GNN-based spatiotemporal resilience prediction for Maritime Silk Road transportation network

doi:10.1016/j.trd.2026.105347

GNN-based spatiotemporal resilience prediction for Maritime Silk Road transportation network

2026-04-03

0

PRE

AI

DOI：10.1016/j.trd.2026.105347

Original

Original request for help

Save

Abstract

En 中文

The resilience of the 21st Century Maritime Silk Road transportation network is crucial for maritime safety. This study develops a resilience theory framework and proposes a graph neural network–based model to quantify and predict the spatiotemporal dynamics of Maritime Silk Road transportation network resilience. A four-layer weighted graph convolution model integrates multi-source heterogeneous data, including ship, environmental, and historical accident information. Using spatiotemporal network analysis and dynamic simulation, annual resilience performance is assessed, and local resilience variations under tropical cyclone impacts are predicted. Results indicate that the Maritime Silk Road transportation network maintains strong overall resilience, with good adaptability and recovery capability. Resilience is generally higher in shipping segments closer to coastlines and exhibits marked spatial heterogeneity and periodic fluctuations. The proposed model provides decision support for optimizing transport resource allocation and enhancing maritime security cooperation along the Maritime Silk Road.

Keywords:

Maritime Silk Road

network resilience

graph neural network

spatiotemporal dynamics

transportation network

GNN-based spatiotemporal resilience prediction for Maritime Silk Road transportation network

Abstract

Journal

Organization