Fundamental Frequency Estimation Using Wavelet Denoising Techniques
Fundamental Frequency Estimation Using Wavelet Denoising Techniques
Fundamental Frequency Estimation Using Wavelet Denoising Techniques
1
interharmonics, which strongly deteriorate the quality of the
Abstract—Harmonic and interharmonic analysis in power power supply voltage. Periodicity intervals in the presence of
systems are usually based on the nominal frequency of 50Hz or interharmonics can be very long. Parameter estimation of the
60Hz, which is an approximation of the fundamental frequency.
components is very important for control and protection tasks.
Many spectrum analyzers consider this frequency constant and
therefore, their measurement is not very accurate. This paper The design of harmonics filters relies on the measurement of
presents a new fundamental frequency estimation method based distortions in both current and voltage waveforms [1].
on an improved zero crossing algorithm. The electric signal is Digital control and protection of power systems require the
first filtered using the wavelet transform and the frequency is estimation of supply frequency and its variation in real-time.
then computed using a method that counts the number of zero Variations in system frequency from its normal value indicate
crossings in time using an adaptive window that detects the false the occurrence of a corrective action for its restoration. A
zero crossings. The paper continues with a virtual instrument large number of numerical methods is available for frequency
that computes the frequency from a sampled signal. Several
estimation from the digitized samples of the system voltage.
computer simulations test results are presented in the paper to
highlight the usefulness of this approach in estimating near Discrete Fourier transforms , Least error squares technique,
nominal power system frequencies. Kalman filtering , Recursive Newton-type algorithm , adaptive
notch filters etc. are known signal processing techniques used
Index Terms—Fundamental Frequency, Harmonics and for frequency measurements of power system signals [2].
Interharmonics, Virtual Instrument, Wavelet Denoising, Zero The real-time performance of a fundamental frequency
Crossings estimation algorithm depends not only on its computational
efficiency but also on its ability to obtain accurate estimates
I. INTRODUCTION from short signal segments.
S PECTRUM estimation of discretely sampled processes is The algorithm proposed in this paper is based on a Discrete
usually based on procedures employing the Fast Fourier Wavelet Transform filter that attenuates the high frequency
harmonics which would create false zero crossings near the
Transform (FFT). This approach is computationally efficient
real one, an adaptive window of search and an algorithm that
and produces reasonable results for a large class of signal
tracks the fundamental frequency and approximates it each
processes. However, there are several performance limitations
period.
of the FFT.
The most prominent limitation is that of frequency resolution, II. WAVELET FILTER DESIGN
i.e. the ability to distinguish the spectral responses of two or
more signals. A second limitation is caused by data A. Discrete Wavelet Transform
windowing, which manifests as leakage in the spectral The Wavelet Series is just a sampled version of Continuous
domain. These performance limitations are particularly Wavelet Transform and its computation may consume
troublesome when analyzing short data records, which occur significant amount of time and resources, depending on the
frequently in practice, because many measured processes are resolution required. The Discrete Wavelet Transform (DWT),
brief. which is based on sub-band coding is found to yield a fast
Modern frequency converters generate a wide spectrum of computation of the Wavelet Transform. It is easy to
harmonic components. Large converter systems and arc implement and reduces the computation time and resources
furnaces can also generate non-characteristic harmonics and required.
Filters are one of the most widely used signal processing
All authors are with the Electrical Power Systems Department, Technical functions. Wavelets can be realized by iteration of filters with
University of Cluj-Napoca, 15 C. Daicoviciu St., RO 400020, Cluj – Napoca. rescaling. The resolution of the signal, which is a measure of
D. Gheorghe and R.B. Vasiliu are PhD students (e-mails: the amount of detail information in the signal, is determined
[email protected], [email protected]), A.
Cziker is an associate professor (e-mail:[email protected]) and
by the filtering operations, and the scale is determined by
M. Chindriş is a full professor (e-mail: [email protected]). upsampling and downsampling (subsampling) operations[3].
The firs author D. Gheorghe received the License Degree in Electrical The DWT is computed by successive lowpass and highpass
Engineering from the Technical University of Cluj-Napoca, Cluj-Napoca, filtering of the discrete time-domain signal as shown in figure
Romania in 2009. Since 2009 he is working towards his Ph.D. at the Electrical
1. This is called the Mallat algorithm or Mallat-tree
Power Systems Department of Technical University of Cluj-Napoca. His
research project is concentrated on studying the power quality parameters. decomposition. Its significance is in the manner it connects
148
3RD INTERNATIONAL CONFERENCE ON MODERN POWER SYSTEMS MPS 2010, 18-21 MAY 2010, CLUJ-NAPOCA, ROMANIA
depends on the length of the signal. The DWT of the original The n indices represent the translation through time will the
signal is then obtained by concatenating all the coefficients, m indices represent the Dilation. The actual wavelets behave
a[n] and d[n], starting from the last level of decomposition. like a band pass filter and the scaling behaves like a low pass
filter.
∞
d (t ) = ∑ Tm,nψ n ,m (t ) (6)
m −∞
∞
a (t ) = ∑ S m,nφn ,m (t ) (7)
m −∞
149
3RD INTERNATIONAL CONFERENCE ON MODERN POWER SYSTEMS MPS 2010, 18-21 MAY 2010, CLUJ-NAPOCA, ROMANIA
Magnitude [V]
200
2
100
200 Voltage Waveform
Magnitude [V]
0
− 100
0 200 400
Time [s]
− 200
Coefficient 2 + 250V
0 0.005 0.01 0.015 0.02 Coefficient 4 +100V
Coefficient 5
Time [s]
Fig. 3. Voltage Waveform Fig. 6. Discrete Wavelet Transform coefficients (2, 4 and 5)
Original Waveform
200 Filtered Waveform
Magnitude [V]
Where S is the scaling function and W is the wavelet 0 0.005 0.01 0.015 0.02
150
3RD INTERNATIONAL CONFERENCE ON MODERN POWER SYSTEMS MPS 2010, 18-21 MAY 2010, CLUJ-NAPOCA, ROMANIA
Fig. 11. Virtual instrument front panel. The wavelet filter is activated.
Fig. 10. First zero crossing and the adaptive search window
151
3RD INTERNATIONAL CONFERENCE ON MODERN POWER SYSTEMS MPS 2010, 18-21 MAY 2010, CLUJ-NAPOCA, ROMANIA
TABLE I
FREQUENCY ESTIMATION RESULTS
152