Robust Identification of EV Charging Profiles

摘要

Electric Vehicle (EV) charging profiles are important for electric energy distribution network operation, load forecasting, as well as utility grid expansion and planning. A typical individual EV charging profile typically includes its start time, its initial battery state-of-charge (SOC), and its (total) charging time. For distribution networks, a network/regional EV charging profile also includes number of EVs at different times (e.g., each hour) of the day. Most existing work on EV charging profiles assume that EV charging profiles follow certain statistical distributions in a universal manner. However, EV charging profiles are highly related to weather conditions, traffic patterns, rate tariffs, population density growth, and population active time periods, for which different regions (residential vs commercial, rural vs suburban, and developed vs developing) are fundamentally different. Instead of make simplification, this paper proposes to extend authors work on non-intrusive load modeling, utilize ubiquitous advanced metering infrastructure (AMI) data such as smart meters, and build robust and accurate EV charging profiles for any networks under consideration.