Customized Bus Service Design With Holding Control and Heterogeneous Fleet: A Column-Generation-Based Decomposition Algorithm

As a promising urban shared transport mode, the Customized Bus (CB) system has the potential to improve diversity and service quality in urban transportation. This paper is driven by the objective of minimizing costs while fulfilling all service requests. A mixed integer nonlinear programming model is developed for the CB service design problem that jointly optimizes routes, timetables (including the arrival and holding time of each vehicle at each stop), and request-route assignment schemes, with particular consideration for a heterogeneous fleet. The model is subsequently linearized and solved using a Column Generation (CG) based decomposition algorithm, which produces precise solutions for small and medium-scale cases. To address the challenge of solving large-scale cases, we hybridize an Improved Genetic Algorithm (IGA) into the CG framework (CG-IGA) to enhance efficiency in solving the pricing subproblem. Finally, two sets of numerical experiments, involving the Sioux Falls network and a real-world road network in Beijing, are conducted. Computational results show that: (1) the optimality can be achieved for small and medium-scale cases when applying the CG algorithm; (2) the CG-IGA exhibits an exceptional performance compared to other solving methods for large-scale cases in terms of optimality and time-efficiency; (3) the holding control strategy allows for trade-offs between timeout costs and operating costs while improving the flexibility of CB services; and (4) the application of heterogeneous fleets bring at least 17.28% reduction of operating costs and ensures high utilization of transport resources.