Understanding charging dynamics of fully-electrified taxi services using large-scale trajectory data

Modeling the preference of electric shared mobility drivers in choosing charging stations

“…The DCA will first compromise a set of EVSE via the USB port and targets the delay of charging services for the SEV fleet. This deviation from normal operation can hardly be detected by SEV drivers and standard AD techniques due to the wide range of charging duration, from several minutes to 1-2 hours [3]. These minor delays in individual charging activities will result in local congestion at EVSEs and unavailability of SEVs, eventually leading to a cascading failure in ESMS.…”

“…For the former, the anomalies in our study are assumed to be contextual [34]. For instance, a short charging duration (e.g., 20 min) may be considered anomalous during the nighttime but acceptable during the peak hours [3]. For the latter, we treat the historical charging performance as the baseline (e.g., training set), assuming that it only includes data instances collected during the normal operation before the DCA launched.…”

“…Electrifying the fleet for shared mobility service is a promising direction to lower operation costs and reduce greenhouse gas (GHG) emissions [1], [2]. As an example, Shenzhen, China has a fully-electrified taxi fleet by the end of 2019 [3]. Moreover, New York City (NYC) will embrace 100% electrification of its for-hire vehicles fleet by 2030 [4], which further requires the deployment of over 1,750 DC-Fast charging ports [5].…”