used to display and cap-
ture the output waveforms. The experimental inverter output volt-
age waveform for induction motor drive using GA and EP are given
Source Rectifier is developed using six diodes as bridge. This recti- in Figs. 4a and 4b respectively. The waveform of the output voltage
fier is being connected to Voltage Source Inverter through the shows that there is excellent correlation between the simulated re-
Inductor and Capacitor which is acting as a DC link between the sults waveform and experimental results waveform.
rectifier and inverter. The load to the proposed converter is a three Harmonic Spectrum is obtained for the inverter output voltage
phase squirrel cage induction motor. In this method, dual trans- with induction motor drive for both Genetic Algorithm and Evolu-
former and 12 pulse rectifier are avoided to eliminate certain lower tionary Programming method. The Harmonics Spectrum of output
order harmonics. High capacity dual transformer connections are voltage with induction motor drive is shown in Figs. 5a and 5b
also avoided to eliminate the lower order harmonics using Genetic respectively. Simulation results of output voltage harmonic spec-
Algorithm and Evolutionary Programming approach. Simulations trum shows that harmonics up to 19th are eliminated using GA

Fig. 6a. Experimental harmonics spectrum for inverter output voltage for induction motor drive (after eliminating 5th, 7th, 11th, 13th, 17th and 19th harmonics using GA) at
M = 0.9.

Fig. 6b. Experimental harmonics spectrum for inverter output voltage for induction motor drive (after eliminating 5th, 7th, 11th, 13th, 17th and 19th harmonics using EP) at
M = 0.9.
698 V. Jegathesan, J. Jerome / Expert Systems with Applications 38 (2011) 692–699

clearly shows the absence of the targeted harmonics while control-

ling the fundamental at a predefined value. The target is achieved
and harmonics up to 19th in Voltage Source Inverter feeding induc-
tion motor drive were totally eliminated. This highlights the
PWM–SHE method, which provides a clean power converter envi-
ronment and meets most accepted standards.

8. Conclusion

An efficient technique of calculating switching angles through

the Genetic Algorithm method is illustrated. A six pulse converter
is proposed as the power circuit for the three phase drive system.
Fig. 7a. Total harmonic distortion of output voltage using Genetic Algorithm.
PWM–SHE switching is proposed for three phase Inverter circuit.
This method avoids using 12 pulse rectifier and the traditional
complex calculations. Analysis of the harmonics spectrum shows
that, 5th, 7th, 11th, 13th 17th and 19th harmonics are eliminated
through Genetic Algorithm and Evolutionary Programming. In this
approach, the dual transformers are not used to eliminate the 5th
and 7th order harmonics. High capacity dual transformer connec-
tions are also avoided to eliminate the lower order harmonics
using Genetic Algorithm and Evolutionary Programming approach.
It remains to be a topic for further investigations to design similar
real-time PWM systems.

References

Agelidis, V. G., Balouktsis, A. I., & Cossar, C. (2008). On attaining the multiple
solutions of selective harmonic elimination PWM three level waveforms
through function minimization. IEEE Transactions on Industrial Electronics, 55,
996–1004.
Fig. 7b. Total harmonic distortion of output voltage using Evolutionary
Barkati, S., Baghali, L., Berkouk, E. M., & Boucherit, M.-S. (2008). Harmonics
Programming.
elimination in diode clamped multilevel inverter using evolutionary algorithms.
Electronics Power System Research, 78, 1736–1746.
Chiasson, J. N., Tolbert, L. M., McKenzie, K. J., & Du, Z. (2004). A complete solution to
and EP approach. The experimental harmonic spectrum of inverter the harmonic elimination problem. IEEE Transactions on Power Electronics, 19,
output voltage for induction motor drive using GA and EP are given 491–499.
Dohidah, M. S. A., Agelidis, V. G., & Rao, M. V. (2006). An abolishing symmetry
in Figs. 6a and 6b respectively. The lowest harmonics available in requirements in the formulation of a five level selective harmonic elimination
the output voltage is 23rd, which is considered to be the higher or- PWM technique. IEEE Transactions on Power Electronics, 21, 1833–1837.
der harmonics and that can be easily removed using LC filter cir- Eltamaly, A. M. (2008). A modified harmonics reduction technique for a three phase
controlled converter. IEEE Transactions on Industrial Electronics, 55, 1190–1197.
cuits. The harmonics spectrum figures show that almost all 5th, Enjeti, P. N., Ziogas, P. D., & Lindsay, J. F. (1990). Programmed PWM techniques to
7th, 11th, 13th, 17th and 19th harmonics are eliminated for the eliminate harmonics: A critical evaluation. IEEE Transactions on Industrial
modulation index value 0.9. These are the characteristic lower or- Applications, 26, 302–316.
Holtz, J. (1992). Pulse width modulation – A survey. IEEE Transactions on Industrial
der harmonics to be eliminated for the six pulse converter. Electronics, 39, 410–420.
IEEE Recommended Practices and Requirements for Harmonic Control in Electrical
Power Systems, IEEE Std. 519-1992, 1992.
7. Discussion
Maswood, A. I., Wei, S., & Rahman, M. A. (2001). A flexible way to generate PWM–
SHE switching pattern using genetic algorithm. In: Proceedings of the IEEE
This paper points out that GA and EP shows great potential in applied power electronics conference, Anheim, CA, USA (pp. 1130–1134).
solving not only the required switching angles, but also in the opti- Maswood, A. I., Neo, T. G., & Rahman, M. A. (2001). An online optimal approach to
PWM SHE gating signal generation. IEEE Power Engineering Review, 61–62.
mization of nonlinear converter switching characteristics in gen- Maswood, A. I., & Wei, S. (2005). Genetic algorithm based solution in PWM
eral. MATLAB 7.0 solutions coupled with PSIM 6.1 converter switching. IEE Proceedings on Electric Power Applications, 152,
implementation results verify and yield good output voltage and 473–478.
Ozpineci, B., Tolbert, L. M., & Chiasson, J. N. (2005). Harmonic optimization of
current waveforms. multilevel converter using genetic algorithms. IEEE Power Electronics Letters, 3,
The novelty of the paper is elimination of lower order harmon- 92–95.
ics of low frequency inverter feeding induction motor drive with- Patel, H. S., & Hoft, R. G. (1973). Generalized techniques of harmonic elimination
and voltage control in thyristor inverters – Part I – Harmonic elimination. IEEE
out using dual transformer. Comparing Genetic Algorithm and Transactions on Industrial Application, 1A-9, 310–317.
Evolutionary Programming, the total Harmonic distortion is greatly Patel, H. S., & Hoft, R. G. (1974). Generalized techniques of harmonic elimination
reduced using Genetic Algorithm approach. The maximum value of and voltage control in thyristor inverters – Part II – Voltage control techniques.
IEEE Transactions on Industrial Application, 1A-10, 666–673.
Total Harmonic Distortion of output voltage using Genetic Algo-
Shen, W., & Maswood, A. I. (2000). A novel current source PWM drive topology with
rithm is obtained from Fig. 7a is around 2.2 and the maximum va- specific harmonic elimination switching. In: Proceedings of the Canadian
lue of Total Harmonic Distortion using Evolutionary Programming conference on electrical and computer engineering, Halifax, Canada (pp. 53–55).
Shi, K. L., & Li, H. (2005). Optimized PWM strategy based on genetic algorithms. IEEE
can be calculated from Fig. 7b is around 3.8. So we can say that the
Transactions on Industrial Electronics, 52, 1458–1461.
Total Harmonic Distortion of the output voltage is reduced more Sinha, N., Chakrabarti, R., & Chattopadhyay, P. K. (2003). Evolutionary programming
using the GA approach and that is more efficient than Evolutionary techniques for economic load dispatch. IEEE Transactions on Evolutionary
Programming. The spectrum of the output voltage obtained with a Computation, 7, 301–306.
Sundareswaran, K., & Chandra, M. (2002). Evolutionary approach for line current
feature of the THD shows very good correlation between experi- harmonic reduction in AC/DC converters. IEEE Transactions on Industrial
mental and simulation results. The spectrum of the output voltage Electronics, 49, 716–719.
 V. Jegathesan, J. Jerome / Expert Systems with Applications 38 (2011) 692–699 699

Sun, J., & Grotsotollen, H. (1992). Solving nonlinear equations for selective harmonic Wells, J. R., Nee, B. M., Chapman, P. L., & Krein, P. T. (2005). Selective harmonic
eliminated PWM using predicted initial values. In: Proceedings of the IEEE IECON, control: A general problems formulations and selected solutions. IEEE
San Diego, CA, USA (pp. 259–264). Transactions on Power Electronics, 20, 1337–1345.
Sun, J., Beineke, S., & Grotsllen, H. (1994). DSP based real-time harmonic elimination Zacharia, P., Menti, A., & Zacharias, T. (2008). Genetic algorithm based optimal
of PWM Inverters. In: Proceedings of the IEEE power electronics specialists design of shunt compensators in the presence of harmonics. Electric Power
conference, Taipei, Taiwan (pp. 679–685). System Research, 80, 728–735.
Wagner, V. E. (1993). Effect of harmonics on equipment. IEEE Transactions on Power
Delivery, 8, 672–680.