Loading [a11y]/accessibility-menu.js
Deep Transfer Learning for Detecting Electric Vehicles Highly Correlated Energy Consumption Parameters | IEEE Journals & Magazine | IEEE Xplore

Deep Transfer Learning for Detecting Electric Vehicles Highly Correlated Energy Consumption Parameters


Impact Statement:The drastic growth in the conventional transportation system raises serious air pollution concerns. Eco-friendly vehicles, in contrast, have been introduced as an alterna...Show More

Abstract:

Implementation of advanced intelligent deep learning techniques for electric vehicles (EVs) energy consumption analysis is obstructed by two main subjects. First, the pro...Show More
Impact Statement:
The drastic growth in the conventional transportation system raises serious air pollution concerns. Eco-friendly vehicles, in contrast, have been introduced as an alternative to alleviate such environmental issues. The aforementioned switch requires proper infrastructure to increase the public’s interest before mass production. Ensuring EV owners’ satisfaction by increasing the quality of experience and desecrating range anxiety is the primary goal for EV producers. One attempt to reach this goal is to provide a precise remaining battery level for EVs based on users’ driving destinations and behaviors. However, achieving accurate battery estimation leveraging AI is limited to the amount of historical data. We, therefore, consider introducing an estimation model which trains insufficient EV data history by transferring previously trained knowledge to increase the precision of remaining battery level prediction.

Abstract:

Implementation of advanced intelligent deep learning techniques for electric vehicles (EVs) energy consumption analysis is obstructed by two main subjects. First, the problem of finding a very similar collection of datasets to the actual EVs energy usage in terms of feature space and data distribution. Second, training a retrained model from scratch requires a massive amount of computational power; however, this does not guarantee to catch rare events included in datasets. To mitigate the aforementioned concerns, this article aims to present a model based on deep transfer learning (DTL) between domain-variant datasets, to reduce the need for the existence of a vast amount of EVs data, including driving characteristics and patterns. Also, this model applies a distributed cooperative learning approach to identify highly correlated energy consumption parameters by building the model on previously acquired knowledge from preceding learning phases in order to enhance the artificial intellig...
Published in: IEEE Transactions on Artificial Intelligence ( Volume: 5, Issue: 8, August 2024)
Page(s): 4087 - 4100
Date of Publication: 25 January 2024
Electronic ISSN: 2691-4581

Funding Agency:


References

References is not available for this document.