Development of EV charging templates: an improved K-prototypes method

• Published in: IET generation, transmission & distribution Vol. 12; no. 20; pp. 4361 - 4367
• Main Authors: Hong, Juhua, Xiang, Yue, Liu, Youbo, Liu, Junyong, Li, Ran, Li, Furong, Gou, Jing
• Format: Journal Article
• Language: English
• Published: The Institution of Engineering and Technology 13.11.2018
• Subjects: clustering analysis; electric vehicle charging; electric vehicle charging behaviour; Euclidean distance; EV charging load profiles; EV charging templates; feature extraction; improved K‐prototypes method; mixed data feature extraction; morphological dissimilarities; pattern clustering; pulse train waves; rectangular pulse train; rough set theory; Special Issue: New Trends in the Planning of Distribution Network with High Penetration of Renewables and Flexible Loads; traditional distance calculation; traditional distance calculation; improved K-prototypes method; EV charging load profiles; mixed data feature extraction; pulse train waves; electric vehicle charging behaviour; morphological dissimilarities; EV charging templates; electric vehicle charging; pattern clustering; feature extraction; rectangular pulse train; Euclidean distance; rough set theory; clustering analysis
• ISSN: 1751-8687; 1751-8695; 1751-8695
• DOI: 10.1049/iet-gtd.2017.1911

In order to manage the charging behaviour of electric vehicles (EVs), this study for the first time develops a set of EV charging load profiles: EV templates. EV charging profiles have unique waveforms similar to a rectangular pulse train. This characteristics significantly limits the performance of clustering analysis in that traditional distance calculation, such as Euclidean distance, which cannot reflect the morphological dissimilarities. This study proposes a novel clustering method using rough set theory to accurately measure the dissimilarity between the EV profiles. The pulse train waves are firstly extracted as mixed data features, which are partitioned by an improved K-prototypes method based on rough set distance. The proposed method is implemented on the real charging load profiles and compared with K-means and traditional K-prototypes. Their clustering performances are evaluated by diverse validity indices. The results show that the proposed method outperforms other comparison methods.