Data-Driven Planning of Electric Vehicle Charging Infrastructure: A Case Study of Sydney, Australia

Abstract

With the increase of vehicle travelling range and the decrease of vehicle expense, Electric Vehicles (EVs) are gaining popularity as a new transport option which offers a great opportunity for creating a smart city in terms of transportation electrification. However, the existing research of EV charger planning primarily concentrates on the optimal location seeking from a perspective of global optimization. To achieve a competitive strategy in future, a market-based mechanism is proposed for the problem of EV charger planning with a case study of Sydney where a predict-then-optimise diagram is introduced to predict the EV charging demand by a multi-relation graph convolutional network (GCN)-based encoder-decoder deep architecture and optimise the competitive resource allocation strategy for the charger planning through a Cournot competition game model. A new parallel computational algorithm is proposed to seek the Cournot competition equilibrium with the convergence analysis. Furthermore, the optimal EV charger sizing problem is considered to obtain each service provider’s EV charger number at each zone of the city. Experiments are implemented by using the Sydney public transportation dataset and the corresponding key economic indicators. The results show the effectiveness of our proposed approach which can fill the gap between data and the optimal EV charging infrastructure planning.