A sequential optimization framework for public transit timetabling and driver scheduling using hybrid metaheuristics

This study addresses two critical and interrelated problems in public transit operations: the Transit Network Timetabling Problem (TNTP) and the Bus Driver Scheduling Problem (BDSP). While integrated approaches exist theoretically, they often suffer from computational intractability when applied to large-scale real-world networks. To bridge the gap between theoretical optimization and practical implementation, we propose a robust sequential optimization framework. First, passenger demand is analyzed using ticket transaction data to formulate the TNTP model, minimizing passenger waiting times. Subsequently, the output constitutes the input for the BDSP model, which aims to optimize driver workload fairness by minimizing the deviation between the actual and ideal driving hours (8 hours) and reducing idle times. To solve the NP-hard BDSP effectively, we develop and compare two hybrid metaheuristics: Genetic Algorithm combined with Simulated Annealing (GA-SA) and Genetic Algorithm with Tabu Search (GA-TS). Empirical results based on real-world transit data demonstrate that the proposed GA-SA hybrid algorithm outperforms standard approaches in terms of solution quality with short calculation time. The study provides transit operators with a decision-support tool that balances passenger satisfaction and driver resource utilization efficiently.