As a knowledge and innovations center, the researchers at ElaadNL periodically contribute to scientific research regarding electric vehicles and Smart Charging. During this summer, two papers co-written by Nazir Refa (Team Leader Data Analytics at ElaadNL) on Smart Charging are published. These papers can be downloaded below.
Smart charging achieves better performance when it is driven by reasonably accurate predictions of charging behaviour. Hence, for a smart charging scheme that dynamically updates a charging schedule, updating the predictions of charging behaviour could be beneficial. In this paper, we explore the potential to improve the accuracy of prediction models of the connection duration to a charging station by updating the predictions as the charging sessions unfold. We propose and compare several strategies for making the prediction updates with the use of state-of-the-art LightGBM machine learning models. More concretely, we compare a single-model to multiple-models combined with regularly and irregularly spaced updates in time. The multiple-model with irregular updates achieves the best performance while improving the prediction accuracy up to 30%, compared to conventional prediction approaches. It appears as efficient to update the predictions with higher frequency in the very early stages of charging sessions. Later on, regular updates are sufficient.
The outcomes of this research are presented at a scientific conference (SpliTech 2022) in July 2022.
We developed state-of-the-art prediction models for the connection duration with the asymmetric loss function and analysed the impact of asymmetries on the performance of smart charging schemes. The stochastic nature of charging demand and renewable generation requires intelligent charging driven by predictions of charging behaviour. The conventional prediction models of charging behaviour usually minimise the symmetric quadratic loss function. Moreover, the adequacy of predictions is almost solely evaluated by accuracy measures, disregarding the consequences of prediction losses in an application context. Here, we study the role of asymmetric prediction losses which enable balancing the over- and under-predictions and adjust predictions to smart charging algorithms. Using the main classes of machine learning methods, we trained prediction models of the connection duration and compared their performance for various asymmetries of the loss function. In addition, we proposed a methodological approach to quantify the consequences of prediction losses on the performance of selected archetypal smart charging schemes. In concrete situations, we demonstrated that an appropriately selected degree of the loss function asymmetry is crucial as it almost doubles the price range where the smart charging is beneficial and increases the extent to which the charging demand is satisfied up to 40%. Additionally, the proposed methods improve charging fairness since the distribution of unmet charging demand across vehicles becomes more homogeneous.
This research has been published in the International Journal of Electrical Power and Energy Systems in August 2022.