Harmonic Source Region Location Based on Fast ICA and Scene Clustering

Abstract

The widespread application of power electronic devices and renewable power generation has resulted in significant harmonic pollution. Harmonic source location is of great importance since it is a prerequisite to mitigate harmonic pollution. In this paper, a scene-based harmonic source region location method is proposed. The proposed method employs the fast independent component analysis (ICA) to locate the harmonic source region as the harmonic measurement matrix is unknown and the measuring equation is not observable. In this way, the range of suspect harmonic sources is narrowed and the cost of harmonic source location is reduced. Furthermore, the improved K-means algorithm is adopted to divide the measurement data into several scenes to improve location accuracy. Finally, a case study is conducted to verify the validity and accuracy of the proposed method.