Minimization of Harmonics in Multi-Level Inverters Using
Bat Algorithm
Akash Tyagi
Electrical Engineering Division, Piro Technologies
Limited, New Delhi, India
akashtyagi0010@gmail.com
Abstract- Minimization of harmonics in Cascade multi-level
inverters is a complex optimization problem. These involve a
large number of nonlinear transcendental equations having
multiple minima. A solution to the harmonic minimization
problem using Bat Algorithm is presented in this paper. When
the number of switching angles is increased, it is not possible to
determine them with the help of a conventional iterative
technique or resultant theory method. But the proposed
algorithm is able to find the optimal values of the switching
angles when their number is increased. Simulations are done for
an 11-level H-bridge inverter and results are compared with the
theoretical results. Results show that the proposed method
effectively minimizes a large number of specific harmonics, and
the output voltage results in very low total harmonic distortion
and switching frequency.
Keywords­  Multilevel   inverters   (MLI),   Selective   Harmonic
Elimination­Pulse   Width   Modulation   (SHE­PWM),   Harmonics,
Bat Algorithm (BA), Particle Swam optimization (PSO)
I. INTRODUCTION
In the field of distributed energy resource, several batteries,
fuel cells, solar cells, and microturbines can be connected to
multi-level inverters without the voltage balancing problems,
due to which they are drawing much attention in the last few
years [1]. MLI is accepted worldwide because of the ability
to reduce the harmonics of the AC Voltage waveforms, high-
voltage capability and operate efficiently. These H-bridge
MLI incorporate the desired output voltage from several
levels of DC voltage as input. An approximately sinusoidal
output voltage waveform can be obtained from several levels
of DC voltage.
The step-wise output of the MLI is a major advantage and
this results in better electromagnetic compatibility, lower
switching losses, higher power quality, higher voltage
capability, and no need of the transformer at distribution
voltage level, thereby reducing the cost of the system. There
Asif Iqbal
Computer Science Division, Piro Technologies, Private
Private Limited, New Delhi, India
asif.iqbal256003@gmail.com
are basically three categories in which MLI is divided:
Diode-clamped, cascade H-bridge multilevel inverters, and
flying capacitors. Amongst the above, cascade multilevel has
the highest power, high reliability, and high voltage levels
due to its modular structure [2]. This MLI is based on series
connected single phase H-bridge inverter, each with their
own isolated bus to get a sinusoidal voltage output [3]. In
high AC power supplies and high-speed drive applications,
these MLI are very popular.
Comparing to the square wave inverter, the output voltage of
MLI is in a stepped form resulting in reduced harmonics [1].
There are several well-known topologies of MLI to further
reduce the harmonics in MLI.
1. Sinusoidal Pulse Width Modulation (SPWM)
2. Optimized Harmonic Stepped-Waveform Technique
(OHSW)
3. Selective Harmonics Eliminated Pulse Width
Modulation (SHE-PWM)
From the techniques discussed above, the optimized
harmonic stepped-waveform is best suited for MLI
topologies. A low THD output waveform and switching
devices with only one switch per cycle can be obtained with
the implementation of this technique along with the multi-
level topology. The problem of switching loss, as well as
EMI, is also improved. Distortion in the electrical power
system is called harmonics. The harmonic current and
reactive power from the AC source affects the quality of the
power. Non-linear loads are the main reason for harmonics,
as they draw the non-sinusoidal current from a sinusoidal
voltage. The well-known loads responsible for producing
harmonics are static-VAR compensators, Inverters, Switch-
mode power supply, and electric arc furnace. These loads
produce harmonics currents which are injected back into the
supply systems and can interact adversely with the equipment
of the power system such as transformers, capacitors, energy,
motor and demand metering causing additional losses (S-1) switching angles are used to eliminate most noticeable
overheating an overloading. These can also cause lower order harmonics. The magnitude of the output phase
interference with telecommunication lines, and affect the voltage is:
normal working of computers, controllers, telephone system,
and sophisticated electronic equipment such as relays V1+V2+V3 (2)
resulting in false operation [4]. The most important aspect of
these inverters is the harmonic content. Because of the
staircase waveform of MLI, the harmonic content is lesser in
these inverters [2].

In many applications of MLI, the THD output voltage of the

inverter is a measurement of the harmonic distortion, which
should be very low. THD is defined as the Summation of all
harmonic component of the voltage or current waveform
compared against the fundamental component of the voltage
or current wave [5]:

√
∞

∑ Vn
2
Fig.1 H-Bridge
n=3,5,7. .
THD=
V1
(1)

II. CASCADE MULTILEVEL INVERTER

The three-level inverter is the basic of MLI and has drawn

the tremendous interest in the industry of the power system.
Compared to the simple inverters, MLI produces the high
voltage and high power. Cascade MLI is the most important
topology among the family of MLI. The components required
in the MLI are less as compared to the other two topologies.
It has a modular structure and simple switching strategy,
require less space [7].

A series of single-phase full-bridge inverter units forms a

cascade MLI as shown in fig-1. The key function of the MLI Fig-2 Output Voltage of 7-Level Cascade Inverter
is to produce the desired voltage from SDCS (separate DC's
source) connected to every single phase full bridge inverter By applying the Fourier series analysis for a cascade MLI
and can generate three different outputs, +Vdc, 0, -Vdc. The with ‘S’ switching angles, the amplitude of any odd nth
AC output voltage of every single phase full bridge is harmonic can be expressed as:
connected in series such that output voltage waveform is the
sum of all the individual H-Bridge Output [8]. By connecting k
the required number of H-Bridge in cascade and using proper n α¿
modulation technique, closely sinusoidal output voltage cos ( n α 1 ) +..+cos ⁡¿ [ sin(nωt ) ]
waveform can be synthesized. The number of levels in the
output voltage waveform is 2S+1, where ‘S’ is the number of Van (wt) = 4 Vdc (3)
¿
H-bridges. For e.g.- the AC output phase voltage waveform nπ
of a seven level MLI using two H-Bridges is shown in fig-2. ∞
In a seven level multi inverter there are three switching ∑ ¿
angles and three non-linear transcendental equations. From n=1,3,5
‘S’ number of switching angles only one switching angle is
used to calculate the fundamental voltage and the remaining
Where ‘n’ is an odd harmonic order, and switching angles are V1
considered equal to the number of dc sources and α k is
4 Vdc
the kth switching angle. The amplitude of the entire even
harmonic is zero. The switching angles should be in the range
M=
( (
3∗
π ))
of 0 to 90 considering 5th and 7th order output phase voltage (6)
to zero. For a seven-level cascade MLI, the fundamental
voltage in terms of switching angles is given: The seven level cascades MLI require two H-bridge. To
obtain the required switching angles and the desired
4 Vdc fundamental voltage nonlinear switching equations are used
V1 =
nπ
[ cos ( nα 1 ) +cos ( nα 2 ) +cos ⁡(nα 3)] as given below:
(4)
Cos ( α 1 ) + Cos ( α 2 ¿ cos ( α 3 )=3 M

Cos ( 5 α 1 ) + Cos (5 α 2 ¿+cos (5 α 3 )=0

Cos ( 7 α 1 ) +Cos (7 α 2 ¿+cos (7 α 3 )=0

(7)

These are the non-linear transcendental equations for

elimination the 5th and the 7th order harmonics and get the
desired fundamental voltage. For the given values of m (from
0 to 1), it is required to get complete and all possible
solutions of these equations for determining the switching
angles and lower THD.

III. CLASSIFICATION OF MULTILEVEL INVERTER

CONTROL STRATEGIES

The key role of the modulation strategy used in MLI is to

synthesize the output voltage as close as possible to the
sinusoidal waveform. To obtain the sinusoidal waveform
from several levels of dc inputs, semiconductors devices
must be switched on and off in a manner that a stepped
output voltage waveform is obtained with fewer harmonics
and losses. To reduce the switching loss and harmonics,
various modulation techniques have been implemented and
are classified according to switching frequency, as shown in
fig-4.
Fig-3 Seven Level Cascade Inverter
The modulation index ‘M’ is defined as the ratio of the MLI Modulation
fundamental voltage (V1) to the maximum obtainable
Strategy
fundamental voltage (V1max) [12]. The maximum
fundamental voltage is obtained when all the switching
angles are zero and V1max is given as:

V1max = 3* ( 4 Vdc
π )
(5)
Fundamental
Switching
Frequency
High Switching
Frequency
PWM

Then the modulation index is given as:

Space Selective Space Sinusoidal SHE

Vector Harmonic Vector PWM PWM
Control Elimination PWM

Fig-4 Classification of Modulation Techniques
Low switching modulation techniques are used in high power
applications to keep the losses below the acceptable values.
During one complete cycle of the output voltage, these
methods perform one or two commutations of the
semiconductor devices, generating the staircase waveform.
Space vector and selective harmonic elimination are the
representatives of this family, thus resulting in better output
waveform and less complexity [9]. In these strategies, firing
angles are selected in such a way that the THD is minimum
in the output voltage waveform.
High switching frequency methods have many commutations
for the power semiconductor devices during one cycle of the
fundamental output voltage. A very popular method in the
industrial application is the carrier-based sinusoidal pulse
width modulation (SPWM) that uses the phase-shifting
technique to reduce the harmonics in the output voltage.
Another method is the space vector control strategy, which
has been used in three-level inverters.
IV. TECHNIQUE TO BE IMPLEMENTED
BAT ALGORITHM
The most powerful algorithms for optimization problems are
the nature-inspired algorithm and based on the echolocation
behavior of bats a new nature-inspired metaheuristic
algorithm called BAT algorithm has been proposed by Yang
2010 [7] [8] [9].
Bats are the only mammals in the world that have wings and
high echolocation capability. Practically, a sound signal is
radiated by bats called sonar/echolocation to detect the
objects surrounding them and find their way in night. The
sound emitted by bats is very loud and they can be also to
listen to the echo that bounces back from the surrounding
objects. Microbats also use short frequency modulated
signals in most of the cases and due to such echolocation
behavior, they can be formulated in such a way that it can be
associated with the objective function to be optimized and
thus can form a new optimization algorithm.
Developing bat algorithm needs some idealization of the
echolocation characteristics and for simplicity, we are
considering the following rules:
1. Echolocation is used by all the bats to sense the
distance and they also know the difference between
food and background barriers.
2. v i at position
Bats fly randomly with velocity
xi f
with a fixed frequency min , varying
wavelength and loudness A o to search for prey.
They can automatically adjust the wavelength (or
frequency) of their emitted pulse and adjust the rate
of pulse emission r in the range of [0, 1], depending
on the proximity of the target.
3. Although the loudness can vary in many ways, we
assume that loudness varies from a large (positive)
A o to minimum constant value A min .
In general the frequency f in a range [ f min , f max ]
corresponds to a range of wavelengths from minimum to
maximum value. Furthermore, we do not necessarily have to
use the wavelengths themselves; instead, we can also vary
the frequency while fixing the wavelength λ. This is because
λ and f are related due to the fact λf is constant. We will use
this later approach in our implementation. For simplicity, we
can assume f ∈ [0, f max ]. We know that higher
frequencies have short wavelengths and travel a shorter
distance. For bats, the typical ranges are a few meters. The
rate of pulse can simply be in the range of [0, 1] where 0
means no pulses at all, and 1 means the maximum rate of
pulse emission. Based on these approximations and
idealization, the basic steps of the Bat Algorithm (BA) can be
summarized. In algorithm we have define the rules for the
updating the positions z i and velocities v i in the
search space. The new solutions i.e. position and velocity is
given by:
f i = f min + ( f max−f min ) α
(8)
v i = v i + ( z i−z o ) f i
(9)
z ti = f i + v i
(10)
Z new = Z old + A t µ
(11)
A. THE PSEUDO CODE OF THE BAT ALGORITHM IS AS
FOLLOWS:
The main steps of proposed BA are as follows-
Step 1: Initialize the bat population or their position zi
and their velocities v i . Define pulse frequency fi at
z i . Initialize pulse rate r and loudness A.
Step 2: Generate new solutions by adjusting frequency, and
updating velocities and locations (equation 7 to equation 10).

Step 3: if (rand > r)

Select a solution among the best solutions. Generate a local
solution around the selected best solution.

Step 4: Else generate a new solution by flying randomly.

Step 5: If (rand < A && f ( z i ) < f ( z 0 )

Where f (.) = Objective function
Accept the new solutions, increase r and reduce A. Fig 5. FFT Analysis of line to line voltage

Step 6: Rank the bats and find the current best z0 .

Steps 7: while (iteration < Max number of iterations)
Post process results and visualization. The algorithm stops
with the total-best solution.

V. RESULTS AND DISCUSSIONS

In [main ref paper] the PSO algorithm is successfully applied

to calculate the optimum switching angles in multilevel
inverters. However, by increasing the number of switching
angle. The algorithm traps in the local minima of the search
space.

The algorithm described in this paper is simulated in

MATLAB/SIMULINK software for an 11-level Cascade H-
Bridge inverter. The voltage dc link for each H-bridge units is
considered to be 100V and simulation is done for k=3 and
s=5 and M=1.
The FFT analysis for line-line voltage and values of
switching angles is shown in fig 5 and 6.
Fig 6. Number of switching angles Vs value of switching
angles
The output results obtained from bat algorithm is compared
with the results in paper [main ref paper]. The parameters
used in the implementation of BAT algorithm is given in
Table 1.

The comparison graph for the BAT and MPSO is shown in

Fig 7.
 θ1 3 2.46 8.35
θ1 4 39.62 33.99
θ1 5 31.69 45.07

Table 3. Comparison of percentage of THD

THD (%) BAT MPSO

2.91 3.30

CONCLUSION

In this paper it has been concluded that employing BAT for

the minimization of harmonics can achieve great heights. As
Fig 7. Comparison of Switching Angles modifies species based particle swarm optimization has given
the best results as compared to other optimization techniques
Table 1. Parameters used in BAT algorithm but by employing BAT algorithm all the aspects of MPSO
can be covered as well as harmonic calculation at location to
PARAMETERS VALUES location could be done at lower level. All feasible solutions
Qmin 0 are closed to average values and hence algorithm is robust
Qmax 2 and has fast convergence as compare to other methods.
Population 10
Maximum number of 500 REFERENCES
iterations
[1] B. Ozpineci, L. M. Tolbert and J. N. Chiasson, “Harmonic
Optimization of Multilevel Converters Using Genetic
The values of optimum switching angles obtained by bat Algorithms,” IEEE Power Electronics Letters, Vol. 3, No. 3,
algorithm given in Table 2. The value of THD obtained is 2005, pp. 92-95.
also given in Table 3 and are compares with the output give
in []. [2] M. Mythili, N. Kayalvizhi, “Harmonic Minimization in
multilevel inverters using selective harmonic elimination
Table 2. Comparison of switching angles PWM technique” International conference on ICRSE, 2013,
pp. 70-74.
SWITCHING BAT MPSO
ANGLES [3] Zhong Du, Leon M. Tolbert, and John N. Chiasson,
θ1 10.12 66.62 “Active harmonic elimination for multilevel converters”
IEEE Transactions on Power Electronics, vol. 21, no. 2,
θ2 2.77 7.16 March 2006, pp.459-469.
θ3 19.57 2.90
θ4 42.09 47.51 [4] Bhim Singh, Kamal Al-Haddad and A. Chandra, "A new
control approach to three phase Harmonics and reactive
θ5 26.90 19.13 power compensation", IEEE Trans.On power systems Vol.
θ6 15.02 24.18 13, No 1 Feb 1998, pp. 133-138.
θ7 66.79 49.54
[5] A. Salami and B. Bayat, "Total Harmonic Distortion
θ8 49.63 21.96 Minimization of Multilevel Converters Using Genetic
θ9 23.93 29.99 Algorithms," Applied Mathematics, Vol. 4 No. 7, 2013, pp.
1023-1027.
θ1 0 73.60 72.60
θ1 1 57.17 53.76 [6] Mariusz Malinowski, K. Gopakumar, Jose Rodriguez, and
θ1 2 30.79 10.24 Marcelo A. Perez, “A survey on cascaded multilevel
inverters” IEEE Transactions on Industrial Electronics, vol.
57, no. 7, July 2010, pp.2197-2206.
[7] Jose Rodriguez, Jih-Sheng Lai, and Fang Zheng Peng,
“Multilevel inverters: A survey of Topologies, controls, and
applications” IEEE Transactions on Industrial Electronics,
vol. 49, no. 4, August 2002, pp.724- 738.
[8] R. H. Baker and L. H. Bannister, “Electric Power
Converter,” U.S. Patent 3 867 643, Feb. 1975.
[9] G. Carrara, S. Gardella, M. Marchesoni, R. Salutari and
G. Sciutto, “A New Multilevel PWM Method: A Theoretical
Analysis”, IEEE Transactions on Power Electronics, Vol. 7,
NO. 3, July 1992, pp.497-505.
[10] J.N. Chaisson, L.M. Tolbert, K.J. Mckenzine and Zhong
Du, “Elimination of harmonics in a multilevel converter
using the theory of symmetric polynomials and resultants,”
IEEE transactions on Control Systems Technology, vol. 13,
no. 2, March 2005.
[11] Jagdish Kumar, “THD analysis for different levels of
cascade multilevel inverters for industrial applications,”
International Journal of Emerging Technology and Advanced
Engineering, vol. 2, issue 10, Oct 2012.
[12] B. Ozpineci, L.M. Tolbert and J.N. Chaisson, “Harmonic
optimization of multilevel converters using genetic
algorithms,” 35th Annual IEEE Power Electronics Specialists
Conference, Germany 2004.
[15] A. Kavousi, B. Vahidi, R.Salehi et al, “Application of the
bee algorithm for selective harmonic elimination strategy in
multilevel inverters,” IEEE Transactions on Power
Electronics, vol. 27, no. 4, April 2012, pp. 1689-1696.
[16] H. Taghizadeh and M.T. Hagh, “Harmonic elimination
of cascade multilevel inverters with non equal dc sources
using particle swarm optimization,” IEEE Transactions on
Industrial Electronics, vol. 57, issue 11, Nov 2010, pp. 3678-
3684.
[17] M.T. Hagh, H. Taghizadeh and K. Razi, “Harmonic
minimization in multilevel inverters using modified species
based particle swarm optimization,” IEEE Transactions on
Power Electronics, vol. 24, issue 10, Oct 2009, pp. 2259-
2267.
[18] Shah-Hosseini. H, “Optimization with the Nature-
Inspired Intelligent Water Drops Algorithm”, Int. Journal of
Intelligent Computing and Cybernetics, Vol. 1, No. 2, pp.
193-212, 2008.
[19] M. Tarafdar Hagh, H. Taghizadeh and K. Razi,
"Harmonic Minimization in Multilevel Inverters Using
Modified Species-Based Particle Swarm Optimization,"
in IEEE Transactions on Power Electronics, vol. 24, no. 10,
2009, pp. 2259-2267.

[13] Jagdish Kumar, Jaimala Gambhir and Anil Kumar,

“Control of switching angles for a cascade multilevel inverter
using ANN,” IEEE conference RAECS-2014 held at UIET,
Chandigarh, India, 6th-8th March, 2014, pp. 1-6.

[14] Suman Debnath, R.N. Ray and T. Ghosh, “Comparison

of different soft techniques applicable to multilevel inverter
for harmonic elimination,” International Journal of Computer
Application, issue 2, vol. 6, Dec 2012.