a new method for measurement of harmonic groups in power systems using wavelet analysis in the iec...
TRANSCRIPT
-
7/27/2019 A New Method for Measurement of Harmonic Groups in Power Systems Using Wavelet Analysis in the IEC Standar
1/9
Electric Power Systems Research 76 (2006) 200208
A new method for measurement of harmonic groups in power systemsusing wavelet analysis in the IEC standard framework
Julio Barros , Ramon I. Diego
Department of Electronics and Computers, University of Cantabria, Santander, Spain
Received 23 December 2004; received in revised form 20 April 2005; accepted 12 June 2005
Available online 2 September 2005
Abstract
The paper presents a new method based on the wavelet-packet transform for the analysis of harmonic groups in power systems, compatible
with the standards of the International Electrotechnical Commission. The paper studies the performance of the new method in the analysis
of harmonic groups and the spectral leakage associated with the presence of interharmonic components, synchronized and non-synchronized
with the fundamental component, and also due to the loss of synchronization between the sampling window width and the fundamental
frequency. Finally, the results obtained in the application of this method to different test waveforms are compared with the results obtained
using the IEC method.
2005 Elsevier B.V. All rights reserved.
Keywords: Harmonic distortion; Power quality; Wavelets
1. Introduction
Since the beginning of the distribution of electricity, the
harmonic distortion has been one of the most widely studied
disturbances in voltage and current waveforms. The study of
the effect of the harmonic distortion has lead to the develop-
ment of standards to limit its magnitude in order to prevent
damage on equipment and on the power system itself. The
interest both, in the development of measurement methods
and in the study of the levels of harmonic distortion in dis-
tribution networks, has prompted interest in other aspects of
power quality to arrive at the present situation. IEEE stan-
dard 519-1992 [1] and IEC standard 61000-4-7 [2] are the
main international standards for measurement and analysis ofharmonics in power systems. In these standards the Fourier
analysis is proposed as the signal-processing tool to obtain
the harmonic components in voltage and current waveforms.
At present, two main factors affect the harmonic distortion
in power supply systems: first, the growing use of non-linear
loads that are increasing the level of harmonic distortion up
Corresponding author.
E-mail address: [email protected] (J. Barros).
to near to the limits defined in the standards, and second, the
increasing use of non-synchronously pulsating loads with thefundamental power system frequency produce interharmonic
components that are covering the spectra with new frequency
components. These new frequency components are not accu-
rately computed using the current processing methods.
To take into account the changing situation, the IEC has
defined a new method for harmonic and interharmonic anal-
ysis in the second edition of standard IEC 61000-4-7. The
main contribution of this standard is the definition of the
harmonic and interharmonic groups and sub-groups where
different frequency components are computed as single mag-
nitudes in order to have a more accurate representation than
the oneutilizedup to now using singleharmonic components.The standard proposes the use of the Fourier analysis as the
processing tool for the reference instrument for harmonic and
interharmonic analysis, but the standard itself states that the
use of the Fourier analysis does not preclude the application
of other analysis principles.
Other processing tools have been proposed in the litera-
ture for the analysis of harmonics. Kalman filters can be used
for measurement and tracking of power system harmonics
[3,4]. The correct application of this method requires
0378-7796/$ see front matter 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.epsr.2005.06.004
-
7/27/2019 A New Method for Measurement of Harmonic Groups in Power Systems Using Wavelet Analysis in the IEC Standar
2/9
-
7/27/2019 A New Method for Measurement of Harmonic Groups in Power Systems Using Wavelet Analysis in the IEC Standar
3/9
202 J. Barros, R.I. Diego / Electric Power Systems Research 76 (2006) 200208
Fig. 2. Wavelet-packet decomposition tree and grouping of output bands
compatible with the harmonic groups defined by the IEC.
to 3.2 kHz, one additional level must be added to the wavelet
decomposition tree ofFig. 2 and double the number of har-
monic groups (up to 30th order) that could be computed in
the input signal.Theuse of theWPT methodproposedin this paper directly
computes all the frequency components included into uni-
form frequencybandsthat correspond to the harmonic groups
of the signal as defined by the IEC. Using the WPT method
the single line harmonic components of the signal cannot be
obtained.
4. Measurement of harmonics and spectral leakage
In order to evaluate the performance of the method pro-
posed in the measurement of harmonics, Table 1 shows the
rms value of each of the grouped output frequency bands of
Fig. 2 when a 1 p.u. single tone at fundamental frequency
and at odd harmonic frequencies from 3rd to 15th orderare introduced as the input signal. The rms value of each
grouped output frequency band (coefficients d1(n) to d15(n))
is obtained using the square root of the mean square of the
coefficientsof thetwo sub-bandsincludedin each band, using
the method proposed in [5].
As can be seen from the results reported in Table 1,
the spectral leakage is higher in the output sub-bands at
the center of the decomposition tree, coefficients d7(n) and
d9(n), respectively. To accurately characterize the frequency
response of the filter bank selected and to compare the perfor-
mance of the method proposed with the IEC method, a single
tone of 1 p.u. in a range from 1 to 800 Hz, in steps of 1 Hz,
has been introduced as the input signal in Fig. 2 and has beenused to compute the harmonic groups from 1st to 15th order
using the IEC method. Fig. 3a and b shows, respectively, the
results obtained using both methods.
As can be seen in Fig. 3a, the frequency response of each
sub-band is different and depends on the type of decomposi-
tion tree and the type and the sequence of filters (LP or HP)
through which the input signal goes. On the other hand, the
frequency response of each harmonic group is the same using
the IEC method, therefore, it is necessary to compare the fre-
quency characteristics of each group separately to assess the
performance of both methods. As an example, Fig. 4 repre-
sents the frequency response of thefirst order harmonic groupusing both methods.
Whenthetoneisintherangefrom25to75Hzthatisinside
the first order harmonic group, the results obtained using the
IEC method have no error when the tone is synchronous with
the window width used that is when the frequency of the tone
is a multiple of 5 Hz. Otherwise, the error in the estimation
Table 1
rms values of the output bands of the wavelet decomposition tree when a 1 p.u. single tone at odd harmonic frequencies is used as the input signal
50 Hz 150 Hz 250 Hz 350 Hz 450 Hz 550 Hz 650 Hz 750 Hz
d1(n): 2575 Hz 0.9916 0.0541 0.0277 0.0176 0.0117 0.0076 0.0043 0.0014
d2(n): 75125 Hz 0.0579 0.0843 0.0300 0.0179 0.0117 0.0075 0.0042 0.0014d3(n): 125175 Hz 0.0205 0.9876 0.0792 0.0207 0.0130 0.0082 0.0046 0.0015
d4(n): 175225 Hz 0.0098 0.0703 0.0806 0.0258 0.0146 0.0088 0.0048 0.0015
d5(n): 225275 Hz 0.0080 0.0881 0.9880 0.0393 0.0189 0.0117 0.0060 0.0019
d6(n): 275325 Hz 0.0047 0.0181 0.0814 0.0935 0.0304 0.0122 0.0063 0.0020
d7(n): 325375 Hz 0.0025 0.0090 0.0263 0.9318 0.3253 0.0074 0.0040 0.0013
d8(n): 375425 Hz 0.0029 0.0099 0.0225 0.0895 0.0896 0.0225 0.0099 0.0029
d9(n): 425475 Hz 0.0031 0.1050 0.0237 0.3315 0.9346 0.0360 0.0136 0.0039
d10(n): 475525 Hz 0.0022 0.0070 0.0147 0.0234 0.0854 0.0858 0.0184 0.0048
d11(n): 525575 Hz 0.0017 0.0054 0.0103 0.0178 0.0369 0.9861 0.0833 0.0071
d12(n): 575625 Hz 0.0017 0.0053 0.0096 0.0159 0.0283 0.0948 0.0782 0.1060
d13(n): 625675 Hz 0.0015 0.0046 0.0082 0.0131 0.0209 0.0825 0.9875 0.0217
d14(n): 675725 Hz 0.0013 0.0041 0.0072 0.0113 0.0174 0.0293 0.0831 0.0505
d15(n): 725775 Hz 0.0014 0.0043 0.0077 0.0118 0.0177 0.0279 0.0536 0.9915
-
7/27/2019 A New Method for Measurement of Harmonic Groups in Power Systems Using Wavelet Analysis in the IEC Standar
4/9
J. Barros, R.I. Diego / Electric Power Systems Research 76 (2 006) 2002 08 203
Fig. 3. Frequency characteristics: (a) WPT method and (b) IEC method.
Fig. 4. Frequency response of the first order harmonic group using the IEC
method and the WPT method.
of the harmonic group is higher using the IEC method than
the one obtained using the WPT method. If the frequency
of the tone is outside the frequency range of the first order
harmonic group, the spectral leakage is null using the IEC
method, when the tone is again synchronous with the window
width, but the error is higher than the one obtained using the
WPT algorithm when the frequency tone is desynchronizedwith the window width.
The results obtained from the comparison of the other har-
monic groups are similar, except in the case of the harmonic
groups at the center of the decomposition tree, where the per-
formance of the wavelet method is worse than that obtained
using the IEC method. The use of filters with better roll-off
characteristics or the use of scaling factors to compensate
the distortion introduced by the filter bank could improve the
performance of the WPT method proposed [810].
The spectral leakage can also be produced when there
is an error in synchronizing the fundamental power system
frequency and the time window used in the measurement sys-
tem. According to IEC 61000-4-7, the maximum permissibledesynchronization error for a harmonic measurement instru-
ment is 0.03% and this requirement should be fulfilled for
measurement within a range of at least 5% of the nomi-
nal system frequency. The loss of synchronisation should be
indicated by the instrument and the data so acquired should
be flagged.
In order to specify the performance of the WPT method
proposed in the case of failure of synchronization, as is
required in standard 61000-4-7, Fig. 5 represents the error in
percentage, in the determination of the magnitude of the fun-
damental component and harmonics of third and fifth order,
when the IEC and the WPT methods are applied to a 1 p.u.single tone in a range form 49.5 to 50.5 Hz (5% of the
nominal system frequency and the frequency range variation
accepted in the European standards). As can be seen from
the results reported in Fig. 5, the spectral leakage produced
Fig. 5. Error in themagnitude of fundamental,thirdand fifth order harmonic
groups in the case of synchronization error using the IEC method and the
WPT method.
-
7/27/2019 A New Method for Measurement of Harmonic Groups in Power Systems Using Wavelet Analysis in the IEC Standar
5/9
204 J. Barros, R.I. Diego / Electric Power Systems Research 76 (2006) 200208
using the IEC method is a function of the magnitude of the
synchronization error and of the harmonic order, whereas the
error in the determination of the magnitude of the fundamen-
taland harmonic components of third andfifth order using the
WPT method is in all cases 0%, showing that this method is
insensitive to the fluctuations of the power system frequency
in the range studied.
5. Comparative study of the WPT and the IEC
methods
In this section a comparative study of the performance of
the two methods, the WPT method and the IEC method, is
done using different test waveforms. Firstly, the two methods
are compared using stationary signals with different har-
monic components and secondly, the comparative study is
done using non-stationary signals and interharmonic com-
ponents synchronized and non-synchronized with the funda-
mental component.
5.1. Comparative study using stationary signals
In order to do this study, we haveused the daily mean mag-
nitude of the harmonic distortion measured in our low voltage
distribution system as the test signal, which is represented in
Table 2 under column input signal.
Table 2 reports the results obtained using this signal as the
input signal in both, the WPT method and the IEC method.
The spectral leakage caused by the filtering characteristics
of the WPT gives very large errors, while the IEC method
gives the exact values. After different simulations we haveverified that the spectral leakage produced using the WPT
method is mainly due to the fundamental component. The
spectral leakage produced by the rest of the harmonic com-
ponents in the neighboring harmonic groups is very small
due to their small magnitude. To compensate the effect of the
fundamental component in the spectral leakage we propose
the following two-stage process: a first stage where the fun-
damental component is estimated and a second stage where
the fundamental component is filtered out in the input sig-
nal and the WPT method is applied to the resultant signal to
compute the rest of the harmonic components. The resultsobtained using this two-stage process are reported in Table 2,
under column filtering + WPT. As can be seen in Table 2,
the errors in the estimation of the harmonic groups from 2nd
to 15th order are very small and they are within the accepted
range for a harmonics measurement instrument.
5.2. Comparative study using fluctuating harmonics and
interharmonic components
In this section a comparative study of the performance
of the two methods is done using different test waveforms
proposed in standard IEC 61000-4-7. The precision in the
determination of the magnitude of the harmonic componentsusing both methods is compared with the total rms value of
each test waveform calculated over 10 cycles of the funda-
mental frequency.
The following case studies are considered: Examples 1
and 2 study the case of fluctuating harmonics. In Example 1,
a large harmonic current fluctuation is considered, whereas
Example2 considersthe case of a harmonic current controlled
by the zero-crossing multicycle method. Examples 35 study
two possible interharmonic producing conditions in a power
system: the case of a communication signal connected to the
power system with a frequency non-integer multiple of the
fundamental frequency not-coexisting and coexisting withdifferent harmonic components (Examples 3 and 4) and the
case of a voltage harmonic with a sinusoidal voltage modula-
tion, as produced by an electronic motor drive with a varying
torque (Example 5).
Table 2
Magnitude of the 15 harmonic groups obtained using the IEC and the WPT methods for a stationary signal corresponding to daily mean magnitude of the
harmonic distortion measured in our low voltage distribution system
Harmonic group Input signal IEC method WPT method Filtering + WPT
V % V % V % V %
1 230 100 230 100 228.0705 99.1611 0.2900 0.1261
2 0 0 0 0 13.0247 5.6629 0.3392 0.14753 1.15 0.50 1.15 0.50 4.5445 1.9759 1.2955 0.5632
4 0 0 0 0 1.8192 0.7910 0.5116 0.2224
5 6.21 2.70 6.21 2.70 6.2541 2.7192 6.1241 2.6626
6 0 0 0 0 1.3703 0.5958 0.3541 0.1539
7 2.53 1.10 2.53 1.10 2.5929 1.1273 2.5441 1.1061
8 0 0 0 0 0.9832 0.4275 0.3165 0.1376
9 0.69 0.30 0.69 0.30 1.0917 0.4746 0.5084 0.2211
10 0 0 0 0 0.6742 0.2931 0.1759 0.0765
11 0.46 0.20 0.46 0.20 0.7095 0.3085 0.4839 0.2104
12 0 0 0 0 0.5538 0.2408 0.1676 0.0729
13 0 0 0 0 0.4642 0.2018 0.1268 0.0551
14 0 0 0 0 0.4084 0.1776 0.1038 0.0451
15 0 0 0 0 0.4306 0.1872 0.1076 0.0468
-
7/27/2019 A New Method for Measurement of Harmonic Groups in Power Systems Using Wavelet Analysis in the IEC Standar
6/9
J. Barros, R.I. Diego / Electric Power Systems Research 76 (2 006) 2002 08 205
Fig. 6. Large fifth harmonic current fluctuation.
Fig.7. Spectral components of the waveformofFig. 6 using theIEC method.
Example 1. Fig. 6 shows the case of rms fifth harmonic cur-
rent fluctuating from 3.536 to 0.7071 A and Fig. 7 shows the
corresponding spectrum using the IEC method. The change
in the magnitude of the current occurs after 21.25 periods of
the fifth harmonic. The total rms value of the time function
calculated over a time interval of 0.2 s is 2.367 A. Table 3
Fig.8. Waveformof third harmoniccurrentcontrolled by zero-crossingmul-
ticycle method.
shows the results obtained in the measurement of the 15 har-
monic groups in the current waveform using the IEC and the
WPT methods.
The magnitude in the estimation of the fifth order har-
monic group using the IEC method is 2.34 A, with an error
of 1.14%. On the other hand, the estimation obtained usingthe WPT method is 2.3267 A. In this case, the error is 1.70%,
slightly higher than the error obtained with the IEC method
but in the range acceptable for a measurement instrument.
The spectral leakage in the rest of harmonic groups is similar
using both methods.
Example 2. Fig. 8 shows the typical waveform of a third
harmonic current produced by a microwave appliance. The
average power is controlled by the zero-crossing multicycle
method with, in this case, a repetition rate of 5 Hz and a duty-
cycle of 50%. The total rms current calculated over 0.2 s is
0.707 A. Fig. 9 shows the corresponding spectrum obtained
using the IEC method. The magnitude of harmonic groupsfrom 1st to 15th order calculated using the IEC and the WPT
methods is reported in Table 3.
Using the IEC method the estimation of rms value of the
third harmonic group is 0.6925 A, with an error of 2.05%.
Table 3
rms values of the 15 harmonic groups using the IEC and the WPT methods for the test waveforms of Examples 15
Harmonic group order Example 1 Example 2 Example 3 Example 4 Example 5
WPT IEC WPT IEC WPT IEC WPT IEC WPT IEC
1 0.0763 0.0629 0.0537 0.0531 0.2490 0.2320 1.3902 1.3830 0.2506 0.0388
2 0.0863 0.0753 0.0831 0.0957 0.2562 0.2689 1.8836 1.9051 0.2721 0.04513 0.1806 0.1094 0.6913 0.6925 0.2931 0.3514 14.1416 15.2743 0.7862 0.0616
4 0.2718 0.2357 0.0713 0.0753 0.4477 0.5568 20.9453 20.2313 0.7352 0.1349
5 2.3267 2.3400 0.0731 0.0294 1.4916 1.7360 11.8734 11.7455 10.166 10.264
6 0.2646 0.2353 0.0183 0.0177 9.6415 9.5459 1.2804 0.6654 0.7235 0.1227
7 0.1116 0.1079 0.0090 0.0123 0.4606 0.8376 0.5668 0.4098 0.2386 0.0479
8 0.0803 0.0717 0.0099 0.0094 0.2825 0.4560 0.4960 0.2781 0.2036 0.0302
9 0.0645 0.0555 0.0105 0.0076 0.2977 0.3262 0.4969 0.1987 0.2150 0.0221
10 0.0489 0.0469 0.0071 0.0064 0.4143 0.2614 0.3434 0.1458 0.1338 0.0176
11 0.0397 0.0410 0.0054 0.0056 0.0975 0.2235 0.2619 0.1081 0.0931 0.0148
12 0.0401 0.0368 0.0053 0.0051 0.0876 0.1997 0.2510 0.0796 0.0869 0.0130
13 0.0365 0.0344 0.0046 0.0047 0.0737 0.1844 0.2332 0.0569 0.0749 0.0119
14 0.0352 0.0331 0.0040 0.0044 0.0624 0.1748 0.1997 0.0385 0.0657 0.0111
15 0.0350 0.0320 0.0040 0.0043 0.0637 0.1695 0.2118 0.0239 0.0695 0.0107
-
7/27/2019 A New Method for Measurement of Harmonic Groups in Power Systems Using Wavelet Analysis in the IEC Standar
7/9
206 J. Barros, R.I. Diego / Electric Power Systems Research 76 (2006) 200208
Fig.9. Spectral components of thewaveform ofFig. 8 usingthe IEC method.
The estimation of the rms magnitude of the third harmonicgroup obtained using the WPT method is 0.6913 A. The error
in this case is 2.22% once again slightly higher than that
obtained using the IEC method but in a range acceptable for
a measurement instrument. The spectral leakage in the rest
of harmonic groups is similar using both methods.
Example 3. In this case a communication signal of 9.8 V at
287Hz is considered. Fig. 10 shows the waveform of this sig-
nal and Fig. 11 shows the corresponding spectrum obtained
using the IEC method. Table 3 shows the magnitude of the
harmonic groups obtained using the IEC and the WPT meth-
ods.
The magnitude of the sixth harmonic group using the IEC
method is 9.5459 and 9.6415 V using the WPT method. The
error incurred using the IEC method is 2.59% whereas the
error observed using the WPT method is only 1.61%. The
spectral leakage in the rest of harmonic groups is higher using
the IEC method than the WPT method.
Example 4. Fig. 12 shows the waveform of a communi-
cation signal of an interharmonic of 178 Hz with constant
magnitude of 23 V rms superimposed on a third and fifth
harmonic of 11.5 V each. Fig. 13 shows the corresponding
spectrum obtained using the IEC method and Table 3 shows
Fig. 10. Interharmonic of 9.8V at 287 Hz.
Fig. 11. Spectral components of the waveform of Fig. 10 using the IEC
method.
Fig. 12. Thirdandfifthharmonicwitha interharmonicof 23V rmsat 178Hz.
the magnitude of the harmonic groups obtained using bothmethods.
The magnitude of the fourth order harmonic group is
20.2313 V using the IEC method and 20.9453 V using the
Fig. 13. Spectral components of the waveform of Fig. 12 using the IEC
method.
-
7/27/2019 A New Method for Measurement of Harmonic Groups in Power Systems Using Wavelet Analysis in the IEC Standar
8/9
-
7/27/2019 A New Method for Measurement of Harmonic Groups in Power Systems Using Wavelet Analysis in the IEC Standar
9/9
208 J. Barros, R.I. Diego / Electric Power Systems Research 76 (2006) 200208
[10] V.L. Pham, K.P. Wong, Antidistortion method for wavelet trans-
form filter banks and nonstationay power system waveform harmonic
analysis, in: IEE Proceedings on Generation, Transmission and Dis-
tribution, vol. 148, 2001, pp. 117122.
Julio Barros received the M.Sc. and Ph.D. degrees in Physics in 1978 and
1989, respectively, both from the University of Cantabria, Spain. In 1989,
he joined the Department of Electronics and Computers of the University
of Cantabria, where he is currently an associate professor. His research
areas are real-time computer applications in power systems, harmonics
and power quality.
Ramon I. Diego was born in Santander (Spain) in 1973. He received
the M.Sc. degree in physics in 2000 from the University of Cantabria,
Spain. His main research interests include electromagnetic compatibility
and digital signal processing applied to power quality. He is currently
working on his Ph.D. thesis.