Abstract

Electric vehicles (EVs) presence in the power grid can bring about pivotal concerns regarding their energy requirements. EVs charging behaviors can be affected by several aspects including socio-economics, psychological, seasonal among others. This work proposes a case study to analyze seasonal effects on charging patterns, using a public real-world based dataset that contains information from the aggregated load of the total charging stations of Boulder, Colorado. Our approach targets to forecast and recognize EVs demand considering seasonal factors. Principal component analysis (PCA) was used to provide a visual representation of the variables and their contribution and the correlation among them. Then, twelve classification models were trained and tested to discriminate among seasons the charging load of electric vehicles. Later, a benchmark stage is presented for regression as well as for classification results. For regression models, examined through Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE), the random Forest provides better prediction than quasi-Poisson model widely. However, it was observed that for large variations in electric vehicles’ charging load, quasi-Poisson fits better than random forest. For the classification models, evaluated through Accuracy and the Area under the Curve, the Lasso and elastic-net regularized generalized linear (GLMNET) model provided the best global performance with accuracy up to 100% when evaluated on the test dataset. The results of this work offer great insights for enhancing demand response strategies that involve PEV charging regarding charging habits across seasons. Document type: Article

Abstract

The understanding of charging behavior has been recognized as a crucial element in optimizing roll out of charging infrastructure. While current literature provides charging choices and categorizations of charging behavior, these seem oversimplified and limitedly based on charging data. In this research we provide a typology of charging behavior and electric vehicle user types based on 4.9 million charging transactions from January 2017 until March 2019 and 27,000 users on 7079 Charging Points the public level 2 charging infrastructure of 4 largest cities and metropolitan areas of the Netherlands. We overcome predefined stereotypical expectations of user behavior by using a bottom-up data driven two-step clustering approach that first clusters charging sessions and thereafter portfolios of charging sessions per user. From the first clustering (Gaussian Mixture) 13 distinct charging session types were found; 7 types of daytime charging sessions (4 short, 3 medium duration) and 6 types of overnight charging sessions. The second clustering (Partition Around Medoids) clustering result in 9 user types based on their distinct portfolio of charging session types. We found (i) 3 daytime office hours charging user types (ii) 3 overnight user types and (iii) 3 non-typical user types (mixed day and overnight chargers, visitors and car sharing). Three user types show significant peaks at larger battery sizes which affects the time between sessions. Results show that none of the user types display solely stereotypical behavior as the range of behaviors is more varied and more subtle. Analysis of population composition over time revealed that large battery users increase over time in the population. From this we expect that shifts charging portfolios will be observed in future, while the types of charging remain stable.