J Shanghai Jiaotong Univ Sci ›› 2023, Vol. 28 ›› Issue (1): 136-149.doi: 10.1007/s12204-023-2576-0

Previous Articles     Next Articles

Robust Charging Demand Prediction and Charging Network Planning for Heterogeneous Behavior of Electric Vehicles


ZHANG Yilun1‡ (张轶伦), XU Sikun2‡ (徐思坤), XU Jie1 (徐 捷), ZENG Xueqi3 (曾学奇), LI Zheng4 (李 铮), XIE Chi5∗ (谢 驰)   

  1. (1. Department of Industrial Engineering and Management, Shanghai Jiao Tong University, Shanghai 200240, China; 2. Olin Business School, Washington University in St. Louis, St. Louis, MO 63130, USA; 3. Urban Mobility Institute, Tongji University, Shanghai 201804, China; 4. Antai College of Economics and Management, Shanghai Jiao Tong University, Shanghai 200030, China; 5. College of Transportation Engineering, Tongji University, Shanghai 201804, China)
  2. (1. 上海交通大学 工业工程与管理系,上海200240;2. 华盛顿大学圣路易斯分校 商学院,美国 圣路易斯 63130;3. 同济大学 城市交通研究院,上海201804;4. 上海交通大学 安泰经济与管理学院,上海200030;5. 同济大学 交通运输工程学院,上海201804)
  • Received:2021-12-27 Online:2023-01-28 Published:2023-02-10

Abstract: This study addresses a new charging station network planning problem for smart connected electric vehicles. We embed a charging station choice model into a charging network planning model that explicitly considers the heterogeneity of the charging behavior in a data-driven manner. To cope with the deficiencies from a small size and sparse behavioral data, we propose a robust charging demand prediction method that can significantly reduce the impact of sample errors and missing data. On the basis of these two building blocks, we form and solve a new optimal charging station location and capacity problem by minimizing the construction and charging costs while considering the charging service level, construction budget, and limit to the number of chargers. We use a case study of planning charging stations in Shanghai to validate our contributions and provide managerial insight in this area.

Key words: electric vehicle, charging network planning, charging behavior, robust demand prediction

摘要: 研究了智能网联电动汽车背景下的充电站网络规划问题。从数据驱动的角度,考虑充电行为的异质性构建了用户对充电站的选择模型,并将其嵌入到网络规划模型中。同时,提出了鲁棒性充电需求预测方法,以解决充电行为数据规模较小且稀疏所带来的样本偏差和数据丢失问题。在上述基础上,以最小化建造和充电成本为目标,考虑充电服务水平、建设预算、充电桩个数限制等约束,建立了充电站选址定容的数学模型。以上海实际数据进行了数值实验,结论充分佐证了创新点的合理性,并提供了相应的管理启示。

关键词: 电动汽车,充电网络规划,充电行为,鲁棒性需求预测

CLC Number: