Electric vehicle (EV) adoption rates are rising in California because of successful state and federal air pollution mitigation regulations such as the Zero Emission Vehicle mandate, Low Emission Vehicle Regulation III for CA, the U.S. Greenhouse Gas Standard, U.S. Tier 3 Ozone Standard, and the Corporate Average Fuel Economy Standard. While this is a great feat for reducing local air pollution, improperly managed EV charging can lead to peak electricity demand loads that can strain local distribution networks and potentially increase point source emissions from power plants. Managed/smart charging is one of the most promising approaches to prevent suboptimal EV charging as it enables utility operators to control EV charging to reduce load spikes and take full advantage of renewable power. This study aims to develop a Recurrent Neural Network (RNN), an artificial neural network that is highly effective at learning sequential data such as trip logs, to forecast EVs’ trip destinations and charging behavior—information that is essential for electricity load aggregators to effectively manage charging loads. The RNN will be configured using a robust dataset, consisting of trips and charges that have occurred in California between 2015 and 2020 from 358Plug-in Electric vehicles, spanning 9 models.
