The continuous growth and evolve of vehicle electrification causes the electric power systems to confront new challenges, since the load profile changes, and new parameters are being set. With the number of EVs gradually rising, problems may occur in technical characteristics of the network, like bus voltages and line congestion. Therefore, it is necessary to develop EV management systems so as to prevent such phenomena. The effectiveness of such systems is heavily depended on the early knowledge of future demand. This knowledge can be provided by accurate EV load forecasting techniques. In this paper the use of data mining methods for forecasting the EV charging demand was studied. Two different realistic study cases where considered, and the performance of four different data mining methods was evaluated. In the first study case the day-ahead charging demand of 3,000 EVs was forecasted and compared to the actual data. The second study case considered a fleet of 2,130 EVs, and predicted the charging demand of a whole week on a half-hourly basis. The results showed that data mining methods can be used for forecasting the EV charging load, with increased accuracy especially when the configuration parameters of each method are carefully selected. However, more cases have to be studied, in order to clearly understand the key attributes that indicate the choice of one data mining method over another.
Electric vehicle load forecasting using data mining methods
2013
6 Seiten, Bilder, Tabellen, 25 Quellen
Conference paper
English
Electric Vehicle Charging Load Forecasting based on Monte Carlo
VDE-Verlag | 2022
|Daily Electric Load Forecasting Using Artificial Neural Network
Online Contents | 1995
|