A Hierarchical Spatio-Temporal Framework for Sustainable and Equitable EV Charging Station Location Optimization: A Case Study of Wuhan

Deploying public EV charging infrastructure while balancing efficiency, equity, and implementation feasibility remains a key challenge for sustainable urban mobility. This study develops an integrated, grid-based planning framework for Wuhan that combines attention-enhanced ConvLSTM demand forecasting with a trajectory-derived, rank-based accessibility index to support equitable network expansion. Using large-scale charging-platform status observations and citywide ride-hailing mobility traces, we generate grid-level demand surfaces and an accessibility layer that helps reveal structurally connected yet underserved areas, including demand-sparse zones that may be overlooked by utilization-only planning. We screen feasible grid cells to construct a new-station candidate set and formulate expansion as a constrained three-objective optimization problem solved by NSGA-II: maximizing demand-weighted neighborhood service coverage, minimizing the Group Parity Gap between low-accessibility populations and the citywide population, and minimizing grid-connection friction proxied by road-network distance to the nearest power substation. Practical deployment plans for 15 and 30 sites are selected from the Pareto set using TOPSIS under an explicit weighting scheme. Benchmarking against random selection and single-objective greedy baselines under identical candidate pools, constraints, and evaluation metrics demonstrates a persistent coverage–equity–cost tension: coverage-driven heuristics improve demand capture but worsen parity, whereas equity-prioritizing strategies reduce gaps at the expense of coverage and feasibility.