Work Section Planning for Accompanying Construction of Extra-Long Tunnel and Ballastless Track of Mountainous High-Speed Railways

Reasonable construction methods and construction work section planning (WSP) are the key influencing factors in ensuring that the construction of extra-long railway tunnels is completed in time. This study proposes a new construction planning method for accompanying the construction of tunnel and ballastless track (AC-TBT) in high-speed railways, which allows the construction of a tunnel and ballastless track to occur with time overlaps and shortens the construction period of extra-long tunnels. To further evaluate the advantages of AC-TBT, this study first designs a deep Q-network (DQN) algorithm to find the optimal WSP scheme under dual-objective optimization for time constraints and number of construction teams limitations. Second, a construction planning model based on integer linear programming (ILP), and a Monte Carlo random sampling is established to calculate construction schedules and the number of construction teams involved in construction for different WSP schemes. Monte Carlo random sampling is used to generate safety buffer times to mitigate the impact of uncertainties in tunnel construction and is included as an important parameter in the ILP model. The results of the case study show that, compared with the conventional construction of tunnel and ballastless track, the two WSP schemes based on AC-TBT save 80 and 40 days of construction duration, respectively. The sensitivity analysis of the reward function weights shows that the weight level directly influences the DQN’s decision-making tendency regarding different construction schedules and construction team allocations. Finally, DQN demonstrates a faster convergence rate and better performance compared to Q-learning.