Predicting future loads of electric vehicles in the UK

Roy, Rahul and Dokka Venkata Satyanaraya, Trivikram and Ellis, David and Hazas, Mike (2020) Predicting future loads of electric vehicles in the UK. Masters thesis, Lancaster University.

[thumbnail of 2019rahulroymsc(research)]
Text (2019rahulroymsc(research))
2019rahulroymsc_research_.pdf - Published Version
Available under License Creative Commons Attribution-NonCommercial-NoDerivs.

Download (4MB)


This thesis aims to propose a robust statistical model to predict the future energy demand on low voltage distribution networks based on the data obtained from the EV (electric vehicle) trials of Electric Nation project, conducted from 2017 to 2018. While the ultimate objective of Electric Nation is to assess the impact of EV charging on distribution networks and enable the distribution network operators (DNOs) to make informed decisions on demand management, this research project, as part of Electric Nation, aims to build relevant statistical models that would help the industry partner, EA Technology, to forecast the quantum of energy consumption, with high accuracy, that EV charging would lead to. In current research, we develop four statistical models based on four different algorithms: we start with time series regression as the benchmark model and iteratively improve the forecast accuracy of the benchmark model by boosting methodology. In addition, we also explore deep learning models (LSTM networks as the data is sequential) and identify that such models, with little hyperparameter tuning, deliver the best forecast accuracy among all the models.

Item Type:
Thesis (Masters)
?? electric vehicle (ev); ev charging; time series; forecasting; regression; arima; lstm networks. ??
ID Code:
Deposited By:
Deposited On:
01 Apr 2020 11:45
Last Modified:
18 Jul 2024 00:19