Modern spectrum estimation methods can provide in this case more reliable results of investigation because of high-accuracy even for short data lengths, immunity to noise and independence of accuracy on length of analysis window and on other signal parameters (1 - 3). Abstract - Spectral representation of waveform distor- tions has been of great interest in power systems for many years since the disturbing loads generate a wide spectrum of components which deteriorate the quality of power supply. When spectral components are time-varying in amplitude and/or in frequency - such as in presence of arc furnaces and adjustable speed drives - high resolution methods should be applied due to their capability of cap- ture these changes in time. In this paper the Prony and ESPRIT high-resolution methods are used to analyse simulated non-stationary waveforms in a supply system of a DC arc furnace. The analysis has been completed before and after the inclusion in the system of several compensa- tion devices (STATCOM, Active Filter, Passive Filter and a hybrid of STATCOM and Active Filter). The main con- clusion of the paper is that Prony and ESPRIT methods are useful tools also for the control system design; in fact, they allow capturing time variations of spectra compo- nents better than most usual techniques such as Fourier- based methods. Current and voltage waveforms are obtained from a simulated dc arc furnace plant with and without com- pensation devices, such as: passive filters, active filters, static synchronous compensator (STATCOM) and a hybrid of active filter and STATCOM (4). Analysis of current and voltage waveforms is carried out using the Prony model and ESPRIT method. All above methods are parametric high-resolution spectral methods, which can provide time-varying values of amplitude and/or frequency of spectral components. In the paper, numerical applications of these methods are presented in order to illustrate their implementation problems and practical applications. To investigate the methods, several experiments are performed using non- stationary voltage and current waveforms in a supply system of a dc arc furnace with different compensation devices. Conclusions and recommendations are formu- lated concerning each of the compensation and investi- gation methods.
[1]
Tadeusz Lobos,et al.
Advanced spectrum estimation methods for signal analysis in power electronics
,
2003,
IEEE Trans. Ind. Electron..
[2]
Athina P. Petropulu,et al.
HARMONIC DECOMPOSITION OF TRANSIENT DISTURBANCES USING THE LS PRONY AND ESPRIT-BASED METHODS
,
2002
.
[3]
Zbigniew Leonowicz,et al.
WAVEFORM DISTORTIONS DUE TO AC/DC CONVERTERS FEEDING DC ARC FURNACES
,
2007
.
[4]
Gordon K. Smyth,et al.
A Modified Prony Algorithm for Exponential Function Fitting
,
1995,
SIAM J. Sci. Comput..
[5]
Tadeusz Lobos,et al.
Harmonics and interharmonics estimation using advanced signal processing methods
,
2000,
Ninth International Conference on Harmonics and Quality of Power. Proceedings (Cat. No.00EX441).
[6]
G. Carpinelli,et al.
DC Arc Furnaces: Comparison of Arc Models to Evaluate Waveform Distortion and Voltage Fluctuations
,
2001
.
[7]
Guido Carpinelli,et al.
Comparison of some active devices for the compensation of DC arc furnaces
,
2003,
2003 IEEE Bologna Power Tech Conference Proceedings,.
[8]
Charles W. Therrien,et al.
Discrete Random Signals and Statistical Signal Processing
,
1992
.
[9]
Thomas Kailath,et al.
ESPRIT-estimation of signal parameters via rotational invariance techniques
,
1989,
IEEE Trans. Acoust. Speech Signal Process..