A Data-Driven Framework for Improving Public EV Charging Infrastructure: Modeling and Forecasting

This work presents an investigation and assessment framework, which, supported by realistic data, aims at provisioning operators with in-depth insights into the consumer-perceived Quality-of-Experience (QoE) at public Electric Vehicle (EV) charging infrastructures. Motivated by the unprecedented EV...

Full description

Saved in:
Bibliographic Details
Published inarXiv.org
Main Authors Al-Dahabreh, Nassr, Mohammad Ali Sayed, Sarieddine, Khaled, Elhattab, Mohamed, Khabbaz, Maurice, Atallah, Ribal, Assi, Chadi
Format Paper
LanguageEnglish
Published Ithaca Cornell University Library, arXiv.org 08.12.2023
Subjects
Electric vehicle charging
Forecasting
Infrastructure
Machine learning
Mathematical models
Model accuracy
Operators
Performance evaluation
Performance measurement
Provisioning
Online AccessGet full text

Cover

More Information
Summary:This work presents an investigation and assessment framework, which, supported by realistic data, aims at provisioning operators with in-depth insights into the consumer-perceived Quality-of-Experience (QoE) at public Electric Vehicle (EV) charging infrastructures. Motivated by the unprecedented EV market growth, it is suspected that the existing charging infrastructure will soon be no longer capable of sustaining the rapidly growing charging demands; let alone that the currently adopted ad hoc infrastructure expansion strategies seem to be far from contributing any quality service sustainability solutions that tangibly reduce (ultimately mitigate) the severity of this problem. Without suitable QoE metrics, operators, today, face remarkable difficulty in assessing the performance of EV Charging Stations (EVCSs) in this regard. This paper aims at filling this gap through the formulation of novel and original critical QoE performance metrics that provide operators with visibility into the per-EVCS operational dynamics and allow for the optimization of these stations' respective utilization. Such metrics shall then be used as inputs to a Machine Learning model finely tailored and trained using recent real-world data sets for the purpose of forecasting future long-term EVCS loads. This will, in turn, allow for making informed optimal EV charging infrastructure expansions that will be capable of reliably coping with the rising EV charging demands and maintaining acceptable QoE levels. The model's accuracy has been tested and extensive simulations are conducted to evaluate the achieved performance in terms of the above listed metrics and show the suitability of the recommended infrastructure expansions.
ISSN:2331-8422