Stochastic user equilibrium based spatial-temporal distribution prediction of electric vehicle charging load

As the number of electric vehicles (EVs) connected to the grid increases, the EV electricity demand rises dramatically, affecting the grid’s planning and operation and deepening the coupling of the power and transportation systems. Therefore, accurate spatial–temporal distribution prediction of EV charging load is vital for both power system and coupled power-transportation system studies. This paper proposes a novel method based on stochastic user equilibrium (SUE) for predicting the accurate spatial–temporal distribution of EV charging load synchronized with traffic states. A prediction framework of EV charging load based on SUE and trip chain is proposed, which can effectively reflect the actual behavior of EVs in synchronous traffic states. Then, the extended logit-based SUE and equivalent mathematical model are proposed to obtain more detailed traffic states with intersection turning flows and delays. Meanwhile, the unified reachability and charging models are established to ensure that the trip chain is reachable and the charging characteristics are suitable for different EV types. Finally, the method of the successive averages (MSA) and the Dijkstra-based K-shortest paths algorithms are integrated to solve the proposed framework iteratively with stable convergence. Test results on a realistic traffic network show that the proposed method can effectively reflect the charging and trip characteristics of different EV types while ensuring reachability. And it can also accurately predict the overall and individual EV travel costs and total charging loads in detailed synchronous traffic states. In particular, even in the case of high EV penetration with higher peak-to-valley differences and charging demand, the convergence of the prediction is still stable with even more remarkable prediction effectiveness, especially during peak load hours. Furthermore, the quantitative analysis based on proposed criticality indexes reveals that traffic network failures will affect the network-wide traffic states and EV charging loads with different node-level impact characteristics, which should be considered in joint power-transportation restoration scheduling.