Prediction of EV Charging Behavior Using Machine Learning

Authors

  • Preethi Singireddy, D. Aruna Kumari,

DOI:

https://doi.org/10.17762/msea.v71i4.1163

Abstract

Electric vehicles (EVs), a foundational element about smart mobility in applications for smart cities, are gaining popularity due to their role in lowering greenhouse gas emissions. However, one about major difficulties isburden that widespread EV deployment places oninfrastructure about electrical grid. Utilizing clever scheduling algorithms to controlrising demand for public charging isanswer to this problem. Scheduling algorithms can be made better by using data-driven tools&machine learning techniques to studycharging behaviour about EVs.use about historical charging data to forecast behaviour, such as departure time&energy requirements, has received a lot about attention from researchers. However, factors that have been mostly ignored, such as weather, traffic,&local occurrences, might add relevant representations& offer improved predictions. Because about this, we propose in this studyuse about past charging data along with weather, traffic,&events data to forecastlength about an EV session&its energy consumption using well-known machine learning methods including random forest, SVM, XGBoost,&deep neural networks. An ensemble learning model outperforms previous research inliterature in terms about predictive performance, with SMAPE scoresabout 9.9%&11.6 percent for session length&energy usage, respectively. We show that both forecasts significantly outperform earlier research onsame dataset,&we emphasizesignificance about traffic&weather data for charging behaviour predictions.

Downloads

Published

2022-10-31

How to Cite

Preethi Singireddy, D. Aruna Kumari,. (2022). Prediction of EV Charging Behavior Using Machine Learning. Mathematical Statistician and Engineering Applications, 71(4), 5664–5675. https://doi.org/10.17762/msea.v71i4.1163

Issue

Section

Articles