Clustering analysis of residential electricity demand profiles

摘要: Abstract Little is known about variations in electricity use at finely-resolved timescales, or the drivers for those variations. Using measured data from 103 homes Austin, TX, this analysis sought to (1) determine shape of seasonally-resolved residential demand profiles, (2) optimal number normalized representative profiles within each season, and (3) draw correlations different based on survey occupants homes. Within with similar hourly patterns were clustered into groups using k-means clustering algorithm. Then probit regression was performed if homeowner responses could serve as predictors results. This found that Austin fall one two seasonal some more expensive (from a wholesale market perspective) than others. Regression results indicate variables such someone works home, hours television watched per week, education levels have significant average profile shape, but might vary across seasons. The herein also policies time-of-use real-time structures be likely affect lower income households during high parts year.