•  
  •  
 

Abstract

In terms of waters, vessel density distribution is a significant factor to evaluate the complexity of marine traffic and the collision risk. In previous studies, scholars frequently discovered the high-density vessel clusters according to density-based algorithms. Nevertheless, these algorithms were normally based on Euclidean or Hausdroff distance, etc., in which the encountering situation was prone to be ignored. Apparently, the heavydensity vessels in the traffic separation scheme don’t have a high risk due to their well organization, while the micro-traffic relationships such as approaching, receding, head-on and crossing should be crucial factors in clustering. Therefore, this paper majorly focuses on the complexity of vessel couple and it’s clustering using data mining technology. The complexity model of vessel couple is improved by taking the following factors into consideration: length overall, distance, movement trend and crossing angle. On the basis of traffic complexity and risk factors analysis, a clustering method of ship to ship encountering risk is presented by proposing a new distance definition, which can more effectively calculate the complexity of a mass of ships in an area.

Download
COinS