Probability-based hierarchical matching approach for stochastic electric vehicle scheduling considering power grid load

With the rapid adoption of electric vehicles, the electric vehicle scheduling problem (EVSP) has become a critical challenge in public transport. Many studies at the operational scheduling level have considered stochastic traffic or power grid security in isolation, overlooking their interdependence, whereby stochastic trip times affect charging demand, exacerbate peak loads and reduce schedule reliability. To fill this gap, this article investigates the stochastic EVSP with power grid load considerations and proposes a probabilistic model that jointly minimizes fleet size, operating cost and charging peak load while maximizing on-time performance. A probability-based hierarchical matching (P-HM) approach is developed, partitioning the timetable into tiers and matching adjacent tiers based on compatibility probabilities, combined with a greedy local search to mitigate peak-load violations. Numerical results demonstrate that P-HM significantly outperforms benchmarks, particularly in reducing fleet size, and that the proposed model improves robustness and grid security compared to existing EVSP formulations.