Intelligent anti-collision in multi-vessel scenarios: a spatial DPAD decision-making model via accident data-driven

One of the core technologies for water transportation safety is the intelligent anti-collision technology for vessels. Although significant progress has been made in the research of intelligent anti-collision for vessels, the problem of multi-vessel adaptive anti-collision in complex waters has not been fully explored. In view of this, this paper proposes an intelligent decision-making model for multi-vessel anti-collision based on the Domain Predicted Area of Danger (DPAD). Initially, concerning the mechanisms of vessel collisions and in conjunction with the requirements of the International Regulations for Preventing Collisions at Sea (COLREGs) regarding vessel maneuvering actions, a Predicted Collision Risk Area (PCRA) model based on DPAD is constructed. Subsequently, to address the requirements for resuming navigation post-anti-collision, a threshold for resumption evaluation based on Potential Collision Risk (PCR) is proposed. Additionally, Grey Wolf Optimization (GWO) and Bézier curve techniques are employed to optimize the recovery navigation path schemes after anti-collision. Finally, by incorporating representative vessel collision cases, complex scenarios involving single-vessel and multi-vessel encounters are constructed to validate the intelligent anti-collision decision model based on DPAD, thereby achieving adaptive control of vessel avoidance maneuvers. The application results show that the spatial DPAD-based anti-collision decision model is reliable during multi-vessel encounters, producing avoidance outcomes that satisfy maritime safety requirements. This model provides a novel approach to intelligent vessel anti-collision by synergistically addressing adaptive anti-collision in spatial multi-vessel encounters.