A spatiotemporal distribution prediction model for electric vehicles charging load in transportation power coupled network.

With increasing demand of electric vehicles (EVs), problems such as insufficient EV charging piles and unreasonable layout of EV charging stations are also becoming prominent. New challenges are introduced to the planning of urban EV charging infrastructures. To effectively plan the charging facilities, accurately predicting EV charging loads is essential. The present study proposes a spatio-temporal distribution prediction model for EV charging loads in transportation-power coupled network (TPCN). First, path planning is performed separately using the Dijkstra algorithm and refined origin-destination (OD) probability matrix based on the travel characteristics of various vehicle types. The charging selection model is then formulated considering multiple compelling factors, such as transportation conditions, ambient temperature, rest days and so on. Furthermore, the transportation-power coupled network is established based on the graph-theoretic analysis approach, and the spatial and temporal distribution characteristics of charging loads are predicted by Monte Carlo simulation. Finally, a case study is conducted in an actual urban region. The results show that EV charging load presents significant differences in different functional areas, different time periods and scenarios, and the proposed method can accurately predict the spatial-temporal distribution of charging load. This study represents a reliable approach for predicting charging demand in a certain region, and also provides powerful support for the rational planning of EV charging stations.