A data-driven approach to identify households with plug-in electrical vehicles (PEVs). (15th December 2015)

• Record Type: Journal Article
• Authors: Verma, Anoop; Asadi, Ali; Yang, Kai; Tyagi, Satish
• Is Part Of: Applied energy. Volume 160(2015:Dec. 15)
• Page Start: 71
• Page End: 79
• Publication Date: 2015-12-15
• Subjects: Plug-in electric vehicles (PEVs) -- Energy envelopes -- Data mining -- Classification; Power (Mechanics) -- Periodicals; Energy conservation -- Periodicals; Energy conversion -- Periodicals; 621.042
• DOI: 10.1016/j.apenergy.2015.09.013
• Languages: English
• ISSNs: 0306-2619

Abstract

Highlights: A data-driven approach with 80% or greater accuracy in identifying households charging PEVs. A transformed energy envelope metric with delta threshold to improve classification results. PEV user's classification based on real-world utility usage data only. Random forest based classification ensuring quick convergence. Abstract: In recent years popularity of plug-in electric (PEV) vehicles has grown significantly. Charging of such vehicles is typically done at home from a standard outlet or at corporate car locations and thus adds extra load on the distribution grid. Due to high power consumption of PEV charging, the utility industries face enormous challenges to provide this extra demand. The identification of charging patterns of PEV is thus of paramount importance to balance the electric load and assure coordinated charging. More specifically, there is a need to identify users with PEVs to better manage the load distribution. In the present research, an analysis based on energy envelopes of the usage patterns is performed. A set of well-known data mining algorithms are used to identify the best classifier to help identify customers with PEVs.