A Novel Time-Frame Regional Collision Risk Model Based on Dynamic Time Warping

The quantification of collision risk in the water area is crucial for marine traffic safety. However, most existing studies primarily adopt a statistical perspective or only consider collision risks at discrete time points, neglecting the dynamic nature and temporal characteristics of collision risk. To overcome the problem, this paper proposed a novel time-frame regional collision risk model based on dynamic time warping (DTW). The time series of the ships in the water area were generated by modeling the collision risk from the perspectives of geometric encounters and traffic situation pressure. Dynamic time warping was used to mine the eigenvalue of each sequence in representing collision risk. The time-frame regional collision risk can be obtained by synthesizing the collision risk eigenvalues through considering the contribution of each ship to the entire collision risk. To validate the proposed model, some experiments were carried out by using the real AIS data in the Bohai Strait. The results show the capability of the proposed model in reasonably and effectively identifying the time-frame regional collision risk and prove its merits compared with the traditional methods. Therefore, it can better assist the supervision and analysis of regional collision risk so as to further enhance marine traffic safety.