Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 27, NO.

2, APRIL 2012

971

A Method to Improve the Interharmonic Grouping


Scheme Adopted by IEC Standard 61000-4-7
Jin Hui, Honggeng Yang, Member, IEEE, Wilsun Xu, Fellow, IEEE, and Yamei Liu

AbstractThe interharmonic group concept adopted by the IEC


61000-4-7 is very useful for addressing the need to monitor interharmonics in power systems. This paper shows that grouping
spectral bins into interharmonics and harmonic components of 5
Hz resolution could yield misleading information for cases where
an interharmonic components frequency is very close to a harmonic frequency. Since interharmonics in such cases can cause
more severe waveform modulation, the interharmonic grouping
scheme needs to be improved. A signal-processing method that can
separate interharmonic and harmonic components in close proximity is, therefore, proposed in this paper to address this need.
The method is based on the estimation of leakage values caused by
interharmonics, at harmonic frequencies. Simulation studies and
field data analysis showed that the proposed method was able to
extract interharmonic and harmonic components correctly. Moreover, this paper also discusses how to apply the method to enhance
the spectral grouping scheme of the IEC 610004-7.
Index TermsDiscrete Fourier transform (DFT), harmonic
analysis, interharmonics, spectral leakage.

I. INTRODUCTION
ARMONIC and interharmonic measurements are one of
the common tasks for power-quality (PQ) monitoring and
troubleshooting [1], [2]. The discrete Fourier transform (DFT)
and its fast algorithm fast Fourier transform (FFT) are the most
commonly used techniques to process the measured data. However, these techniques have accuracy problems when a waveform contains interharmonics. These problems are due to the
spectral leakage and picket fence effects that cannot be avoided
when interharmonics are present.
To address this problem and standardize the measurement approach, the International Electrotechnical Commission established a harmonic and interharmonic measurement protocol in
IEC 61000-4-7 [3]. DFT is still the processing tool, and the recommended window width is 12 fundamental frequency cycles
(for a 60 Hz system). This width yields a frequency resolution of
5 Hz. The DFT results are grouped into harmonics and interharmonics components. This approach is easy to implement and
can deal with spectral leakage issues in most cases. However,

Manuscript received August 12, 2011; revised November 17, 2011; accepted
December 27, 2011. Date of publication February 20, 2012; date of current version March 28, 2012. Paper no. TPWRD-00684-2011.
J. Hui, H. Yang, and Y. Liu are with the College of Electrical Engineering and
Information Technology, Sichuan University, Chengdu 610065, China (e-mail:
hj4655@163.com; yangsi@mail.sc.cninfo.net; huijean@live.cn).
W. Xu is with the Department of Electrical and Computer Engineering, University of Alberta, Edmonton, AB T6G 2V4 Canada (e-mail: wxu@ualberta.ca).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TPWRD.2012.2183394

if the frequency of an interharmonic component is very close


to a harmonic frequency, the spectral leakage problem caused
by 5 Hz frequency resolution can corrupt both the harmonic
and interharmonic group results. The harmonic results are often
overestimated, and the interharmonic results underestimated.
As a result, the interharmonic component cannot be characterized properly [1]. Researchers have found that interharmonics
in close proximity of a harmonic frequency can cause much
more significant waveform fluctuations (modulations) [4], [5]
and, therefore, have the highest need for their proper measurement and characterization. Some studies have been published
on reducing the impact of spectral leakage [6][15]. However,
if an interharmonic component has a distance less than 5 Hz
(a frequency resolution) from a harmonic frequency, strong interharmonic-harmonic spectral interference will occur and will
be beyond the processing capability of the methods proposed in
these studies.
In this paper, the authors propose a method to solve the aforementioned problem and to improve the accuracy of the IEC interharmonic group results. The idea is to group harmonics based
on the harmonic spectrum (the total spectrum of all harmonic
components) and to group interharmonics based on the interharmonic spectrum (the total spectrum of all interharmonic
components). In this way, the spectral interference between the
harmonics and interharmonics is eliminated. The determination
of the harmonic spectrum and interharmonic spectrum is
achieved by calculating the interharmonic leakage contributions
at harmonic frequencies.
This paper is structured as follows. Section II introduces
the interharmonic group concept defined in IEC 61000-4-7. In
Section III, the proposed method is described. In Section IV,
the simulations and field experiment results are provided to
show the effectiveness of the proposed method. Section V
concludes this paper.
II. IEC INTERHARMONIC GROUP CONCEPT
The IEC defined harmonic subgroup includes one harmonic
bin and two bins adjacent to the central harmonic. For example,
115, 120, 125 Hz form the 2nd harmonic subgroup for a 60
Hz system. The magnitude of this group is defined as the rms
value of the three bins
(1)

and represent the magnitude of the DFT results


where
refers to the frequency
and the harmonic order, respectively,

0885-8977/$31.00 2012 IEEE

972

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 27, NO. 2, APRIL 2012

Fig. 1. Harmonic and interharmonic subgroups based on IEC 61000-4-7.

resolution (about 5 Hz), and


refers to the fundamental frequency (about 60 Hz).
The other spectral bins between the two adjacent harmonic
subgroups form the interharmonic subgroup. For example,
70, 75, 80,
110 Hz form the 1.5th interharmonic subgroup. The magnitude of this group is defined as the rms value
of all these nine bins

Fig. 2. Separated harmonic and interharmonic spectra. (a) Harmonic spectrum.


(b) Interharmonic spectrum.

III. PROPOSED SPECTRAL-PROCESSING TECHNIQUE


(2)
Fig. 1 illustrates the group concept defined before, where
HSG and IHSG are short for harmonic subgroup and interharmonic subgroup, respectively. The spectrum is obtained by
performing DFT on a 12-cycle waveform snapshot of a signal
containing five frequency components (Harmonics: 60, 120, 180
Hz. Interharmonics: 117.5, 177.5 Hz).
If interharmonics are in close proximity with a harmonic, the
aforementioned grouping approach can yield misleading results.
Fig. 1 is an example of this phenomenon. According to the definition of a harmonic subgroup, the DFT bins located at 115
Hz (bin ), 120 Hz (bin ), and 125 Hz (bin ) are all included
as the 2nd harmonic component (subgroup). A large interharmonic component (bin ) is thus misclassified as a harmonic
component. On the other hand, the magnitude of the 1.5th interharmonic subgroup will be underestimated. Based on these
results, a PQ meter user could mistakenly conclude that the measured waveform has a large harmonic component at 120 Hz and
an insignificant interharmonic component between 60 Hz and
120 Hz. Such a conclusion can easily mislead troubleshooting
efforts.
Interharmonics, which produce worst flicker levels, can be
seen as the most harmful interharmonics. The literature [4]
shows an experimental result from a study of the acceptable
magnitude of a single interharmonic component which causes
perceptible light flicker by various lamps. The result reveals that
the frequencies of the lowest acceptable magnitudes are all close
to the harmonic frequencies.
In summary, the accuracy of the IEC interharmonic and harmonic group results needs to be improved. One of the direct
benefits will be the availability of accurate measurement results
for PQ troubleshooting.

The proposed method uses only the 12-cycle data available


to an IEC-compliant measurement device. Its basic idea is to
separate the DFT spectrum produced from a 12-cycle snapshot
into a harmonic spectrum and an interharmonic spectrum.
The RMS values of the harmonic and interharmonic groups are
calculated based on the two separated spectra. Fig. 2 shows the
separated spectra from the spectrum shown in Fig. 1. Note that
all of these calculations are based on the IEC frequency resolution (5 Hz).
A. Harmonic and Interharmonic Spectra Separation
The IEC standard requires the DFT window length to
be synchronized with the real system frequency by using a
phase-locked loop (PLL) or other frequency synchronization
techniques. Therefore, the harmonic energies will be all concentrated only at the harmonic frequencies in the original DFT
spectrum, and no leakage will take place. In contrast, the interharmonic energies will still leak at all frequencies. Thus, the
spectral bins at the harmonic frequencies will be determined
by the harmonics and the interharmonics while the bins at
the interharmonic frequencies will be determined only by the
interharmonics. For instance, bin in Fig. 1 is determined by
both the 2nd harmonic and the two interharmonics while bin
and bin are determined only by the two interharmonics.
Therefore, the whole spectrum separation problem can be
transformed into the problem of separating the harmonic and
interharmonic contributions at the harmonic frequencies (e.g.,
by separating bin in Fig. 1 into bin and bin in Fig. 2).
From a mathematical perspective, the interharmonic leakage
values (contributions) at the harmonic frequencies (e.g., ) can
be calculated by knowing the spectral lines at the interharmonic
frequencies (e.g., bins and ), as long as the number of interharmonic components in the signal is known. Thus, the DFT results at the interharmonic frequencies and the calculated leakage

HUI et al.: A METHOD TO IMPROVE THE INTERHARMONIC GROUPING SCHEME ADOPTED BY IEC STANDARD 61000-4-7

values at harmonic frequencies form the final interharmonic


spectrum, and the remaining spectrum forms the harmonic
spectrum.
In the following text, the leakage calculation processes for
considering one, two, and three interharmonics in the signal are
introduced. Since the DFT result is nonlinear with respect to the
frequency, the calculation process will be complex and timeconsuming. To speed up and simplify the calculation, this paper
simplifies the DFT results to transform the nonlinear problem
into a simple liner problem.
1) Simplification of the Interharmonic DFT Result: Consider
an interharmonic component (ignore the negative frequency
part)

973

Based on (8), the leakage contribution of this interharmonic


at a frequency of times the frequency resolution away from
can be expressed as
the th harmonic

(9)
where

(10)
(11)

(3)
are the frequency, amplitude, and phase
where , , and
angle of the component, respectively.
Digitize the signal with the sample interval of
(4)
,
is the sample number per
where
is the cycle number in the time window.
cycle, and
Its DFT is

(5)
where
Deform the above equation

is represented by
For the readability,
in the following text.
2) Interharmonic Leakage Calculation by Considering one
Interharmonic Component: If only one interharmonic is considered (with frequency ) around the th harmonic, the specHz will be contributed by the considered
tral values around
interharmonic only
(12)
means the total spectral value contributed by all inwhere
means the spectral value only conterharmonics, and
tributed by the interharmonic with frequency .
As stated before, the interharmonic spectral values adjacent
Hz directly correspond to the DFT results and have the
to
expressions
(13)

(6)
(14)
.
where
As long as the frequency of the item to be summed in (6) (i.e.,
) is very low in comparison with the sampling frequency , the summation in (6) can be replaced by integrating

After simple mathematical transformation, two linear equations can be formed as


(15)

(7)
(8)
The concerned spectral bin, in reality, is usually located close
enough to make the above assumption true. For instance, when
253 Hz,
,
37 Hz 7.4 ,
12,
64, the result of
by using (7) is
, while the exact value by using (5) is
.
The difference is very small compared to the absolute value.
Based on the simulation studies by the authors, the sample frequency of more than 100 points per cycle is acceptable to make
(7) hold.

where
(16)
By solving the above linear equations, the interharmonic
leakage value at
Hz
can be obtained by
(17)
A simple numerical example will be used here to illustrate the
calculation process. In this example, the interharmonic leakage
value at 120 Hz in Fig. 1 is calculated by considering only the
117.5 Hz interharmonic component, as the impact of the 177.5

974

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 27, NO. 2, APRIL 2012

Hz component is small. The spectral values located at 115 Hz


and 125 Hz (bins and ) are
(18)
(19)
The interharmonic leakage value at 120 Hz
(2) (bin
in Fig. 2) can be obtained by using (15) and (17) with
(2) and
(2). The result is
the two known values
, which is a good estimation compared to the
. The difference is due to the
real value of
influence of the 177.5 Hz component.
3) Interharmonic Leakage Calculation by Considering two
Interharmonic Components: If two interharmonics are considered around the th harmonic, the total leakage values conHz are
tributed by the two interharmonics around

(20)
where
and
refer to the frequencies of the two interharmonics.
Through a series of mathematical transformations, the four
Hz will meet the following linear
spectral lines surrounding
equations:

(21)
where

(22)

As a result, the interharmonic leakage value at


solved by

Hz can be

(23)
The interharmonic leakage value at 120 Hz in Fig. 2 (bin )
was again calculated by considering two interharmonic components. In this case, the four DFT results centered with 120 Hz
,
were used. The calculated leakage result is
which is better than the result from considering only one interharmonic since the new result includes the impact of the 177.5
Hz interharmonic. The small error is due to the negative frequency components.
4) Calculate Interharmonic Leakage Values by Considering
Three Interharmonic Components: Similar to the two aforementioned cases, if three interharmonic components are conHz will satisfy six
sidered, the six DFT results adjacent to
linear equations (the mathematical expression is omitted here
to save space), and the interharmonic leakage values can be obtained by solving the corresponding equations.
Here, the interharmonic leakage value at 120 Hz in Fig. 2
(bin ) was estimated once again by considering three inter-

, which is
harmonic components. The result is
exactly the same as the true value. It is better than both the previous results since more interharmonic components (three instead of only one or two) are included.
The aforementioned results show that as long as the real interharmonic number is less than or equal to the considered number,
accurate results can be obtained. However, the number of interharmonic components is usually unknown before measuring a
raw signal. How to determine the number of the considered interharmonics will be discussed further in Section IV.
B. Improved Grouping Scheme
After calculating the interharmonic leakage values at harmonic frequencies one-by-one, the pure harmonic and interharmonic spectra can thus be established. Then, the group results
can be calculated based on the separated spectra:
1) For harmonics, the spectral lines at the harmonic frequencies of the harmonic spectrum can be obtained by subtracting the calculated interharmonic leakage values at the
harmonic frequencies from the original DFT and the results directly correspond to the real harmonic results.
2) However, for interharmonics, their energy is not concentrated at specific frequencies in the interharmonic spectrum, but is usually spread out to all frequency points
[shown in Fig. 2(b)] due to asynchronous truncation. If an
interharmonic constituent is near a harmonic, the main energy of the interharmonics will distribute around the harmonic frequency. Therefore, the three spectral bins centered with the harmonic frequency [as shown in Fig. 2(b) in
the interharmonic spectrum] cannot be ignored and should
be grouped into the right interharmonic group. Based on
the attenuation trend of a components spectral bins, the
three bins should be included into the interharmonic group
to which the larger one of the two side bins belongs. For
instance, in Fig. 2(b), bins , , and should be grouped
into the 1.5th interharmonic group, since bin is larger
than bin . Similarly, the spectral lines at 175 180, and 185
Hz should be included in the 2.5th interharmonic group.
C. Complete Method
The complete method is summarized as follows.
1) Capture a 12-cycle snapshot of the waveform and perform
DFT on the snapshot.
2) Calculate the interharmonic leakage values at the harmonic
frequencies one-by-one.
3) Obtain separated harmonic and interharmonic spectra.
4) Calculate harmonic and interharmonic group results based
on their own spectra.
If one needs to know the frequencies, magnitudes, and phase
angles of the main interharmonics, interpolation algorithms
[8][15] can be employed to meet this requirement since the
interference from the harmonics has already been eliminated
by this spectrum separation technique discussed before.
IV. VERIFICATION STUDIES
In this section, we use two groups of data to assess the performance of the IEC recommended method and the proposed

HUI et al.: A METHOD TO IMPROVE THE INTERHARMONIC GROUPING SCHEME ADOPTED BY IEC STANDARD 61000-4-7

TABLE I
PARAMETERS OF THE TEST SIGNAL

975

TABLE II
CALCULATED INTERHARMONIC LEAKAGE VALUES
AT HARMONIC FREQUENCIES.

Fig. 3. DFT spectrum of the 0.2-s waveform snapshot.


TABLE III
HARMONIC AND INTERHARMONIC GROUP RESULTS

improved method. The first group is from the computer simulation, which represents the ideal cases. The second group is a
real-world case study of an industrial VFD.
A. Simulation Study
Assume that the test signal consists of six tones, the parameters of which are listed in Table I. For the sake of simplicity
and without loss of generality, all phase angles of the frequency
components are set to be zeros.
Fig. 3 shows the 12-cycle spectrum based on the recommendation of the IEC standard, which reveals that the DFT results
at the harmonic frequencies no longer correspond to the actual harmonic parameters even though the sampling is synchronized with the fundamental frequency. This finding occurred because the spectral lines at harmonic frequencies were interfered
with by interharmonics. The severity of the interference is determined by the closeness of the harmonic to the nearby interharmonic and its relative amplitude. For the fundamental component, since it has the strongest energy and the interharmonics
are relatively far away, the interharmonic interference has little
impact and can be ignored. However, for the second and third
harmonic tones, since they are located in close proximity to the
interharmonics, whose energies are comparable to theirs, the interference is strong. On the other hand, for the interharmonics
with frequencies 117.5 Hz and 182.5 Hz, the spectral lines in
their main lobes are also disturbed by the two harmonics.
To eliminate the interference discussed before, the proposed
spectrum separation method was used to obtain the pure harmonic and interharmonic spectra. The first step of the separation method is to calculate the interharmonic leakage values at
harmonic frequencies. Table II lists the calculated results (IH is
short for the interharmonic, and H is short for the harmonic
in the following). The test results show that for each harmonic,
the three results obtained by considering different interharmonic
numbers are comparable, and all are around the actual values.
When three interharmonics are considered, the calculated IH
leakage values are almost the same as the true values since the
real number of interharmonic components is also three.
After obtaining the interharmonic leakage values at the
harmonic frequencies, the pure interharmonic spectrum can be
established. Then, the harmonic spectrum can also be formed
by subtracting the calculated interharmonic leakage values from

the original DFT harmonic spectrum. Based on the improved


grouping scheme, the group results for both harmonics and
interharmonics were calculated, and the results are shown in
Table III.
The results show that for the fundamental 0.5th and 2.5th
groups, the results from the IEC method are acceptable and close
to the real values. However, for the 2nd, 3rd, 1.5th, and 3.5th
groups, the results are far from the true values because the interharmonic constituents in the two interharmonic groups are located close to the two nearby harmonics. The IEC measurement
method irrationally grouped the interharmonic bins of 115 Hz
and 125 Hz into the second harmonic group and grouped the interharmonic bins of 175 Hz and 185 Hz into the third harmonic
group. Thus, the harmonic results were overestimated, and the
interharmonic results were underestimated. In contrast, the results from the improved method are all acceptable and accurate
since it groups the main bins of the interharmonics into the correct interharmonic groups.
Furthermore, to verify the antinoise performance, the calculation was also performed for a noise environment in which
the signal-to-noise ratios (SNRs) varied in the range of
dB at an increment of 20 dB. The results are shown in Fig. 4.
(Improved methods 13 refer to the proposed methods by
considering 13 IHs, respectively, which suggests that the proposed method was still effective in dealing with noise-disturbed
signals.)
If one wants to know the magnitudes, frequencies, and phase
angles of the main interharmonics, the interpolation algorithm
developed in other studies can meet this requirement. In this
paper, the transformed discrete Fourier transform (TDFT) developed by [10] was used to obtain the detailed information

976

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 27, NO. 2, APRIL 2012

if no real interharmonics are present. Too small interharmonic bins may produce large errors in estimating the
true interharmonic leakage impact on certain harmonics.
Consequently, for each harmonic, a criterion has to be
specified to filter out the noise and the ignorable interharmonic interference
(24)

Fig. 4. Results of interharmonic groups with the presence of white Gaussian


noise. (a) The 0.5th interharmonic group. (b) The 1.5th interharmonic group. (c)
The 2.5th interharmonic group. (d) The 3.5th interharmonic group.

TABLE IV
RESULTS OF INTERHARMONIC PARAMETERS

Practical investigations by the authors show that the impact


of the measurement noise is usually much larger than that
of the instrument acquiring resolution for real power signals, so the coefficient in (24) should be user-defined and
determined by the noise level.
2) The test signal in the simulation section contains exactly
three interharmonic constituents. Strictly, three interharmonics should be considered in calculating the interharmonic leakage values at the three harmonic frequencies.
However, the result implies that considering one interharmonic tone is enough since the numbers of the interharmonics close to the harmonics are all less than two in this
case. For practical signals, the number of dominant interharmonics interfered with by harmonics is also usually less
than two. Therefore, considering one interharmonic can
meet the measurement requirement for practical uses.
3) On the other hand, the results obtained by considering two
and three interharmonics are also needed since they can
be used to confirm the number of interharmonics near certain harmonics and to crosscheck the results by considering
one interharmonic component. If only one interharmonic
is near a harmonic, the results by considering one, two,
and three interharmonics will be almost the same since the
other interharmonics effects are ignorable. If the three results are all far from each other, the interharmonic bins adjacent to the harmonic frequencies are likely to be caused
by the irregular fluctuation of the harmonic component. In
these cases, the IEC subgroup method can be used directly,
and the adjacent interharmonic bins should be grouped into
the harmonic groups to evaluate the final rms values.
C. Real Case Study

of the two main interharmonics. The results are tabulated in


Table IV, which shows that the TDFT gave accurate interharmonic parameters. Note that the spectrum separation makes the
interpolation algorithms, such as TDFT, usable. If the interharmonic and harmonic spectra are interfered with, as is shown in
Fig. 3, the details of the interharmonic tones cannot be located
by using any interpolation methods.
B. Practical Issues and the Solutions
To put the proposed method into practice, some practical issues must be discussed and addressed as follows.
1) In practice, due to the limit acquiring resolution of the
instrument and the measurement noise, the spectral bins
around the harmonic frequencies are always nonzero even

The real case involves a 25-kV distribution feeder commonly


seen in rural North America. The VFD is one of the largest loads
supplied by the feeder. Since the VFD had been installed, lighter
flicker problems have been reported due to significant interharmonics.
A measurement was taken at the secondary side of a stepdown transformer connecting the VFD. The sampling frequency
is 11520 Hz. Fig. 5 shows the waveform snapshots and the
12-cycle DFT results. The DFT results show that the 120 Hz
and 240 Hz harmonics are strongly interfered by the nearby interharmonics. Thus, it can be concluded that the IEC 61000-4-7based group scheme will give misleading answers in calculating
the 2nd and 4th harmonic groups and the interharmonic groups
nearby.
Ten continuous 12-cycle waveform snapshots were studied.
By filtering out the negligible interharmonic leakage effect and
noise using (24) ( was chosen as 0.3% in this case), only the

HUI et al.: A METHOD TO IMPROVE THE INTERHARMONIC GROUPING SCHEME ADOPTED BY IEC STANDARD 61000-4-7

977

Fig. 5. Waveforms and spectrum of the measured signal. (a) The 2-s waveform
snapshot. (b) The 0.2-s waveform snapshot. (c) DFT spectrum of the 0.2-s waveform snapshot.
Fig. 7. Calculated interharmonic leakage values at the 4th harmonic frequency.
(a) Magnitudes of the leakage values. (b) Phase angles of the leakage values.

Fig. 6. Calculated interharmonic leakage values at the 2nd harmonic frequency. (a) Magnitudes of the leakage values. (b) Phase angles of the leakage
values.

interharmonic interferences at 120 Hz and 240 Hz were considered and extracted for most of the snapshots. The calculated
interharmonic DFT results at the two harmonic frequencies are
shown in Figs. 6 and 7. These figures show that the three leakage
results by considering three different interharmonic numbers
are consistent and concentrated. This finding not only implies
that the results are correct, but also indicates only one dominant
interharmonic was in close proximity to the nearby harmonic.
Thus, for this case, considering one interharmonic is enough and
credible.
After separating the original DFT spectrum, the harmonic
and interharmonic group results were calculated for each snapshot. In order to verify the accuracy of the calculated results,
the group results of the 2-s waveform snapshot were used as the
references. The mean group magnitudes obtained by using the
three different methods are shown in Fig. 8. For the harmonics,

Fig. 8. Comparison among group results from different methods. (a) Results
of harmonic groups. (b) Results of interharmonic groups.

the IEC standard procedure gives always overestimated results


especially when interharmonics are nearby. The results reach an
overestimation of about 70% for the 2nd harmonic group and of
320% for the 4th harmonic group. In contrast, for the interharmonic groups, the IEC standard procedure gives underestimated
results for the 1.5th and the 3.5th interharmonic group results.
By comparison, the proposed method gives accurate results for
the harmonic and interharmonic groups. It was discovered that
the 1.5th and 3.5th interharmonics are significant.
Furthermore, the magnitudes, frequencies, and phase angles
of the two interharmonic tones near 120 Hz and 240 Hz were
calculated based on the separated interharmonic spectrum and
the TDFT algorithm [10]. The results are shown in Fig. 9. The

978

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 27, NO. 2, APRIL 2012

can remain as recommended by IEC (i.e., 5 Hz), and provide


more accurate information on the harmonics and interharmonics present in the signal. The specific characteristics of the
proposed method are:
1) harmonics and interharmonics can be processed separately
in the frequency domain with the proposed spectrum separation technique, even if the frequency difference between
a harmonic and an interharmonic is less than one frequency
resolution (5 Hz);
2) the new interharmonic grouping scheme can group the
main energy of the interharmonic tones into the correct
groups when the interharmonic is in close proximity with
a harmonic;
3) more accurate signal-processing techniques, such as spectral interpolation, can be performed to obtain the detailed
interharmonic information since the interference from harmonics is strongly restrained.
REFERENCES

Fig. 9. Parameters of the two studied interharmonics. (a) Results of the magnitudes. (b) Results of the frequencies. (c) Results of the phase angles.

most obvious finding is that the frequency and phase angle distances between the 120 Hz tone and the nearby interharmonic
are almost the same as the ones between the 240 Hz tone and
the nearby interharmonic. This phenomenon exactly fits the
index [16]. (The frequency distance between the two main
interharmonics produced by VFD is equal to twice the fundamental frequency.) From this perspective, the interharmonic
answers are correct. Moreover, along with this verification,
other evidence also shows the correctness of the results. In
Fig. 5(a), the fluctuation frequency of the waveform presents a
down and up trend. As is known, for most electrical signals, the
waveform fluctuation frequency is determined mainly by the
frequency distance between the main interharmonics and the
nearby harmonics [17]. Therefore, the interharmonic frequency
should also have the same trend, and the results shown in Fig. 9
are consistent with the waveform fluctuation trend.
V. CONCLUSION
A comprehensive solution to the problems caused by the
leakage and picket fence effects is to select a frequency resolution that is a common divider of the frequencies of all
components in the signal. This would not be practical (among
other reasons) because the window width would become too
large. If the window width being examined becomes too large,
the risk of dealing with nonstationary signals increases. This
paper proposed a method to improve the grouping scheme
adopted by IEC. By using the method, the frequency resolution

[1] C. Li, W. Xu, and T. Tayjasanant, Interharmonics: Basic concepts and


techniques for their detection and measurement, Elect. Power Syst.
Res., vol. 66, pp. 3948, 2003.
[2] T. X. Zhu, Exact harmonics/interharmonics calculation using adaptive
window width, IEEE Trans. Power Del., vol. 22, no. 4, pp. 22792288,
Oct. 2007.
[3] General Guide on Harmonics and Interharmonics Measurements for
Power Supply Systems and Equipment Connected Thereto, IEC Standard 61000-4-7, 2002.
[4] D. Gallo, R. Langella, and A. Testa, Light flicker prediction based on
voltage spectral analysis, presented at the IEEE Power Tech. Conf.,
Porto, Portugal, Sep. 2001.
[5] T. Kim, M. Rylander, E. J. Powers, W. M. Gary, and A. Arapostathis,
LED lamp flicker caused by interharmonics, presented at the IEEE
Instrum. Meas. Technol. Conf., Victoria, BC, Canada, May 2008.
[6] D. Gallo, R. Langella, and A. Testa, Desynchronized processing technique for harmonic and interharmonic analysis, IEEE Trans. Power
Del., vol. 19, no. 3, pp. 9931001, Jul. 2004.
[7] Z. Liu, J. Himmel, and K. W. Bonfig, Improved processing of harmonics and interharmonics by time-domain averaging, IEEE Trans.
Power Del., vol. 20, no. 4, pp. 23702380, Oct. 2005.
[8] G. W. Chang, C. Y. Chen, and M. C. Wu, A modified algorithm for
harmonics and interharmonics measurement, presented at the IEEE
Power Eng. Soc. Gen. Meeting, Tampa, FL, Jun. 2007.
[9] G. Andria, M. Savino, and A. Trotta, Windows and interpolation algorithm to improve electrical measurement accuracy, IEEE Trans. Instrum. Meas., vol. 38, no. 4, pp. 856863, Aug. 1989.
[10] R. Yang and X. Hui, A novel algorithm for accurate frequency measurement using transformed consecutive points of DFT, IEEE Trans.
Power Syst., vol. 23, no. 3, pp. 10571062, Aug. 2008.
[11] B. Zeng, Z. Teng, Y. Cai, S. Guo, and B. Qing, Harmonic phasor
analysis based on improved FFT algorithm, IEEE Trans. Smart Grid,
vol. 2, no. 1, pp. 5159, Mar. 2011.
[12] D. Belega and D. Dallet, High-Accuracy frequency estimation via
weighted multipoint interpolated DFT, Inst. Eng. Technol. Sci., Meas.
Technol., vol. 2, pp. 18, 2008.
[13] G. W. Chang, C. I. Chen, Y. J. Liu, and M. C. Wu, Measuring power
system harmonics and interharmonics by an improved fast Fourier
transform-based algorithm, Inst. Eng. Technol. Gen. Transm. Distrib.,
vol. 2, no. 2, pp. 192201, 2008.
[14] D. Agrez, Weighted multipoint interpolated DFT to improve amplitude estimation of multifrequency signal, IEEE Trans. Instrum. Meas.,
vol. 51, no. 2, pp. 287292, Apr. 2002.
[15] H. Qian, R. Zhao, and T. Chen, Interharmonic analysis based on interpolating windowed FFT algorithm, IEEE Trans. Power Del., vol. 22,
no. 2, pp. 10641069, Apr. 2007.
[16] J. Yong, T. Tayjasanant, W. Xu, and C. Sun, Characterizing voltage
fluctuations caused by a pair of interharmonics, IEEE Trans. Power
Del., vol. 23, no. 1, pp. 319327, Jan. 2008.
[17] L. Tang, D. Hall, M. Samotyj, and J. Randolph, Analysis of DC arc furnace operation and flicker caused by 187 Hz voltage distortion, IEEE
Trans. Power Del., vol. 9, no. 2, pp. 10981107, Apr. 1994.

HUI et al.: A METHOD TO IMPROVE THE INTERHARMONIC GROUPING SCHEME ADOPTED BY IEC STANDARD 61000-4-7

979

Jin Hui was born in Wuxi, China, in 1985. He received the B.S. degree in electrical engineering from the College of Electrical Engineering and Information
Technology, Sichuan University, Chengdu, China, in 2007 and is currently pursuing the Ph.D. degree in electrical engineering at Sichuan University, Chengdu,
China.
His main research interests are power system harmonic analysis and evaluation.

Wilsun Xu (M90SM95F05) received the Ph.D. degree in power engineering from the University of British Columbia, Vancouver, BC, Canada, in
1989.
He was an Engineer with BC Hydro from 1900 to 1996. Currently, he is a
Professor and an ESERC/iCORE Industrial Research Chair at the University of
Alberta, Edmonton, AB, Canada. His research interests are power quality and
harmonics.

Honggeng Yang (M07) was bon in Chengdu, China, in 1949. He received


the M.S. degree in electrical engineering from Harbin Institute of Technology,
Harbin, China, in 1985, and the Ph.D. degree in electrical engineering from
Liege University, Leige, Belgium, in 1997.
Since 2002, he has been a Professor at Sichuan University, Chengdu. His
research interests include power quality and reactive power/voltage control.

Yamei Liu received the M.S. degree from Chengdu University of Science and
Technology, Chengdu, China, in 1999, and is currently pursuing the Ph.D. degree in electrical engineering at Sichuan University, Chengdu.
Her main research interests are power system harmonics and interharmonic
analysis.

You might also like