Finding Load Profile Primitives by using Independent Component Analysis
Tadahiro Taniguchi, Shiro Yano, Haruka Kobayashi · 2013
In this paper, we describe our work on finding load profile primitives. Analyzing home electricity consumption is an important issue for designing, managing, and analyzing a future smart grid. Load profile data have additive property because of the same property of electric current. It will enable us to decompose observed data into a small number of elemental components, named load profile primitives. We measured several sets of home load profiles and applied independent component analysis (ICA) to extract load profile primitives. We investigate the possibility of finding load profile primitives by analyzing the measured data set with independent component scores. We evaluate its performance by comparing with originally observed data, qualitatively. We also adopted ICA to load profiles of one hundred houses measured in Higashiomi city and shows that the method can extract the seasonal changes in load profiles of local grid.