Robust Service and Charging Plan for Dynamic Electric Demand-Responsive Transit Systems

This study proposes a robust route optimization model for electric Demand-Responsive Transit (e-DRT) services, where dispatched vehicles may deviate from the determined plan to serve real-time demands. In particular, online partial charging strategies are coordinated with flexible service schedules. To benefit the productivity of the e-DRT system, the route schedule and charging time are changed dynamically. A two-phase Adaptive Large Neighborhood Search (ALNS) -based heuristic is proposed to effectively solve the proposed problem. The baseline case and large-scale cases are presented to verify the effectiveness and accuracy of the proposed method. Comparisons between CPLEX and the proposed algorithm suggest that the proposed algorithm can considerably improve computational efficiency. A comparative analysis shows the proposed model takes 21% less total cost than the alternative non-robust model. Further, two sensitivity tests are designed to unveil the impacts of unmet real-time requests and the charging rate on the e-DRT's performance.