Supply Chain Resilience in Maritime Logistics Networks Integrating Blockchain Technology and Machine Learning Disruption Prediction

Abstract

Maritime transportation, the cornerstone of global trade, faces significant resilience challenges due to inherent complexities and vulnerability to disruptions, often exacerbated by limited transparency and predictive capabilities. This research addresses the critical need for enhanced resilience by developing and evaluating an integrated framework combining blockchain technology for transparency and machine learning for disruption prediction. The objective was to quantify improvements in efficiency, predictive accuracy, and recovery capabilities within maritime logistics networks. A permissioned hybrid blockchain architecture using Hyperledger Fabric was implemented across eight international container terminals, interfacing with existing systems to create a transparent, immutable record of supply chain events. Concurrently, an Extreme Gradient Boosting (XGBoost) machine learning model was trained on five years of historical data to predict port congestion 72 hours in advance. The integrated framework's performance was validated through discrete-event simulations of disruption scenarios and controlled field tests. The blockchain component reduced average document processing time by 66% and dispute resolution time by 81%. The XGBoost model achieved 87% accuracy and 0.93 AUC-ROC in predicting congestion on test data. Simulation results indicated the integrated system reduced post-disruption container dwell time by 40% and improved resource allocation efficiency by 54%. Field tests corroborated these efficiency gains and predictive performance. The integration of blockchain and machine learning significantly enhances maritime supply chain resilience. This framework provides unprecedented transparency and enables proactive, data-driven responses to potential disruptions, fundamentally shifting maritime logistics from reactive to anticipatory operational models, thereby improving efficiency and robustness.