Full Citation

Title: Short-term Forecasting for Utilization Rates of Electric Vehicle Charging Stations

Citation Type: Journal Article

Publication Year: 2021

ISBN: 9781665449199

ISSN:

DOI: 10.1109/ISC253183.2021.9562948

NSFID:

PMCID:

PMID:

Abstract: Accurate forecasts for the utilization rates of electric vehicle charging stations (CSs) are crucial to coordinating the operations of on-site distributed energy resources. In this paper, we propose to forecast the CS utilization rates by considering key explanatory variables such as historical utilization rates, traffic flows, demographic properties, the number of EV registrations, and points of interest. Three machine learning models, namely random forest, feed-forward neural network, and long short-term memory (LSTM) are adopted for the forecasting task. The proposed algorithms are validated using the real-world utilization data collected from around 130 CSs in Contra Costa County, California. The numerical study results show that the LSTM model achieves the best prediction performance. The lagged CS utilization rates and traffic flows are the two most influential features. More interestingly, the traffic flow plays a more important role in predicting the utilization rates of DC Fast CSs than that of the level 1 (L1) and level 2 (L2) CSs.

Url: https://ieeexplore.ieee.org/abstract/document/9562948

User Submitted?: No

Authors: Ye, Zuzhao; Wei, Ran; Yu, Nanpeng

Periodical (Full): IEEE

Issue:

Volume:

Pages: 1-7

Data Collections: IPUMS NHGIS

Topics: Methodology and Data Collection, Other, Population Data Science

Countries: