Multi-phase location and capacity planning of electric-hydrogen charging stations with GCN in coupled power-traffic system

With the rapid development of urban electric–hydrogen coupling systems and the increasing adoption of electric and hydrogen vehicles, the coordinated expansion of integrated charging stations, transportation networks, and hydrogen systems has become a pressing need. This paper proposes a novel methodology that reformulates the Electric–Hydrogen User Equilibrium–Traffic Assignment Problem into a set of Mixed-Integer Linear Programming constraints using a Graph Convolutional Network. This transformation overcomes the inherent complexity of coupling the upper- and lower-level structures of the user equilibrium model with other optimization frameworks, thereby substantially enhancing computational efficiency in the planning process. To broaden the solution space, an Extended User Equilibrium–Traffic Assignment Problem is further introduced, which achieves traffic equilibrium while incorporating the heterogeneous charging and refuelling preferences of electric and hydrogen vehicle users. The proposed framework is particularly suitable for multi-phase planning, enabling coordinated development of charging stations, power distribution networks, and hydrogen production facilities. It also adapts to varying levels of financial support and regional demand fluctuations, providing urban planners with flexible and robust decision-making tools. The validity and effectiveness of the approach are demonstrated through comprehensive case studies, highlighting its potential for practical applications in sustainable urban infrastructure planning and optimization.