Predicting residential electricity consumption patterns based on smart meter and household data: A case study from the Republic of Ireland. (December 2022)

• Record Type: Journal Article
• Authors: Guo, Zhifeng; O'Hanley, Jesse R.; Gibson, Stuart
• Is Part Of: Utilities policy. Volume 79(2022)
• Publication Date: 2022-12
• Subjects: Residential electricity consumption -- Household load profiles -- Machine learning; Public utilities -- United States -- Periodicals; Public utilities -- Government policy -- United States -- Periodicals; 363.6
• DOI: 10.1016/j.jup.2022.101446
• Languages: English
• ISSNs: 0957-1787

Abstract

Abstract: We use machine learning algorithms to investigate various aspects of residential electricity consumption for households in the Republic of Ireland. Temperature, day of week, and month of year have an apparent causal effect on consumption. The prevalence of six distinct intra-day load profiles, identified by clustering, changes dramatically between weekdays and weekends as well as seasonally. Key socio-demographic and dwelling characteristics associated with annual load profiles include household makeup and size and occupation of the primary income earner. We further discuss policy and management implications of our findings and propose avenues for future research. Highlights: Temperature, day of week, and seasonal strongly affect household electricity usage. Clustering reveals distinct intra-day load profiles. Prevalence of intra-day load profiles varies over the week and by season. Annual load profiles are associated with distinct mixes of intra-day patterns. Electricity use is tied to household level characteristics and historical usage.