Marine traffic profile for enhancing situational awareness based on complex network theory

Abstract

Enhancing Situation Awareness (SA) is essential for the safety of marine traffic. Many indicators are developed to evaluate the status of the traffic and to facilitate SA of the Vessel Traffic Service Operators (VTSOs), such as collision risk, traffic complexity, etc. Many of these indicators are based on the pairwise encounters between ships, without considering the influence from the third objects. However, the mechanism to look beyond the pairwise encounters is important for marine traffic management. In this paper, a framework for evaluating the Marine Traffic Situation (MTS) is developed, which treats the traffic as an entire object instead of pairs of ships. The complex network theory is employed and the traffic is modelled as a virtual network, called Marine Traffic Situation Complex Network (MTSCN). The topological properties of the MTSCN compose the state vector reflecting the profile of the MTS. The MTS having similar profiles are categorized as one traffic pattern using Random Forest Algorithm, which can help the VTSOs to understand the traffic situation and to take control measures. Simulations are introduced to investigate the sensitivity of the proposed method with respect to parameters, and real traffic data from Yangtze River is used to demonstrate the performance of the proposed method. The MTS patterns in upstream area and downstream area can be divided into 3 classes and 2 classes respectively. The results are consistent with expert evaluation results and show that this method can construct a state vector describing the traffic profile and identify different traffic patterns reflecting key characteristics of the traffic. Additionally, the identified profile and pattern enhance the SA of VTSOs by supporting decision making.