This paper addresses the large-scale cross-regional multimodal scheduling problem for refined products, and considers the spatiotemporal differences between the continuous sequential transportation process in pipelines and the containerized transportation modes of rail, waterway and highway. Using a daily time scale and with the objective of minimizing transportation costs, a mixed-integer linear programming (MILP) model is formulated for the coordinated optimization of pipeline-rail-water-road multimodal transportation. To overcome the computational inefficiency bottleneck when solving the MILP model for large-scale, long-term and cross-regional logistics scheduling scenarios, this paper innovatively proposes an efficient hierarchical "parent-child" problem decomposition algorithm. The approach divides the original problem into multiple sets of parent-child subproblems through time window segmentation, while utilizing batch movement information from parent model solutions to update dynamic batch sets in child models. As time windows expand, this method effectively controls the growth of related binary variables and constraint adjustments, significantly reducing the original problem’s variable search space and remarkably accelerating the optimization process. Finally, using a refined petroleum product logistics system in western China as a case study, the results demonstrate the superior computational efficiency and solution quality of the proposed approach. Compared to directly solving the MILP model using Gurobi, the proposed algorithm reduces the number of binary variables and constraints by an average of 68.0% and 60.4% respectively during the iteration process. and also reduces the computation time by an average of 86.7% under the premise of maintaining optimality.