Background: Freight allocation is a vital decision in distribution logistics to minimize costs and gain environmental benefits. In this paper, we address the problem of freight allocation optimization on an HSR intermodal network with application for the Ottawa–Quebec City corridor where the HSR system will be constructed. Methods: We develop a novel allocation method in which GNNs encode the intermodal network topology and spatial features, while RL agents learn adaptive freight routing policies through reward optimization, which is enhanced by fractal accessibility metrics for spatial connectivity and MCDM for balancing cost, emissions, and service objectives as well as optimizing dynamic freight flows. The model incorporates geospatial data (population, distance), operational factors (demand, costs), and environmental or policy considerations. Addressing the gap in dynamic, multi-criteria cold-climate HSR freight allocation models for North America, we test our framework on the Ottawa–Quebec corridor. Results: The result shows that compared to traditional methods, the five-hub configuration reduces costs by 15–22% and emissions by 20–28%, while the 11-hub model maintains 94%+ service coverage with an 8–12% efficiency trade-off. Conclusions: The conclusion indicates that the HSR intermodal network is more efficient than road only. Sensitivity analysis highlights that key allocation offers policymakers and logistics planners actionable insights for balancing efficiency and accessibility in HSR freight networks.
Ren et al. (Thu,) studied this question.