Cross-domain feature selection and coding for household energy behavior. (15th July 2016)

• Record Type: Journal Article
• Authors: Tong, Xing; Li, Ran; Li, Furong; Kang, Chongqing
• Is Part Of: Energy. Volume 107(2016)
• Page Start: 9
• Page End: 16
• Publication Date: 2016-07-15
• Subjects: Household energy behavior -- Demographic factors -- Customer classification -- Feature selection and coding; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2016.03.135
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: Household energy behavior is a key factor that dictates energy consumption, efficiency and conservation. In the past, household energy behavior was typically unknown because conventional meters only recorded the total amount of energy consumed for a household over a significant period of time. The rollout of smart meters enabled real-time household energy consumption to be recorded and analyzed. This paper uses smart meter readings from more than 5000 Irish households to identify energy behavior indicators through a cross-domain feature selection and coding approach. The idea is to extract and connect customers' features from energy domain and demography domain, i.e., smart metering data and household information. Smart metering data are characterized by typical energy spectral patterns, whereas household information is encoded as the energy behavior indicator. The results show that employment status and internet usage are highly correlated with household energy behavior in Ireland because employment status and internet usage have an important effect on lifestyle, including when to work, play, and rest, and hence yield a difference in electricity use style. The proposed approach offers a simple, transparent and effective alternative to a challenging cross-domain matching problem with massive smart metering data and energy behavior indicators. Highlights: A cross-domain feature selection and coding method for household energy behavior is proposed. The method extracts and connects customers' features from smart metering data and demographic data. The results show that employment status and internet usage are highly correlated with household energy behavior in Ireland.