Vol. 5(1), pp.

9-22, May 2013

DOI 10.5897/JEEER2013.0436
ISSN 2141 2367 2013 Academic Journals
http://www.academicjournals.org/JEEER

Journal of Electrical and Electronics

Engineering Research

Review

Fundamentals and literature review of Fourier

transform in power quality issues
Ltf SARIBULUT1*, Ahmet TEKE2 and Mehmet TMAY2
1

Department of Electrical-Electronics Engineering, Adana Science and Technology University, Adana/TURKEY.

2
Department of Electrical-Electronics Engineering, ukurova University, Adana/TURKEY.
Accepted 22 May, 2013

Economic impact of power quality (PQ) disturbances on industry is increasing day by day with
widespread use of power electronics based devices. These disturbances should be detected quickly
and accurately. Fourier transform (FT) has become a powerful tool to detect, classify and analyze
various types of PQ disturbances in power systems and signal processing. In this paper, the
fundamentals of Fourier series and FT with their types and distinctions over other disturbance
detection methods are clearly discussed. An extensive literature review of more than 50 research
papers on FT is also presented.
Key words: Power quality disturbances, literature review, harmonics, Fourier transform.
INTRODUCTION
Overview of Fourier series and Fourier transform

Fourier expansion corresponding to f(x) is given in the

following.

Fourier series
Fourier analysis is named after Jean Baptiste Joseph
Fourier (1768 to 1830), a French mathematician and
physicist. Joseph Fourier, while studying the propagation
of heat in the early 1800's, introduced the idea of a
harmonic series that can describe any periodic motion
regardless of its complexity. Later, the spelling of Fourier
analysis gave place to Fourier transform (FT) and many
methods derived from FT are proposed by researchers.
In general, FT is a mathematical process that relates the
measured signal to its frequency content Heideman et al.
(1985). The Fourier series is described in theory and
problems of advanced calculus as follows:
Let, f(x) is the periodic function. f(x) can be defined in an
interval (-T, T) and outside this interval by f(x+2T)=f(x),
that is, assume that f(x) has 2T. The Fourier series or
*Corresponding author. E-mail: [email protected].

f ( x) =

a0
n x
n x
+ an cos
+ bn s in

2 n =1
T
T

(1)

where an and bn are the coefficients of Fourier series, a0

is the first term of an for n=0.

an =

1 T
n x
f ( x ) cos
dx,

T
T
T

0n

(2)

bn =

1 T
n x
f ( x ) sin
dx,

T
T
T

1 n

(3)

A simple example for Fourier series is given in the

following.

10

J. Electrical Electron. Eng. Res.

f(t)

-8

-6

-4

-2

(s)

Figure 1. The waveform of a periodic triangular signal.

Example 1: Calculate the Fourier series of a triangular

periodic signal given in Figure 1. The equation of signal is
given as f(t)=t in the interval -2 t < 2.
Solution 1: The period (T) of triangular periodic signal is
calculated as 4 (T=4), then = (2 / 4) = 1/ 2 is
obtained. The function f(t) is an odd function by
considering the situations f(-t)=(-t) and -f(t)=(-t). Because
of this situation, the coefficients of an are calculated as
zero. The coefficients of bn can be calculated as:
2

1
n t
2
t sin
dt =

20
2
n

The coefficient values of bn are calculated for the first five

terms of positive integers given in the following.

b1 =

4
b2 =
,
2

4
b3 =
,
3

4
b4 =
,
4

4
b5 =
5

The last statement of f(t) is written by using Fourier

explanation in the series form and it is given in the
following.

f (t) = bn sin nt = b1 sint +b2 sin2t +b3 sin3t +b4 sin4t +b5 sin5t +...
n=1

41
1
1
1
1
= ( sint sin2t + sin3t sin4t + sin5t +...)
1
2
3
4
5
4 (1)n+1
=
sin nt
n=1 n

The Fourier transform is the one of the several

mathematical tools for analyzing the signals. It involves
the decomposition of the signals in the frequency-domain
in terms of sinusoidal or cosinusoidal components. The
mathematical definition of a continuous FT (CFT) is given
in the following.

X ( ) =

1
2

x ( t ) e jt dt

(4)

n t 2 n t

t cos 2 + n sin 2
0
0
2
4
=
(2 cos n 0) =
cos n
n
n

bn = 2

Fourier transform

where x(t) is the original signal, X() is the representation

of signal in the frequency-domain, j is the imaginary
number, is the angular frequency and t is the time
index.
The inverse of the continuous Fourier transform (ICFT)
is defined as;

x (t ) =

1
2

X ( ) e j t d t

(5)

A simple example for Fourier transform is given in the

following.
Example 2: Calculate the continuous Fourier transform
of a single rectangular pulse given in Figure 2. The

1,

<t<
2
2
0, elsewhere

equation of signal is given as x(t ) =

and it is calculated by regarding the width of seconds.

Solution 2: The Fourier transform (X(w)) of x(t) is

Saribulut

11

F(t)

1.0
0.8
0.6
0.4
0.2
0.0

0.0

(s)

Figure 2. A single rectangular pulse.

F(f)

(Hz)

0
Figure 3. Fourier transform of a rectangular pulse.

DISCRETE FOURIER TRANSFORM

calculated as:
/2

X () =

jt

/ 2

/2
1
1
dt = e jt = ( e j /2 e j /2 )
/2
j
j

sin ( /2)
2
2
= sin ( / 2) = sin ( / 2) =

/2
The function

sin ( / 2 )
/ 2

is named as sinc function

by the mathematicians. In addition, the mathematical

equation of sinc function can be rewritten as

X ( ) = sinc ( / 2 ) . Then, Fourier transform of

the rectangular pulse can be given in Figure 3.

The continuous-time periodic signals and finite-energy

non-periodic signals can be defined with Fourier series
representation. In addition, the discrete-time signals can
be represented within finite duration in practice. An
alternative transformation is called Discrete Fourier
Transform (DFT) for a finite-length signal, which is
discretized in frequency.
The frequency range of discrete-time signals is defined
over the interval between to . A periodic digital
signal consisted of N samples is to separate the
frequency components into 2/N radians intervals by
dividing the frequency-domain. Then, Fourier series
representation of the discrete-time signal will consist of N
frequency components Kuo et al. (2001). The general
Fourier series representation of a periodic signal (x(n)) is
expressed as:

12

J. Electrical Electron. Eng. Res.

N -1

x(n)= ck e

jk(2 /N)n

(6)

multiplications. These multiplications can be given in the

matrix form.

k=0

where N is the harmonic index related with the

jk(2n/N)n
exponentials function (e
) for k=0, 1,,N-1, ck is the
coefficients of the Fourier series.
The coefficients ck are calculated as:

1
ck =
N

N -1

x(n)e

jk(2 /N)n

(7)

W40
0
W4
W 0
4
W40

kn

The coefficients ck of Fourier series are the form of a

periodic sequence of fundamental period N. The timedomain spectrum of a periodic signal can be represented
as periodic sequence with DFT. The frequency analysis
of discrete-time periodic signals (sin(nt) and cos(nt))
involves Fourier transform of the time-domain signal. DFT
is defined as multiplication of N samples x(n) with N
discrete frequencies. These samples are taken at
discrete frequencies (k=2k/N), where k=0,1...N-1,
between 02. This means that X() is evaluated at
the successive samples by equally spaced frequencies.
X() is given in the following.

x ( n) e

j (2 / N ) kn

(8)

n=

DFT is a mapping between N samples x(n) of the timedomain into N samples X() of the frequency-domain.
This gives the opportunity to compute DFT of the periodic
and the finite-length signals. The frequency-domain
spectrum of a periodic sequence can be re-obtained as
the periodic signal by using inverse discrete-time Fourier
transform (IDFT). IDFT can be defined by using the
frequency samples of X(k). It is given in the following.

1 N -1
x (n) = X(k)WN kn , n =0, 1, ..., N -1
N k=0

W40 W40 W40 x (0)

W41 W42 W43 x (1)

W42 W44 W46 x (2)

W43 W46 W49 x (3)

The mathematical calculations of four-point DFT can be

decreased
by
using
the
periodicity
property
k ( n+ N )

( WN = WN

n=0

X( k ) =

X (0)
X (1)

=
X (2)

X (3)

kn + N / 2
=
N

(W

(k + N )n

= WN

kn
N )

and

symmetry

of DFT. With the periodicity property

( W = W = 1,

W41 = W49 = j , W42 = W46 = 1 and

0
4

4
4

W43 = j ), the matrix form of four-point DFT can be

rewritten given in the following.

1 1 1 x (0)
X (0) 1
X (1) 1 j 1 j x (1)

X (2) 1 1
1 1 x (2)

j 1 j x (3)
X (3) 1
After the using of periodicity property, the equations given
above are also rewritten by using the symmetry property
and the last equations can be given in the following form.
X (0) = x(0) + x(1) + x(2) + x(3) = [ x(0) + x(2) ] + [ x(1) + x(3) ]
14
4244
3 14243
g1

IDFT shows that there is no loss information by

transforming the frequency spectrum of X(k) back into the
original time sequence of x(n). A simple example for DFT
is given in the following.
Example 3: Calculate an efficient algorithm for four
samples DFT given in the following form.
3

X (k ) = x(n)W , k = 0,1, 2, 3
kn
N

n=0

Solution 3: The four-point DFT includes 16 complex

g2

X (1) = x(0) jx(1) + x(2) + jx (3) = [ x(0) x(2) ] j [ x (1) x (3) ]

14243
14243
h1

h2

X (2) = x(0) x(1) + x(2) x(3) = [ x (0) + x(2) ] [ x(1) + x(3) ]

14
4244
3 14243
g1

g2

X (3) = x(0) + jx(1) x(2) jx(3) = [ x (0) x(2) ] + j [ x(1) x(3) ]

14243
14243
h1

(9)

property

h2

Hence, the equations of new algorithm become as

follows:

Stage 1

Stage 2

g1 = x (0) + x (2)

X (0) = g1 + g 2

g 2 = x (1) + x (3)

X (1) = h1 jh2

h1 = x (0) x (2)

X (0) = g1 g 2

h2 = x (1) x (3)

X (1) = h1 + jh2

The new algorithm includes only the two complex

multiplications. The simple structure of this algorithm is
illustrated as graphically in Figure 4.

Saribulut

x(0)

x(2)

h1
-j

x(1)

x(3)

X(0)

g1

-1

-1

13

g2

-1

h2

X(1)

X(2)

X(3)

Figure 4. DFT algorithm for four samples.

Fast Fourier transform

Fast Fourier transform (FFT) is an efficient algorithm to
compute the discrete Fourier transform and its inverse. A
DFT decomposes a sequence of values to the
components of different frequencies. This operation is
useful in many fields but computing it directly from the
definition is often too slow to be practical. In other words,
DFT is a native way to compute the Fourier transform of
N samples by using the general definition. FFT is a fast
way to compute the same result with DFT. DFT takes N2
complex multiplications and N2-N complex additions are
required operations, while FFT makes it possible to
calculate DFT with Nlog2N operations instead of N2
operations (Wikipedia, 2013).
The most common FFT algorithm is the CooleyTukey
algorithm (Cooley and Tukey, 1965). This algorithm
divides the transform into two pieces of size N/2 at each
step. Therefore, it is limited to power-of-two sizes. These
are called radix-2 and mixed-radix cases, respectively. In
other words, CooleyTukey algorithm breaks DFT into
smaller DFTs. It can be combined arbitrarily with any
other algorithm for the DFT.
The two FFT algorithms are proposed.
(i) Decimation in time (DIT)
(ii) Decimation in frequency (DIF)
In DIT and DIF algorithms, the input time sequence is
successively divided into smaller sequences and DFTs of
these subsequences are combined in a certain pattern to
yield the required DFT of the entire sequence with fewer
operations. Since this algorithm was derived by

separating the time-domain and frequency-domain

sequences successively into the smaller sets, the
resulting algorithm is referred to as decimation in time
algorithm (Kuo et al., 2001).
i. Radix-2 Decimation in Time
Firstly, DIT computes Fourier transform of even-indexed
numbers x(0), x(2), ...,x(n-2) and odd-indexed numbers
x(1), x(3),...,x(n-1). The sets of even and odd sequences
can be defined as:

x1 (m) = x(2m), m = 0,1,..., ( N / 2) 1

(11)

x2 (m) = x(2m + 1), m = 0,1,..., ( N / 2) 1

(12)

These two results are combined to produce the Fourier

transform of the whole sequence. Then, this idea can be
performed recursively to reduce the overall runtime. This
simplified form assumes that N is a power of two, since
the number of N samples can be usually chosen freely by
the application. This is often not an important restriction
(Sundararajan, 2001). The name "butterfly" comes from
the shape of the data-flow diagram in the radix-2 case. A
basic butterfly operation of DIT algorithm required only
twiddle-factor multiplies per stage is given in Figure 5.
Each butterfly involves just a single complex
k

multiplication by a twiddle factor WN . The twiddle factor

converts N points DFT into an N/2-points DFTs. In other
words, N points DFT sequence are divided into N/2 point
subsequences, X(2k), X(2k+1) where k = 0,1,...,N/2-1.

14

J. Electrical Electron. Eng. Res.

X(A)=x(a)+x(b)W
WNNk

x(a)

WNk

x(b)

-1

X(B)=x(a)-x(b)W
WNNkk

Figure 5. Twiddle-factor of 2-point DIT algorithm of FFT.

ii. Radix-2 decimation in frequency

A second variant of the radix-2 FFT is the decimation-infrequency algorithm. In order to derive this algorithm, the
input sequence is divided into the first and second halves
of N/2 samples. The process of decomposition is
continued until the last stage is made up of two-point
DFTs. The decomposition proceeds from left to right for
the decimation-in-frequency algorithm and the symmetry
relationships are reversed from the decimation-in-time
algorithm Mertins (1996). A basic butterfly operation of
DIF algorithm is given in Figure 6.
A simple example for DIF algorithm of FFT is given in
the following.
Example 4: Calculate the fast Fourier transform of
discrete signal x(n) by regarding the decimation in time
algorithm. It is given in the following.

x(n) = {1,3, 0, 2, 4,1, 0, 2}

Solution 4: The sequence of x(n) is divided into
subsequences included the odd (o) and even (e) index
terms until at least a group had two terms left. The
division of signal is given in the following.
st

1 division: xo ( n) = {1, 0, 4, 0}

xe (n) = {3, 2,1, 2}

2nd division: xoo (n) ={1,4} xoe (n) ={0,0} xeo (n) ={3,1} xee (n) ={2,2}
In the first step, 2-Radix FFT of subsequences is taken
by using the definition of FFT.
2-1

j(2 )kn
N

For, X(k) = x(n)WNkn = x(n)e

n=0
1

j (2 )0n
8

Xoo (0) =xoo(n)e

n=0

=xoo(0) + xoo (1) =1+4=5

n=0

k =0,1,..., N 1

j(2 )1n
8

Xoo (1) = xoe(n)e

n=0

j(2 )0n
8

=xoo(0)e

j(2 )1n
8

+ xoo(1)e

=1+4(10 j) =3, then

Xoo (k) =[5 3]; Xoe (k) =[0 0]

Xeo(k) =[4 2]; Xee (k) =[4 0]
The second step is the 4-Radix FFT of other
subsequences calculated by using DIT algorithm. The
equations of 4-Radix FFT are used to calculate the odd
terms of X(k) and the results are given in the following.

foroddterms, then
Xo (k + N ) = Xoo (k) W4k Xoe (k), k = 0,1
2

Xo (k) = Xoo (k) +W4k Xoe (k), k = 0,1

Xo (k) = [5 3 5 3]
The equations of 4-Radix FFT are used to calculate the
even terms of X(k) and the results are given in the
following.

foreven terms, then

Xe (k + N ) = Xeo (k) W4k Xee (k), k = 0,1
2

Xe (k) = Xeo (k) +W4k Xee (k), k = 0,1

Xe (k) = [8 2 0 2]
The third step is the calculation of 8-Radix FFT for other
subsequences. The equations of 8-Radix FFT are used
to calculate FFT by utilizing the odd and even terms of
X(k). The results of 8-Radix FFT are given in the
following.

Saribulut

15

X(A)=x(a)+x(b)

x(a)

WNk

-1

x(b)

WNNk[x(a)-x(b)]
X(B)=W
k

Figure 6. Twiddle-factor of 2-point DIF algorithm of FFT.

For, X (k ) = X o ( k ) + W4k X e ( k ), k = 0,1, 2,3 and

X ( k + N ) = X o ( k ) W4k X e ( k ), k = 0,1, 2,3, then
2
X ( k ) = [13 ( 1.5858 j1.412) 5 (4.4142 j1.4142)

3 ( 4.4142 j1.4142) 5 (1.5858 j1.412)]

Inverse fast Fourier transform
The inverse FFT (IFFT) is calculated by using the
equations based on FFT with simple operations (Kuo et
al., 2001; Thede, 2004). Basically, FFT algorithm is used
to compute IFFT of a sequence X(k). In the first step, the
complex conjugating (X(k)*) of X(k) is used in FFT
algorithm and then the results are divided by N (the
length of FFT) to obtain x*(k). In the second step, FFT
algorithm is also used to compute the inverse DFT by
utilizing the complex conjugating of x*(k) for obtaining the
sequence of x(n). If the signal is real-valued, the final
conjugation operation is not required. The mathematical
expression of IFFT is given in the following.

x* ( n ) =

1
N

N -1

(k)WNkn , n=0, 1, ..., N -1

(13)

k=0

All steps of FFT algorithms are based on two-input and

two-output butterfly computations. These steps are
classified as radix-2 complex FFT algorithms. Different
radix butterflies can be combined with radix FFT
algorithms.
Advantages and disadvantages of FFT
The advantages and disadvantages
summarized as given in the following.

of

FFT

are

(i) FFT is the easiest way to calculate the Fourier

transform.
(ii) The fundamental harmonic and its multiply of whole
number harmonics (except inter harmonics) are extracted

easily by FFT.
(iii) FFT algorithm is insufficient to extract the interharmonics and variable-frequency harmonics of a signal.
New FFT algorithms should be improved to calculate the
inter-harmonics and variable-frequency harmonics.
(iv) To prevent the aliasing, the anti-aliasing filters are
used before the sampling of the signal. It is resulted with
the delaying at the response time of FFT.
(v) FFT can be calculated the harmonics after the one
period of the signal left. Consequently, harmonics cannot
be extracted simultaneous with reference signal.
(vi) To calculate the higher harmonics, the signal should
be sampled at higher frequency.
LITERATURE REVIEW
Methods dependent on DFT and FFT
The references given in the following are related with the
DFT and FFT based control algorithms in power quality
issues.
A method for the calculation of bus voltage transients in
an electric power system is presented (Heydt, 1989). The
essence of the method is the Fourier transform of Ohms
law. The fast Fourier transform is used in order to give
computational efficiency. Two approximations are found
for the calculation of this transfer impedance and one of
these is found to be applicable to the cited problem.
Examples are used to illustrate the calculation of bus
voltage transients and harmonic content. The Gibbs
phenomenon appears in a discrete Fourier transform due
to incomplete periodic, the waveform has not reached a
full cycle within its period. Data flipping furnishes a
complete periodic cycle to the waveform and thus
suppresses the Gibbs phenomenon. This facilitates the
design of a digital filter using fast Fourier transform
without windowing. The filter does low-pass, band-pass,
high-pass, band-stop, notch or single-frequency-pass
simply by manipulating the band limits. The filter can be
affected by spectral resolution and the slope discontinuity
at the end data points.
The reduction of such effects and an alternative design

16

J. Electrical Electron. Eng. Res.

are discussed (Pan, 1993). The evaluation of DFT

spectra usually yields more detailed information than
evaluation in the time domain is presented (Breitenbach,
1999). The evaluation of time-discrete spectra, hampered
by leakage, which occurs if a non-integral number of
periods, is present in the sampled data set. Minimization
of spectral leakage is an important prerequisite for
spectrum analysis. The use of non-rectangular timedomain windowing functions offers some improvement
but also invites unwanted side-effects. Spectral leakage
can be avoided entirely by ensuring that an integer
number of periods falls into the sampling time (e.g. by the
use of coherent sampling). The application of the
windowed fast Fourier transform to electric power quality
assessment is presented (Heydt et al., 1999). The
windowed FFT is a time windowed version of the discrete
time Fourier transform. The window width may be
adjusted and shifted to scan through large volumes of
power quality data. Narrow window widths are used for
detailed analyses and wide window widths are used to
move rapidly across archived power quality data
measurements. The mathematics of the method is
discussed and applications are illustrated.
A method for power signal harmonic analysis is
proposed for the frequency and phasor estimating
algorithm (Yang et al., 2005). The major components of
the method are a frequency and phasor estimating
algorithm, a finite-impulse-response comb filter and a
correction factor. It also combines other methods to
enhance our performance, such as discrete Fourier
transform and least square error method. To verify
proposed method, it compares FFT. An advantages and
practical achievements of the frequency-domain
qualification of dynamic properties based on the step
response data obtained both experimentally and
analytically are pointed out (Matyas et al., 2005).
Measurement errors of the impulse peak value and time
parameters have been evaluated via FFT, convolution
and IFFT for a commercial divider. Direct determination
of the high voltage impulse measurement range, no need
for any additional parameters and illustrative graphical
presentation of results are the main advantages of this
approach.
Limitations
of
different
qualification
approaches are discussed.
The problem of differentiating inrush currents from fault
currents that are observed for a feeder at a distribution
substation is discussed Baran et al. 2006. This paper
shows that the two approaches the Fourier transform and
the Wavelet transform can be adopted to extract features
that make it possible to distinguish them from each other
by using an artificial-neural-network-based classifier. The
paper also illustrates how to address the issues for
successful implementation of these schemes; such as
prescreening of data, how to apply FFT and wavelet
transform on the data and the training of the artificial
neural network in order to maximize the performance of
the classifier. These issues are illustrated using the

actual field data. An approach to simplify the design of

IFFT/FFT cores for OFDM applications is presented
(Cortes et al., 2006). A novel software tool is proposed,
called AFORE. It is able to generate efficient single and
multiple mode IFFT/FFT processors. AFORE employs a
parallel architecture, where the degree of parallelism can
be varied. In order to assess the quality of the proposed
approach, results are provided for some of the most
widely used OFDM standards, such as, WLAN 802.11
a/g, WMAN 802.16a, DVB-T.
The sliding DFT algorithm as an alternative for typical
DFT used for spectrum analysis and synthesis is
proposed Heriakian (2006). As an example, a control
circuit for a three-phase 75kVA parallel Active Power
Filter (APF) is used. The presented control APF algorithm
allows selection of control parameters in mains currents:
imbalance, reactive power or harmonics contents. In the
proposed circuit transient performance of APF is
improved
using
non-causal
predictive
current
compensation. An improved FFT-based algorithm to
measure harmonics and interharmonics accurately is
proposed (Qian et al., 2007). In the proposed algorithm, a
frequency-domain interpolation approach is adopted to
determine the system fundamental frequency and the
inter-polatory polynomial method is applied to reconstruct
the sampled time-domain signal; it is followed by using
the FFT to calculate the actual harmonic components.
The performance of the proposed algorithm is validated
by testing the actual measured waveforms. Results are
compared with those obtained by directly applying a
typical FFT algorithm and by the International
Electrotechnical Commission (IEC) grouping method. The
uncertainty analysis of the Root-Mean-Square (RMS)
value and phase computed from the DFT spectrum of the
non-coherently sampled signal using cosine windows are
investigated (Novotny et al., 2007). The analysis is
focused on investigating the influence of quantization
noise.
A Direct Current-Space-Vector Control Scheme is
presented for a three-level, neutral-point-clamped voltage
source inverter, which is employed as an active power
filter (Vodyakho et al., 2008). The proposed method
generates the compensation current reference indirectly
generating an equivalent ohmic conductance for the
fundamental component by means of the APFs dc-link
voltage control. Based on the fast Fourier transform the
compensation of the reactive fundamental current and
selectable harmonics can be cancelled, confining the
operation to only harmonic compensation and thus
saving the APFs apparent power. The sliding discrete
Fourier transform is proposed (Sumathi et al., 2008). It
splits periodic signals into selected harmonic
components, as on-line time functions. Ordinarily, the
sampling frequency is equal to the product of the nominal
signal frequency and the window width. However, when
the signal frequency drifts, to avoid the phase and
magnitude errors, the sampling frequency can be

Saribulut

adaptively adjusted using the phase-error itself. An

integrated phase-locked loop scheme and its parameters
like hold-in, pull-in ranges, lock time, steady-state errors
are presented in this brief.
A strategy of group-harmonic weighting distribution is
proposed for system-wide inter-harmonic evaluation in
power systems (Lin, 2008). The proposed algorithm can
restore the dispersing spectral leakage energy caused by
the FFT and calculate the power distribution proportion
around the adjacent frequencies at each harmonic to
determine the inter-harmonic frequency. The numerical
examples are presented to verify the performance of the
proposed algorithm.
A precise and adaptive algorithm for interharmonics
detection based on iterative DFT is proposed (Zhang et
al., 2008). Practical formulas to calculate the parameters
of harmonics and interharmonics in electric power
systems are presented. An algorithm for fundamental
frequency and harmonic components detection is
presented (Lavopa et al., 2009). The technique is based
on a real-time implementation of discrete Fourier
transform and it allows fast and accurate estimation of
fundamental frequency and harmonics of a distorted
signal with variable fundamental frequency. The
proposed is compared with a Phase Locked Loop
algorithm for frequency detection and the synchronous
DQ reference method for harmonic detection. A spectral
correction-based
algorithm
for
inter-harmonic
computation is proposed (Salor, 2009) for especially
highly fluctuating fundamental frequency cases in the
power system. It has been observed and reported that
fluctuating demands of some loads such as arc furnaces
or disturbances and subsequent system transients make
the fundamental frequency of the power system deviate
and this causes non-existing interharmonics to appear in
the spectrum due to grid-effect when a standard window
length is used for the entire FFT process. The proposed
method uses a synthetic waveform produced at the
fundamental frequency and amplitude to determine the
amount of the leakage due to the grid-effect at each
frequency. Then the leakage is subtracted from the
original FFT of the signal to correct the frequency
spectrum. Both simulative and field data tests have been
performed.
The technique of pre-calculation process for real-time
FFT is presented by Yena et al. (2009). The real-time
FFT algorithm simultaneously constructs and computes
the butterfly modules while the incoming data is collected.
The depending on the computing capability of the
processor, different number of pre-calculation stages for
better performance is also suggested. An algorithm for
the estimation of the frequency of single-tone signals is
presented by Radil et al. (2009). The algorithm works in
the frequency domain and is based on best fitting a
theoretical spectrum of a single-tone signal that is
windowed using a rectangular window on the spectrum of
the sampled signal. The influence of noise and harmonic

17

and inter-harmonic distortions on the proposed algorithm

was investigated and is reported. The algorithms
performance was compared with several other frequencyestimation algorithms (mostly those working in the
frequency domain). A space vector (SV) current control
scheme for shunt APFs is introduced with a three-level
neutral point-clamped voltage source inverter (VSI) as
well as a standard two-level VSI (Vodyakho et al., 2009).
The proposed control can selectively choose harmonic
current components by using a real-time fast Fourier
transform to generate the compensation current. The
proposed current control utilizes a rotating coordinate
system, processing the information of the actual position
of the grid-voltage SV, which is remarkably important in
APF applications and chooses switching states from the
switching table implemented in a field programmable gate
array.
The transient performance and stability of a recurrent
DFT based control methods for a series active filter
integrated with a 12-pulse diode rectifier are considered
by Le Roux et al. (2009). The control method targets
specific harmonics and/or the negative sequence
fundamental component of the supply current and is
intended for use with non-sinusoidal/unbalanced supply
voltages. The proposed control method is based on DFTs
instead of the DQ-method and a simple approach is used
to account for small frequency variations found in
practical power systems. The three-phase four-wire APF
based on a three-level neutral-point-clamped inverter is
presented by Vodyakho and Kim (2009). To regulate and
balance the split DC-capacitor voltages, a control using
the sign cubical hysteresis controller is proposed. The
control method discerns the harmonic currents by fast
Fourier transform choosing switching states from a
switching table based on the hysteresis control. An
analysis for simultaneous sensing of multiple primary
user activity in cognitive radios has been presented from
a signal-processing perspective by Sheikh et al. (2009). It
has been shown that a poly-phase DFT filter bank with an
optimal window as prototype filter can lead to reliable as
well as efficient sensing and detection architecture in
cognitive radios with minimal overhead. A detailed
account of challenges, performance limits and
computational complexity comparisons for various
spectrum-sensing solutions is also provided.
An analysis of the natural voltage balancing dynamics
of a three-phase flying capacitor converter is presented
when supplying an induction motor (McGrath et al.,
2009). The approach substitutes double Fourier harmonic
series for the pulse-width modulation switching
waveforms and the frequency response of the motor, to
create a linear state-space model of this type of load.
Model parameters were measured by applying FFT
analysis to variable frequency square waves injected into
the motor terminals and fitting parameter values to these
measurements using a least squares minimization
method. Experimental verification results using a scaled-

18

J. Electrical Electron. Eng. Res.

down flying capacitor converter drive are included. A

partial cached FFT processor that accounts for the
distribution of allocated resources to the users of the
OFDMA system is designed by Chen et al. (2010). The
constellation- and power-aware twiddle-factor multiplier
for the variable FFT length and modulation order are
designed. A 128 to 1024-point mixed pipelined/cachedFFT processor using a 0.18-m 1P6M CMOS technology
is implemented. The chip measurement results show that
its energy dissipation ranges from 0.09 to 1.90nJ per FFT
point and scales to the allocated resources in the
OFDMA system.
A space-vector discrete-time Fourier transform for fast
and precise detection of the fundamental-frequency and
harmonic positive- and negative sequence vector
components of three-phase input signals is proposed by
De Souza et al. (2010). A recursive algorithm for loweffort online implementation is also presented. The
detection performance for variable-frequency and interharmonic input signals is discussed. A closed-form
analytical approximation of the output harmonic spectrum
of a single-phase two-level inverter under the action of
hysteresis current control is derived by Albanna et al.
(2010). The analytical approach consists of first
describing the error current as a triangular signal of
variable duty cycle and frequency and subsequently,
deriving the Fourier transform of the complex envelope of
the modulated triangular signal. The spectrum of the error
current is given in terms of Bessel functions of the first
kind. An estimating the amplitude of harmonic
components of a harmonically distorted sine wave by the
Interpolated Discrete Fourier Transform (IPDFT) method
with maximum side lobe decay windows is proposed by
Belega et al. (2010). In addition, for a signal corrupted by
stationary white noise, the statistical efficiency of the
IPDFT method is investigated with respect to the singletone unbiased CramrRao Lower Bound. Finally, the
performance of the IPDFT method is compared with that
of the energy-based method on the basis of theoretical,
simulation and experimental results and with that of a
state-of-the-art method according to simulation and
experimental results. An application of Fourier transform
approach in modulation technique of experimental
studies is considered by Khazimullin et al. (2010). The
analysis has been confirmed by simulations and
measurements of a quartz wedge birefringence by means
of the photo elastic modulator. The obtained bias, noise
level and measuring speed are comparable and even
better than in lock-in amplifier technique.
The harmonic components of a carrier-based Pulse
Width-Modulated (PWM) voltage source converter (VSC)
output voltage are theoretically identified when the
modulating wave includes fundamental and baseband
harmonic components (Odavic et al., 2010). The analysis
is based on a double Fourier series expansion in two
variables. This approach to harmonic identification is
evaluated by comparison with a fast Fourier transform

analysis of simulated PWM waveforms. The Reference

Signal Generator (RSG) for voltage-source converters
(VSCs) that enables the maximization of its functionality
is presented by Borisov et al. (2010). The proposed RSG
is based on a combination of the Fortescue
decomposition with recursive discrete Fourier transform.
The method is characterized by computational efficiency,
excellent detection accuracy and fast dynamic response.
An experimental prototype of a multifunctional VSC with
the proposed RSG has been built and experimental
results are presented.
The feasibility and convenience of applying the efficient
DFT techniques are analyzed for the computation of
reference currents in APFs (Ortega et al., 2010). The two
different implementations of the running DFT, including
their recursive and non-recursive versions, are compared
to the SRF approach in terms of steady-state
performance, adequacy of transient response and
computational effort. In addition to simulation results, an
experimental setup is designed to prove the advantages
of resorting to the DFT.
PWM of a sampled signal produces harmonic distortion
is analyzed and proposed expressions are obtained for
the Fourier transform of the modulated output, which are
valid for any input signal (Colodro et al., 2010). In
addition, simple expressions are given to compute the
harmonic distortion factors for a sinusoidal input signal.
Finally, these analytical results are validated by
simulation for two data converters based on sigmadelta
modulation, where the sampled output of a multibitquantizer is pulse width modulated: one digital-to-analog
and one analog-to-digital converter, respectively. The
basics of two techniques, named here as the modified
harmonic domain and the modified dynamic harmonic
domain, proposed for calculating steady and dynamic
states are presented, respectively (Ramirez, 2011).
These techniques have their fundament in the harmonic
domain with a substantial improvement: the inclusion of
interharmonics in either steady or dynamic state. This is
performed through the use of the discrete Fourier
transform which allows an arbitrary frequency-domain
discretization, thus permitting the representation of
interharmonics.
The Cosine Self-Convolution Window (CSCW) is
proposed by Zeng et al. (2011). The main-lobe and sidelobe behaviors of the first to the third order CSCWs are
studied. A CSCW-based improved fast FFT for estimating
power system signal parameters, such as frequency,
phase and amplitude, is given. The effectiveness of the
proposed method was analyzed by means of computer
simulations and practical experiments for multi-frequency
signals with the variations of the power system frequency
as well as the presence of white noise and
interharmonics. A method for transmission line protection
using a variable window short-time Fourier is proposed
by Samantaray et al. (2008). The fault detection, the
impedance to the fault point is calculated using the

Saribulut

estimated phasors of the faulted current and voltage

signals which provide accurate results even with noisy
conditions. The fault location is calculated using
polynomial curve fitting technique with a devised index
obtained from the ratio of spectral energy of the voltage
and current signals, respectively.
Methods dependent on wavelet transform and FFT
The references given in the following are related with the
Wavelet Transform and FFT based control algorithms in
power quality issues. The comparison of Wavelet
Transform and FFT is also investigated.
An algorithm based on the wavelet-packet transform is
proposed for the analysis of harmonics in power systems
(Barros et al., 2008). The paper studies the selection of
the mother wavelet, the sampling frequency and the
frequency characteristics of the wavelet filter bank for the
two most common wavelet functions used for harmonic
analysis and compares the performance of the proposed
method with the results obtained using the DFT analysis
and the harmonic-group concept introduced by the IEC
under different measurement conditions. A modulation
technique is based on a newly designed scaling function
that is capable of supporting a non-uniform recurrent
sampling process by Saleh et al. (2009). This scaling
function generates sets of basic functions that span
spaces, of which a collection constructs a non-dyadictype multi-resolution analysis. Furthermore, the newly
designed scaling function has a dual synthesis scaling
function that is designed to reconstruct continuous-time
signals from their non-uniform recurrent samples. Several
performance tests are conducted for the proposed
wavelet-modulated inverter, when supplying linear,
nonlinear, static and dynamic loads.
A discrete wavelet transform is used as a conditioning
tool to filter the motor current prior to its processing by
the fractional FT (Sanchez et al., 2010). Experimental
results that are obtained with a 1.1-kW three-phase
squirrel-cage induction motor with broken bars are
presented to validate the proposed method. A method of
estimation of components above the harmonic frequency
range up to 9 kHz by wavelet filtering is proposed by
Tarasiuk (2011). The proposed method consists of the
application of FFT for wavelet coefficients after input
signal decomposition and partial synthesis for chosen
frequency bands. The algorithm and its implementation in
real device for power quality monitoring are presented. A
method for the classification of the power system
disturbances using support vector machines (SVMs) is
proposed by Eriti et al. (2010). The input vector is
started with the first best feature and incrementally added
the chosen features. After the addition of each feature,
the performance of the SVM is evaluated. The kernel and
penalty parameters of the SVM are determined by crossvalidation. The parameter set that gives the smallest

19

misclassification error is retained. Both the noisy and

noiseless signals are applied to the classifier.
Methods dependent on neural networks
The references given in the following are related with the
Artificial Neural Network based control algorithms in
power quality issues. The comparisons of Artificial Neural
Network with FFT and Wavelet Transform are also
investigated.
A strategy to estimate harmonic distortion from an AC
line is presented for power electronic converters
(Sekaran et al., 2008). An adaptive linear neural network
(ADALINE) is used to determine precisely the necessary
currents in order to cancel harmonics. The proposed
strategy is based on an original decomposition of the
measured currents to specify the neural network inputs.
This new decomposition is based on the Fourier series
analysis of the current signals and Least Mean Square
training algorithm carries out the weights. The proposed
strategy also allows extracting the harmonics individually.
The method is based on the extraction of fundamental
components of distorted line current using an ADALINE
network. The output of the ADALINE is compared with
distorted supply current to construct modulating signals
and to generate. An alternative method based on artificial
neural networks is presented to determine harmonic
components in the load current of a single-phase electric
power system with nonlinear loads, whose parameters
can vary so much in reason of the loads characteristic
behaviors as because of the human intervention (Oliveira
et al., 2010). The effectiveness of this method is verified
by using it in a single-phase active power filter with
selective compensation of the current drained by an AC
controller. The proposed method is compared with the
fast Fourier transform. The radial-basis-function neural
network is proposed to detect the harmonic amplitudes of
the measured signal (Chang et al., 2010). The proposed
method is compared with several commonly used
methods (FFT, Wavelet ...etc.).
A real-time classification method of PQ disturbances is
proposed by Zhanga et al. (2011). Five distinguished time
frequency statistical features of PQ disturbances are
extracted using RMS method and DFT. The proposed
method is tested using the nine types simulated
waveforms of PQ disturbances including voltage sag,
swell, interruption, harmonic, notch, flicker, oscillatory
transient, sag with harmonics and swell with harmonics. It
compared with common solutions which are usually
based on wavelet transform. The application of a
Complex Adaptive Linear Neural Network (CADALINE) in
tracking the fundamental power system frequency is
proposed Sadinezhada et al. (2009). In this method, by
using stationary-axes Park transformation in addition to
producing a complex input measurement, the decaying
DC offset is eliminated. This paper concludes with the

20

J. Electrical Electron. Eng. Res.

presentation of the representative results obtained in

numerical simulation and simulation in PSCAD/EMTDC
software as well as in practical study.
Methods dependent on fuzzy logic
A fast fuzzy-logic-supported anytime frequency rangeestimation procedure is proposed, which makes it
possible to execute the frequency estimation after one
quarter of the period of the unknown signal, that is, the
adaptation and Fourier analysis can be performed without
any delay (Varkonyi et al., 2009). The design of a tool to
quantify power quality (PQ) parameters using wavelets
and fuzzy sets theory is proposed by Mehera et al.
(2010). The tool merges the best characteristics of these
two theories in establishing a method to analyze PQ
events. The proposed method addresses two issues,
such as selection of discriminative features and classifies
event classes with minimum error. Wavelet features (WF)
of PQ events are extracted using wavelet transform (WT)
and fuzzy classifiers classify events using these features.
Often the captured signals are corrupted by noise.
Varieties of PQ events including voltage sag, swell,
momentary interruption, notch, oscillatory transient and
spikes are considered for performance analysis.
Methods dependent on other methods
Software sampling frequency adaptive algorithm that can
obtain the actual signal frequency more accurately is
presented and then adjusts sampling interval base on the
frequency calculated by software algorithm and modifies
sampling frequency adaptively (Pan et al., 2006). It can
reduce synchronous error and impact of spectral leakage.
This algorithm has high precision just like the simulations
show and it can be practical methods in power system
harmonic analysis since it can be implemented easily. A
new modification of the least squares Pronys method for
Pronys method for power-quality analysis in terms of
estimation of harmonics and interharmonics in an electric
power signal is proposed by Zygarlicki et al. (2010). The
so-called reduced Pronys method can be competitive, in
some specific case, to the Fourier transformation method
and the classical LS Pronys method. The modification
constitutes in a specific selection of a constant frequency
vector in a Fourier-like manner leading to a remarkable
reduction of the computational burden and enabling
online real-time computations.
An efficient procedure that includes a high-resolution
Prony-based method in conjunction with the down
sampling technique for harmonics and interharmonics
detection of the measured power signal is proposed by
Chang and Chen (2010). The performance of the
proposed method is validated by testing the simulated
and actual measured power signals. Results are

compared with those obtained by fast Fourier transform

with and without synchronization, IEC sub-grouping
method and other commonly used linear prediction
approaches adopted in the Pronys method. An approach
for the construction of a family of desired order
continuous polynomial time window functions is
presented without self-convolution of the parent window
(Singla et al., 2010).
Published review papers on FFT
The higher order of continuity of the time window
functions at the boundary of the observation window
helps in suppressing the spectral leakage. Closed-form
expressions for window functions in the time domain and
their corresponding Fourier transform are derived.
Several commonly used methods for time-varying
harmonic and interharmonic detection of measured
waveforms are reviewed and implemented in an
integrated virtual instrumentation (Chen and Chang,
2010). Compared from the aspect of frequency
identification for the reviewed methods, general
guidelines for performing harmonic and inter-harmonic
detection are also developed for the educational purpose.
Major continuous-time, discrete-time and discrete Fourier
related transforms as well as Fourier-related series are
discussed both with real and complex kernels (Ersoy,
1994). The complex Fourier transforms, Fourier series,
cosine, sine, Hartley, Mellin, Laplace transforms and Ztransforms are covered on a comparative basis. The
short-time Fourier-related transforms are discussed for
applications involving non-stationary signals. The onedimensional transforms discussed are also extended to
the two-dimensional transforms. Since fast algorithms for
the DFT were first introduced thirty years ago, they have
had a major impact on signal processing and are now a
basic part of every electrical engineer's education.
However, some of the options and particularly the recent
advances are not as widely known as they deserve.
The fast algorithms for the DFT are reviewed by Blair
(1995a). It considers why the DFT works and looks at the
various fast algorithms for transforms whose orders are a
power of two. The techniques for other orders, at
adapting algorithms for purely real data and at the
problems for fixed-point noise are discussed by Blair
(1995b). The second of two which review the fast
algorithms for the DFT, looks at algorithms for transforms
whose orders are not a power of two. Also discussed are
ways of adapting algorithms for purely real data, the
problems of fixed-point noise and implementation options
with existing hardware.
CONCLUSION
Fourier transform can be used to detect, classify and

Saribulut

Methods of Published
Publication Years
Articles
1989 1993 1999 2005 2006 2007 2008 2009 2010 2011
Fourier Transform
(DFT, Recursive DFT,
FFT,, etc.)
FFT versus Wavelet
Transform
FFT versus Neural
Networks
FFT versus Fuzzy
Logic
Other Methods (PLL,
Prony's Method,
Continous Polynomial)

21

Total

36

Figure 7. Selected methods in the reviewed articles and publishing years.

analyze PQ disturbances with certain accuracy. An

exhaustive review of Fourier transform in power quality
issues is performed to provide a clear understanding on
their applications. Most power quality analyzers also use
FFT-based algorithm to identify the harmonics of the
measured signals.
This paper presents the fundamentals of Fourier series,
Fourier transform, discrete Fourier transform and fast
Fourier transform with simple examples and review of
Fourier transform to provide a clear understanding of its
applications in power quality issues. Most of the selected
articles were published in Scientific Index, Science
Citation Index (SCI) or SCI extended between 2007 and
2011. It is concluded with the review of published papers
on power quality disturbances that most of the selected
methods are included, either Fourier transform or
methods based on Fourier transform. The other methods
are wavelet transform, neural networks and FUZZY logic,
respectively. The selected methods in the reviewed
articles and publishing years are given as shown in
Figure 7.
In the literature, there are few review studies with
published articles. Most of these studies are related to
the explanation of DFT or FFT transforms. The presented
paper is the first review study included both detail
examination of published articles related to power quality
in recent years and classification of these studies
(Sarbulut, 2012). The result of the reviewed published
papers on FFT shows that Fourier transform can be used
for many power quality issues detection and restoration.
In addition, it can be used in image processing tasks,
filtration, disturbance detection, modulation and
demodulation.

REFERENCES
Albanna AZ, Hatziadoniu CJ (2010). Harmonic modeling of hysteresis
inverters in frequency domain. IEEE T. Power. Electron. 25(5):11101114.
Baran ME, Jinsang K (2006). A classifier for distribution feeder over
current analysis. IEEE T. Power Deliv. 21(1):456-462.
Barros J, Diego RI (2008). Analysis of harmonics in power systems
using the wavelet-packet transform. IEEE T. Instrum. Meas. 57(1):6369.
Belega D, Dallet D, Slepicka D (2010). Accurate amplitude estimation of
harmonic components of incoherently sampled signals in the
frequency domain. IEEE T. Instrum. Meas. 59(5):1158-1166.
Blair GM (1995a). A review of the discrete Fourier transform. 1.
Manipulating the powers of two. Electron. Comm. Eng. J. 7(4):169177.
Blair GM (1995b). A review of the discrete Fourier transform. 2. Nonradix algorithms, real transforms and noise. Electron. Comm. Eng. J.
7(5):187-194.
Borisov K, Ginn HL (2010). Multifunctional VSC based on a novel
fortescue reference signal generator. IEEE T. Ind. Electr. 57(3):10021007.
Breitenbach A (1999). Against spectral leakage. Measurement
25(2):135-142.
Chang GW, Chen CI (2010). An accurate time-domain procedure for
harmonics and interharmonics detection. IEEE T. Power Deliv.
25(3):1787-1795.
Chang GW, Chen CI, Teng YF (2010). Radial-basis-function-based
neural network for harmonic detection. IEEE T. Ind. Electron.
57(6):2171-2179.
Chen CI, Chang GW (2010). Virtual instrumentation and educational
platform for time-varying harmonic and interharmonic detection. IEEE
T. Ind. Electron. 57(10):3334-3342.
Chen CM, Hung CC, Huang YH (2010). An energy-efficient partial FFT
processor for the ofdma communication system. IEEE T. Circ. Syst.
57(2):136-140.
Colodro F, Torralba A (2010). Spectral analysis of pulse widthmodulated sampled signals. IEEE T. Circuits II: Express Briefs
57(3):622-626.
Cooley JW, Tukey JW (1965). An algorithm for the machine calculation
of complex Fourier series. Math. Comput. 19:297-301.

22

J. Electrical Electron. Eng. Res.

Cortes A, Velez I, Sevillano JF, Irizar A (2006). An approach to simplify

the design of IFFT/FFT cores for ofdm systems. IEEE T. Consum.
Electron. 52(1):26-32.
De Souza FAS, Bradaschia HEP, Cavalcanti F, Rizo MC, Rodriguez M,
Neves FJ (2010). A space-vector discrete Fourier transform for
unbalanced and distorted three-phase signals. IEEE T. Ind. Electron.
57(8):2858-2867.
Eriti H, Uar A, Demir Y (2010). Wavelet-based feature extraction and
selection for classification of power system disturbances using
support vector machines. Elect. Power Syst. Res. 80(7):743-752.
Ersoy OK (1994). A comparative review of real and complex Fourierrelated transforms. Proc. IEEE 82(3):429-447.
Heideman MT, Johnson DH, Burrus CS (1985). Gauss and the history
of the fast Fourier transform, Archive for History of Exact Sciences.
Arch. Hist. Exact Sci. 34(3):265-277.
Heriakian I (2006). Harmonic compensation using the sliding DFT
algorithm for three-phase active power filter. Electr. Power. Qual.
Utili. J. 12(2):15-20.
Heydt GT (1989). A new method for the calculation of subtransmission
and distribution system transients based on the FFT. IEEE Power
Eng. Rev. 9(7):73.
Heydt GT, Fjeld PS, Liu CC, Pierce D, Tu L, Hensley G (1999).
Applications of the windowed FFT to electric power quality
assessment. IEEE T. Power Deliv. 14(4):1411-1416.
Khazimullin MV, Lebedev YA (2010). Fourier transform approach in
modulation technique of experimental measurements. Rev. Sci. Instr.
81(4):1-8.
Kuo SM, Lee BH (2001). Real-time digital signal processing
implementations
applications
and
experiments
with
the
TMS320C55x. John Wiley & Sons, Inc.
Lavopa E, Zanchetta P, Sumner M, Cupertino F (2009). Real-time
estimation of fundamental frequency and harmonics for active shunt
power filters in aircraft electrical systems. IEEE T. Ind. Electron.
56(8):2875-2884.
Le Roux AD, Mouton H, Akagi H (2009). DFT-based repetitive control of
a series active filter integrated with a 12-pulse diode rectifier. IEEE T.
Power Electron. 24(6):1515-1521.
Lin HC (2008). Inter-harmonic identification using group-harmonic
weighting approach based on the FFT. IEEE T. Power Electron.
23(3):1309-1319.
Matyas Z, Aro M (2005). HV impulse measuring systems analysis and
qualification by estimation of measurement errors via FFT,
convolution and IFFT. IEEE T. Instrum. Meas. 54(5):2013-2019.
McGrath BP, Holmes DG (2009). Natural capacitor voltage balancing
for a flying capacitor converter induction motor drive. IEEE T. Power
Electron. 24(6):1554-1561.
Mehera SK, Pradhanb AK (2010). Fuzzy classifiers for power quality
events analysis. Elect. Power Syst. Res. 80(1):71-76.
Mertins A (1996). Signal analysis wavelets, filter banks, time-frequency
transforms and applications. John Wiley & Sons, Inc.
Novotny M, Slepicka D, Sedlacek M (2007). Uncertainty analysis of the
RMS value and phase in the frequency domain by non-coherent
sampling. IEEE T. Instrum. Meas. 56(3):983-989.
Odavic M, Sumner M, Zanchetta P, Clare JC (2010). A theoretical
analysis of the harmonic content of PWM waveforms for multiplefrequency modulators. IEEE T. Power Electr. 25(1):131-141.
Oliveira CF, Goedtel AA, Silva A, Serni IN, Nascimento PJA (2010).
Neural network-based approach for identification of the harmonic
content of a nonlinear load in a single-phase system. IEEE Latin Am.
T. 8(1):65-73.
Ortega JMM, Macias JAR, Exposito AG, Mannozzi SC, Villarejo MB
(2010). Reference current computation for active power filters by
running DFT techniques. IEEE T. Power Deliv. 25(3):1986-1995.
Pan C (1993). Design of a windowless digital filter using FFT algorithm.
IEEE Trans. Acoust. Speech Sig. Proc., pp. 6221.
Pan LD, Wang F (2006). A software sampling frequency adaptive
algorithm for reducing spectral leakage. Frontiers Electr. Electron.
Eng. China 1(4):375-379.
Qian H, Zhao R, Chen T (2007). Measuring power system harmonics
and interharmonics by an improved fast Fourier transform-based
algorithm. IEEE T. Power. Deliv. 2(2):1064-1069.

Radil T, Ramos PM, Serra AC (2009). New spectrum leakage correction

algorithm for frequency estimation of power system signals. IEEE T.
Instrum. Meas. 58(5):1670-1679.
Ramirez A (2011). The modified harmonic domain: interharmonics.
IEEE T. Power Deliv. 26(1):235-241.
Sadinezhada I, Joorabian M (2009). A novel frequency tracking method
based on complex adaptive linear neural network state vector in
power systems. Elect. Power Syst. Res. 79(8):1216-1225.
Saleh SA, Moloney CR, Rahman MA (2009). Development and testing
of wavelet modulation for single-phase inverters. IEEE T. Ind.
Electron. 56(7):2588-2599.
Salor (2009). Spectral correction-based method for interharmonics
analysis of power signals with fundamental frequency deviation.
Electr. Power Syst. Res. 79(7):1025-1031.
Samantaray SR, Dash PK (2008). Transmission line distance relaying
using a variable window short-time Fourier transform. Elect. Power
Syst. Res. 78(4):595-604.
Sanchez MP, Guasp MR, Daviu JAA, Folch JR, Cruz JP, Panadero RP
(2010). Diagnosis of induction motor faults in the fractional Fourier
domain. IEEE T. Instrum. Meas. 59(8):2065-2075.
Sarbulut L (2012). Detection and mitigation methods of power quality
disturbances. PhD Thesis, ukurova University, Institute of Natural
and Applied Science.
Sekaran EC, Anbalagan P (2008). Comparison of neural network and
fast-Fourier transform based selective harmonic extraction and total
harmonic reduction for power electronic converters. Asian Power
Electron. J. 2(1):1-9.
Sheikh F, Masud S, Bing B (2009). Harmonic power detection in
wideband cognitive radios. IET Sig. Proc., 3(1):40-50.
Singla P, Singh T (2010). Desired order continuous polynomial time
window functions for harmonic analysis. IEEE T. Instrum. Meas.
59(9):2475-2481.
Sumathi P, Janakiraman PA (2008). Integrated phase-locking scheme
for SDFT-based harmonic analysis of periodic signals. IEEE T.
Circuits II: Express Briefs 55(1):51-55.
Sundararajan D (2001). The discrete Fourier transform: theory,
algorithms and applications. World Scientific Publishing Co. Pte. Ltd.
Tarasiuk T (2011). Method, algorithm and device for estimation of
components above the harmonic frequency range up to 9 kHz.
Measurement 44(1):219-229.
Thede L (2004). Practical analog and digital filter design. Artech House
Inc.
Varkonyi KAR (2009). Fast anytime fuzzy-Fourier estimation of
multisine signals. IEEE T. Instrum. Meas. 58(5):1763-1770.
Vodyakho O, Hackstein D, Steimel A, Kim T (2008). Novel direct
current-space-vector control for shunt active power filters based on
the three-level inverter. IEEE T. Power Electron. 23(4):1668-1678.
Vodyakho O, Kim T (2009). Shunt active filter based on three-level
inverter for three-phase four-wire systems. IET Power Electron.
2(3):216-226.
Vodyakho O, Kim T, Kwak S, Edrington CS (2009). Comparison of the
space vector current controls for shunt active power filters. IET Power
Electron. 2(6):653-664.
Wikipedia (2013). (http://en.wikipedia.org/wiki/Fast_Fourier_transform)
Yang JZ, Yu CS, Liu CW (2005). A new method for power signal
harmonic analysis. IEEE T. Power Deliv. 20(2):1235-1239.
Yena WF, Shingchern DY, Changc YC (2009). Real-time FFT with precalculation. Comput. Electr. Eng., 35(3):435-440.
Zeng B, Teng Z (2011). Parameter estimation of power system signals
based on cosine self-convolution window with desirable side-lobe
behavior. IEEE T. Power Deliv. 26(1):250-257.
Zhang Q, Liu H., Chen H, Li Q, Zhang Z (2008). A precise and adaptive
algorithm for interharmonics measurement based on iterative DFT.
IEEE T. Power. Deliv. 23(4):1728-1735.
Zhanga M, Lib K, Hub Y (2011). A real-time classification method of
power quality disturbances. Electric Power Syst. Res. 81(2):660-666.
Zygarlicki J, Zygarlicka M, Mroczka J, Latawiec KJ (2010). A reduced
pronys method in power-quality analysis-parameters selection. IEEE
T. Power Deliv. 25(2):979-986.