A Real-Time Synchronous Dispatching and Recharging Strategy for Multi-Line Electric Bus Systems

Improving the sharing rate is critical for electric bus systems to achieve their goal of reducing greenhouse gas emissions and enhancing system efficiency. This study presents a real-time synchronous dispatching and recharging strategy for multi-line electric bus systems to mitigate the adverse effects of operational randomness. The strategy integrates recharging requirements into real-time control decisions by synchronously determining dispatching and recharging plans for each bus based on continuously updated operational data. To implement this approach, models for bus operational state evolution and an objective function for minimizing passenger waiting time are formulated. These models are then solved using a deep reinforcement learning (DRL) technique that combined clipped double Q-learning with soft actor-critic (SAC). Numerical examples using realistic operational data are conducted to assess the strategy's performance, along with the SAC-based DRL algorithm. The results show that the proposed strategy effectively improves bus service quality while optimizing fleet resources.