Impact of Sample Size on ICA-Based Harmonic Source Estimation
E. Gursoy, Dagmar Niebur · 2006
In this paper, we investigate the effect of sample size to the estimation of harmonic sources in electric power systems. A blind source separation technique called independent component analysis (ICA) is used to estimate the load profiles of harmonic sources. The ICA algorithm is based on the statistical independence of loads. Results are obtained for several sample sizes. Simulation results show that ICA is able to estimate the sources without the knowledge of network topology or parameters even for the small sample sizes