A Valid Data Selection Method in Estimating Harmonic Impact of Individual Loads

MENG Siyu,XIAO Xianyong,ZHANG Yi,HUANG Yong,CHEN Feiyu
DOI: https://doi.org/10.13335/j.1000-3673.pst.2016.2150
2017-01-01
Abstract:Considering possibly existing problem of shortage of valid data sets when estimating impact of individual loads in multiple harmonic source system, a valid data selection method was proposed. In selected data, only one of harmonic loads changed. Firstly, improved k-means clustering method was adopted in a step-by-step way to obtain valid data sets with amplitude and phase angle as two properties of clustering objects. Then contribution of each harmonic source in every cluster was calculated with partial least-squares regression. Final harmonic contribution during the period was obtained by calculating weighted summation of harmonic contribution of every cluster. The proposed method overcomes difficulty of obtaining intersection of several valid data sets. Simulation in IEEE 14-bus system demonstrates accurate performance of the proposed method and its superiority over traditional data selection methods.
What problem does this paper attempt to address?