Advancements in digital technology have shifted the maritime safety paradigm from a reactive approach to one that is predictive and proactive. Artificial Intelligence (AI)—specifically machine learning, deep learning, and explainable AI—holds significant potential to support this transformation. Safety challenges are particularly complex in high-traffic waters due to dynamic ship interactions involving diverse vessel characteristics, maneuvering patterns, meteorological conditions, navigational constraints, and human factors. This study aims to develop a conceptual model for AI implementation to predict ship accident risks in high-traffic waters by utilizing Automatic Identification System (AIS) data, hydrometeorological conditions, vessel characteristics, and navigational indicators such as Distance to Closest Point of Approach (DCPA) and Time to Closest Point of Approach (TCPA). The research employs a design science research approach combined with a literature review covering AI and maritime safety studies from 2016 to 2026. The developed model comprises five key stages: data acquisition, preprocessing and trajectory reconstruction, risk feature extraction, classification/prediction using machine learning algorithms, and result presentation via an early warning system. The research synthesis indicates that AIS serves as a primary data source for modeling ship traffic behavior, while algorithms such as Random Forest, Support Vector Machine, Gradient Boosting, LSTM, ConvLSTM, and CNN offer varying potential depending on data characteristics and prediction objectives. Previous studies demonstrate that integrating weather data enhances accident risk prediction capabilities, whereas explainable AI approaches are crucial for fostering user trust and providing rationales understandable to masters, watch officers, Vessel Traffic Service (VTS) personnel, and control center operators. The proposed implementation model positions AI as a decision-support system rather than a substitute for human responsibility. The implementation is expected to enhance early detection of high-risk situations, strengthen situational awareness, and support safety-related decision-making in waters with high traffic density.
Copyrights © 2025