Predicting Electric Vehicle Charging Station Availability Using Ensemble Machine Learning

Abstract

Electric vehicles may reduce greenhouse gas emissions from individual mobility. Due to the long charging times, accurate planning is necessary, for which the availability of charging infrastructure must be known. In this paper, we show how the occupation status of charging infrastructure can be predicted for the next day using machine learning models— Gradient Boosting Classifier and Random Forest Classifier. Since both are ensemble models, binary training data (occupied vs. available) can be used to provide a certainty measure for predictions. The prediction may be used to adapt prices in a high-load scenario, predict grid stress, or forecast available power for smart or bidirectional charging. The models were chosen based on an evaluation of 13 different, typically used machine learning models. We show that it is necessary to know past charging station usage in order to predict future usage. Other features such as traffic density or weather have a limited effect. We show that a Gradient Boosting Classifier achieves 94.8% accuracy and a Matthews correlation coefficient of 0.838, making ensemble models a suitable tool. We further demonstrate how a model trained on binary data can perform non-binary predictions to give predictions in the categories “low likelihood” to “high likelihood”.

Analysis of electric vehicle charging station usage and profitability in Germany based on empirical data

“…As discussed in Figure 9 , this was to be expected given the low predictive value of the power level, connector type, and surrounding area type alone. 34 Individual circumstances must therefore be kept in mind when planning and operating a PCS.…”

“…Dynamic pricing and other incentive mechanisms should be implemented at PCSs to make sure that the low occupation of PCS during the night and other off-peak periods is avoided. In the project BeNutz LaSA, prediction algorithms 34 and incentive strategies are created to achieve this goal. It is possible to represent PCS usage using a simple set of parameters fitted to the empirical data.…”

“…33 Other external data sources such as weather data, traffic information, or public holidays were also available to the authors, but are not included in this paper as a previous study showed that they had limited explanatory value. 34 The correlation and share of PCS within the different land use categories, power levels, etc., can be found in section III of the supplementary information.…”

Real-time internet of medical things framework for early detection of Covid-19

“…Experimental results outperformed other individual models with 81.13% accuracy, 71.4% sensitivity and 85.9% specificity. Cutler et al [ 15 ] used GBTs and RF ensemble models to estimate the occupancy rates of electric vehicle charging stations. The GBTs model showed a 94.8% accuracy and a 0.838 Matthews correlation to be a suitable model for charge-load estimation.…”

Communication frame work in an electric vehicle charging station supporting solar energy management

“…Maximising the profit through optimal pricing and scheduling, utilizing various parameters like number of vehicles visited EVCS, demand response, the parking time at the station and the various pricing mechanism with reinforcement learning is proposed in [16]. A geographic specific learning is conducted recently [17] in predicting the availability of charging stations and similar inputs are used in [18] to predict the occupancy of the charging station. Consumed energy, number of charging transactions, charging time, facilities at the station, location of the station and the repeatability of the same vehicle using the same charging station have been used in [19] to predict the popularity of a given charging station.…”