Ship Collision Risk Model Based On Data Mining Method

The motion of mandatory installation of automatic identification system of Automatic Identification System(AIS)was passed in the MSC 105(73)meeting of the International Maritime Organization.Mass data of AIS information is recorded every day by AIS data stations around the world,the relevant technologies and applications are enhanced consequently.In recent years,the development of data mining suits to the explosive growth of the data capacity,and the data mining technique can analyze mass and complicate AIS data rapidly and efficiently.In this study,various parameters in AIS data base concerning ship collision are organized and summed up,to study the collision risk of ships when sailing in specific area.An improved ship collision risk evaluation model is proposed based on the original ship collision risk model and the inclusion of new variables in the model.The AIS records of ships in Laotieshan Channel are selected for validation purpose in this study.Figures and tables are used to illustrate the regularities of spatial and temporal distribution of ships collision risk in Laotieshan Channel.The study are intend to provide references for reducing the accidents in specific time and locations,and alleviating the working force of seafarers as well as improving the scientificity of maritime supervision.In this paper,based on the research and study of many scholars and experts,a ship collision risk model is established on the basis of the identification method of collision risk based on fuzzy pattern recognition method.This model takes into account the collision factors DCPA,TCPA,and the ship speed ratio V,according to the different navigation environment conditions,the maritime administration departments of traffic control measures,and taking the possible impact of different navigable waters on ship collision risk into consideration,the risk degree of ship collision in different navigable waters is studied,and the risk degree of collision between ships in different waters is investigated.This model takes DCPA as the analysis reference standard,the collision risk grade division in different navigable waters is preliminarily obtained.This study aims to improve the traditional fuzzy pattern recognition model of collision risk,in order to provide reference for the safety of ship navigation and the reduction of marine traffic accidents.