Investigation of Supra-Harmonics through Signal Processing Methods in Smart Grids

Nowadays supra-harmonic distortion studies are gaining attention day by day in power quality research area. When handling communication systems especially Power Line Carrier (PLC) systems in frequency range 2150 kHz, they are suitable for causing electromagnetic interference (EMI) to other systems. This study shows results of analysis employing advanced method called ensemble empirical mode decomposition (EEMD) to describe supra-harmonic distortion. Unlike the traditional method (short time fourier transformSTFT), EEMD gives extensive representation for supra-harmonic components


I. INTRODUCTION
A novel and significance increasing day by day hazardous risk to smart grid systems called supraharmonics or emissions in 2 kHz -150 kHz frequency band.This threat can affect capacitors, lose communication contacts with smart meters.The important single fault source operations from photovoltaic inverters (PVs).Naturally fuel cells, battery chargers also wind turbines, can produce this serious threat [1,2,3].Supraharmonics also disturb domestic appliances, semi-conductor manufacturing devices, medical equipments, security systems even transportation controls.PLC produces low-impedance for emissions in subharmonics frequency range.The highest levels are commonly by virtue of PLC.As a result, power grids are worked out to transfer power at 50 Hz however, line also carries 2 kHz-150 kHz electromagnetic components [4,5,6,7,17].In this study, National Instruments PWr cRIO Data Recorder was used to acquire distortions in power systems.Sampling frequency was selected 1 MHz for measurements.Table I. gives the information of PV systems components properties.

A. Short Time Fourier Transform
Traditionally STFT is applied to measured data from domestic appliances and Sunny Mini Central PV inverter.The results from the STFT are presented in a spectrogram.Spectrograms used for signal processing owing to show supraharmonics.STFT has drawbacks about representation magnitude and frequency bands [8].In algorithm we used STFT with hamming sliding window (5 ms) for decompositon. .

B. Ensemble Emprical Mode Decomposition (EEMD)
EMD has been profitably performed for non-stationary signal processing.The EMD could decompose the complicated signal function into a number of Intrinsic Mode Functions (IMFs) [9,10,11].The algorithm has major drawbacks of mode mixing, end effects and etc [12,13,14,15,16].Therefore, in this work we performed EEMD method for generating IMFs in order to analyze supra-harmonics.We focused on pattern frequency band which is dominated in power grid.
The mathematical background of EEMD algorithm (Fig 3 ) tn: trial numbers, i: iteration number and j: imf scale [13,14,15].In figure 5 is illustrated continuous and fluctuant frequency emission at 15 kHz and 44.67 kHz bands.Figure 6 shows the frequency spectrum of IMF 3-IMF 5.When considering the spectrum it is comprehensively shown that IMF 3 represents the frequency component (15.14 kHz).Furthermore, IMF 4 also shows frequency (44.92 kHz) band.Table II.shows the relationship between signal and the IMFs components.IMFs are sorted from higher frequency to lower.Algorithm routine stops till the end of getting monotonic function IMF 9 called residual IMF 7 (R=0.9998)strongly related with the signal.

III. CONCLUSIONS
Instead of traditional methods such as STFT, EEMD gives more accurate results, determining the PV inverter and PLC frequency bands with high exactitude.It was shown that EEMD method can be used for spectral analysis of supra-harmonics and can be also applied for pattern recognition of supra-harmonics in smart grids with PV systems.With the help of the proposed method supraharmonic analysis and pattern detection of them easily inquire into.
For future study, this results will discuss with another signal processing methods.Filter design for measuring supra-harmonics in smart grids will be investigated in the light of the analysis.
Fig 1 shows PV System at the Faculty of Electrical Engineering Wroclaw University of Technology.

Fig 1 .
Fig 1. Photovoltaic System at the Faculty of Electrical Engineering Wroclaw University of Technology Respectively fig 2 and 5 shows STFT spectrograms and 2 Hz -120 kHz frequency band for current signals of domestic appliances LCD TV and Laser Printer.

Fig 2 .
Fig 2. Spectrogram of the current LCD TV In figure 2 is illustrated constant or continuous frequency emissions at 17. 6 kHz, 53 kHz and 88 kHz bands.

Fig 5 .
Fig 5. Spectrogram of the current Laser Printer

TABLE I .
PV

TABLE II .
CORRELATION COEFFICIENT BETWEEN SIGNAL AND IMFS