Data-Driven Classification for Residential Coincident Peak Demand Contributors Using Actual Power, Sociological, and Meteorological Data

Residential power consumers consume a significant amount of energy in power systems, compared to the industrial and commercial sectors. Due to the volatility and diversity of residential power demand, the residential sector has a substantial impact on the peak demand of power systems. Unlike those of industrial and commercial power consumers, the load profiles of residential consumers are much more diverse and chaotic. Therefore, understanding residential power demand attributes is critical for effective demand response (DR) programs implementation at residential level to alleviate system peak demand. This paper aims to characterize the residential coincident peak contributors and identify potential residential participants for DR programs. A novel approach, the residential coincident peak contributor disaggregation (RCPCD), is developed in this study to determine the target consumers based on their annual peak demand contribution frequencies. The algorithm of K-Means is employed in the developed RCPCD to obtain typical residential seasonal load profiles. Meanwhile, the correlations among power demand, associated sociological data, and seasonal effects are investigated. This study is conducted based on actual AMI and parcel data from a utility partner.