Reliability of electric vehicle charging infrastructure:A cross-lingual deep learning approach  

作　　者：Yifan Liu Azell Francis Catharina Hollauer M.Cade Lawson Omar Shaikh Ashley Cotsman Khushi Bhardwaj Aline Banboukian Mimi Li Anne Web Omar Isaac Asensio 

机构地区：[1]School of Public Policy,Georgia Institute of Technology,Atlanta,30332,USA [2]Sam Nunn School of International Affairs,Georgia Institute of Technology,Atlanta,30332,USA [3]School of Civil&Environmental Engineering,Georgia Institute of Technology,Atlanta,30332,USA [4]H.Milton Stewart School of Industrial and Systems Engineering,Georgia Institute of Technology,Atlanta,30332,USA [5]School of Computer Science,Georgia Institute of Technology,Atlanta,30332,USA [6]School of Computer Science,Stanford University,Palo Alto,94305,USA [7]School of Economics,Georgia Institute of Technology,Atlanta,30332,USA [8]Institute for Data Engineering&Science(IDEaS),Georgia Institute of Technology,Atlanta,30332,USA

出　　处：《Communications in Transportation Research》2023年第1期81-91,共11页交通研究通讯（英文）

基　　金：supported by funding from the National Science Foundation(Nos.1931980 and 1945332);Microsoft Azure for research;and the U.S.State Department Diplomacy Lab.

摘　　要：Vehicle electrification has emerged as a global strategy to address climate change and emissions externalities from the transportation sector.Deployment of charging infrastructure is needed to accelerate technology adoption;however,managers and policymakers have had limited evidence on the use of public charging stations due to poor data sharing and decentralized ownership across regions.In this article,we use machine learning based classifiers to reveal insights about consumer charging behavior in 72 detected languages including Chinese.We investigate 10 years of consumer reviews in East and Southeast Asia from 2011 to 2021 to enable infrastructure evaluation at a larger geographic scale than previously available.We find evidence that charging stations at government locations result in higher failure rates with consumers compared to charging stations at private points of interest.This evidence contrasts with predictions in the U.S.and European markets,where the performance is closer to parity.We also find that networked stations with communication protocols provide a relatively higher quality of charging services,which favors policy support for connectivity,particularly for underserved or remote areas.

关 键 词：Electric vehicles Consumer behavior Charging infrastructure Public policy Machine learning Natural language processing Transformer algorithms 

分 类 号：U46[机械工程—车辆工程]