Data segmentation algorithms for a time-domain harmonic source modeling method

Nowadays, with a fast increasing in renewable energy applications and vast deployment of plug-in electric vehicle, power quality issue becomes more important than before because of considerable harmonic current injected into grid. To better manage these distributed energy sources/loads and reduce harmonic pollution, establishing more accurate harmonic source models for them becomes very necessary. Traditionally, harmonic source is described as a fixed-parameter model, which is not good enough to describe time-varying harmonic source. In this paper, a time-domain modeling method is firstly explained, which enables the study on time-varying harmonic source model. Then two different data segmentation algorithms to support this model based on the stability of current magnitude and phase angle are developed and explained respectively using a group of data acquired from field. One is developed from statistical perspective while the other is based on slope of curve. Comparison is presented to explain their unique characteristics and synthesizing of both algorithms is also illustrated. The results of algorithms are shown to be satisfactory.