Using Spatial and Temporal Context for Predicting Energy Consumption of Electric Vehicles
Term
4. term
Education
Publication year
2019
Submitted on
2019-06-11
Pages
68
Abstract
The number of electric vehicles in use worldwide is rising every year. Even with improvements in battery technology giving longer range to electric vehicles, many owners still worry about the range of their vehicles. This worry is termed range anxiety. Obtaining more precise energy predictions can combat range anxiety, as they allow the owners of electric vehicles to more precisely gauge the range of their vehicles. In this report, we detail the creation of multiple machine learning models that predict energy consumption of electric vehicles by exploiting contextual information related to trips conducted by these vehicles. We base this project on a large dataset of electric vehicle driving data, which describes trips on the Danish road network. A number of external data sources are integrated with the dataset, such as height data obtained from a height raster of Denmark, and a speedmap from Vejdirektoratet. We conclude that a Recurrent Neural Network architecture is able to successfully exploit the sequential nature of the context data to improve significantly upon a baseline model.
Keywords
Documents