Planning Electric Vehicle Charging Stations Based on User Charging Behavior

Abstract

Electric vehicles (EVs) as a green alternative of fossil-fuel vehicles (FFVs) have been promoted by many governments all over the world. As a result, constructing an efficient charging pile network has become a crucial task for governments and manufacturers to increase EV adoption, as well-planned charging sites can serve more EV users at a lower cost and improve user satisfaction. Unfortunately, most of existing planning approaches for EV charging stations estimate charging demand and optimize locations based on traffic patterns of FFVs, e.g., traffic flow and parking locations, and the patterns of charging behavior are overlooked causing an inefficient network layout for existing EV users. In this paper, we propose and implement a novel algorithm to estimate charging demand and to plan new charging stations. The observations and analysis of the usage data of the charging mobile app developed by the official EV public service platform of Beijing and pile usage data of the charging pile network (CPN) of Beijing are presented. Users' charging-related search behavior and navigation behavior and the pile usage pattern are analyzed and modeled. A Bayesian-inference-based algorithm is proposed to fuse the three models to estimate charging demand. A flexible objective function is introduced to tune the benefit between serving the existing EV users well and attracting more FFV drivers. Finally, a reference system is developed using Beijing as a target city, and providing extensive experiments to demonstrate the performance of our system.