Harmonic separation method based on sparse representation for power system
Liang Fen, Tao Xue-Hua, Fajun Yu · 2016
With the increase of power electronic devices, harmonic pollution in power system is inevitable. It is urgent to separate harmonics in order to improve the power quality. This paper presents a novel harmonic separation method in which the frequency sparsity of harmonic is fully used. First, two channel signals constructed from the original sampled data are transformed to frequency domain with Short-Time Fourier Transform. Then, the mixing matrix is estimated by clustering the points in frequency scatter diagram. Finally, the corresponding harmonic components are separated by the minimal L1 norm decomposition. The simulation results validate the proposed method.