Hogeschool van Amsterdam

Intelligent Data-driven Optimisation of Charging Infrastructure

03. Forecasting

Within this work package, forecasting models are being developed. These models have the aim of informing the professionals about the state of the charging infrastructure in the near future, like predicting the occupancy rate or the kWh consumption. This information enables professionals in the chain of charging infrastructure to decide on whether to expand the infrastructure. Forecasting models can also focus on the end user, like: With how many percent certainty can I charge my car at a specific time and day at a particular charging location?


Prediction models are divided into time series models and causal models. Time series models forecast based on historical data. Causal models base their predictions on correlations that approach the underlying mechanism. Another line of activity within the forecast models are called machine learning models. Based on a large historical data set (charging sessions) an algorithm is generated that makes forecasting possible. In the context of electric vehicles, these models are used for predicting the adoption of electric vehicles in a country or major region.

Published by  IDO-Laad 21 October 2016