Hundreds of thousands of plug‐in electric vehicles (PEVs) have been sold in the United States in the past decade, and this number is projected to grow rapidly [1, 2]. For PEVs, recharging replaces refueling, strengthening the link between transportation systems and electric distribution systems. A common concern regarding increased market penetration of PEVs is that numerous vehicles charging simultaneously or during peak hours will overload electric power generation and distribution systems. In order to avoid this situation, charging must be managed strategically. Vehicle to grid (V2G) technology provides management not only by coordinating charging, but also by allowing PEVs to store and return power back to the grid. PEVs are particularly suited to supply a service known as regulation, which matches the instantaneous demand for electricity to the power supply. Regulation providers must respond rapidly to commands from the grid operator to absorb or return power . Whether a PEV is used for V2G services, or is simply part of a coordination scheme, it is important for the party managing the EVSE to know how the PEV will interact with the system.
If an EVSE could identify a PEV’s type (make, manufacturer and year), it could predict much that vehicle’s behavior, such as maximum and minimum charge rates, or time needed to charge. Even if the PEV will not provide V2G services, optimally incorporating a PEV into a system requires an aggregator to know the types of PEVs joining the system. Figure above illustrates one scenario where information about vehicle type enables an EVSE to provide optimized services. If the EVSE does not know the types of PEVs connected, it will offer both 5 kW. The Volt will not use all the power offered, while the Focus vehicle will be forced to charge at less than its maximum rate. When the EVSE knows both types, it can offer each the appropriate amount of power, and each PEV will charge at is maximum rate. Currently, standard charge protocols do not provide a method for EVSEs to determine the type of vehicle connected. While PEV self‐reporting would be ideal, it is not widely implemented or standardized for this purpose. The substantial and rapidly growing fleet of PEVs in use today cannot identify themselves to an EVSE, and updating all PEVs in the future when protocols including identification are standardized would be financially and logistically infeasible.
We believe that an EVSE may recognize the types of PEVs connected to charge without direct communication, but based on the PEV’s unique implementation of communication protocols, and distinct charge profiles. The goal of this research project are to determine how these implementations manifest themselves in the physical characteristics of communication signals, and in charging profiles in the first few minutes after a PEV connects to and EVSE for charge. We are in a uniquely strong position to perform this research because our university is the center of vehicle‐ to‐grid technology (www.udel.edu/v2g). Our research group builds and deploys electric vehicle charging stations for our campus and for world‐wide v2g technology trials. We also manage a fleet of three dozen electric vehicles operating on our campus. All this infrastructure is readily available to collect perform extensive data collections of actual electric vehicle charging activity.
For collecting data and testing recognition algorithms, we will utilize the University of Delaware V2G project’s extensive network of EVSEs. We will configure several of the EVSEs to perform auxiliary reporting. These reports will be preprocessed via custom programs that we will develop so they may be easily imported into data‐mining tools to define an algorithm for recognizing PEVs. Using the results, we will configure several EVSEs to perform recognition. The accuracy of the recognition algorithm will be verified using planned, observed plug‐in events, and reporting tools such as PlugShare. The algorithm shall be considered sufficiently robust when, for all plug in events, the EVSE recognizes all EVs when the type is in the system, and correctly identifies when an unfamiliar type of EV is connected.
The broader impact of this work is to enable parties managing PEVs (such as a utility or aggregator) to project power demand, avoid faults and circuit overloads, and optimize charging for all vehicles. In addition to publishing this work, we plan to develop open‐source tools that will enable other researchers to duplicate and expand on our experiments. On a broader scope, the outcomes of this project comprise a resource for those developing smart grids. In particular, the knowledge and tools resulting from our research may be used for greater integration of distributed generation in electric distribution systems, implementation of smart metering, and large‐scale load management.