Non-intrusive Electric-vehicle Load Disaggregation Algorithm for a Data-driven EV Integration Strategy

Abstract

Electric vehicle (EV) charger demand has increased from 1.44 kW to between 3.3 kW and 17.2/19.2 kW [1] in the past 10 years – a 3 to 17/19 times the average consumption from a single home. By 2045 EV penetration will on average grow by 34 times (GWh) from today in Southern California Edison’s (SCE) territory [2]. To develop a data-driven utility EV grid integration strategy, EV customer charging behaviors need to be well understood. The ability to disaggregate EV loads, or segregate EV loads from household loads, is very useful in supporting enterprise forecasting and distribution planning, developing distribution standards, and capital request justification for a utility general rate case. EV telemetry and individual metered EV loads are not always available to the utility. In this paper, we present a lightweight efficient EV disaggregation methodology, with several advantages using real power measurements from advanced meter infrastructure (AMI) meters and demonstrated the algorithm and utility applications at scale (approximately 62,000 customers), and showed how the results can support utilities’ strategic need to develop a reliable, affordable, and safe EV integration strategy.