Abstract: In view of the low scheduling efficiency caused by strong process coupling, complex resource constraints and frequent dynamic disturbances in the production of precast box girders for high-speed railways, this paper designed and implemented an automatic production scheduling system for such girders. The system decoupled process logic based on hierarchical task networks and established a four-dimensional constraint model integrating time, process, space and resource dimensions. Meanwhile, it adopted a hybrid scheduling optimization algorithm that combined global search of genetic algorithm, local optimization of tabu search and dynamic response of simulated annealing, and realized multi-objective collaborative optimization and real-time disturbance response. Engineering verifications show that the scheduling schemes generated by this system outperform traditional manual empirical scheduling schemes under various typical disturbance scenarios. The system cuts the construction period by an average of 15.5%, effectively improves the resource synergy efficiency and production robustness of girder yards, and provides an effective scheduling solution for the intelligent construction of high-speed railways.