Vessel Anti-collision Forewarning based on Mean Impact Value and Random Forest

摘要：

　　In order to explore the inherent laws of ship collision and determine the optimal timing of ship collision avoidance,a Vessel Anti-collision Forewarning method was proposed based on mean impact value(MIV)algorithm and random forest(RF)algorithm.First,the initial index system for vessel anti-collision forewarning was established from vessel static information,vessel dynamic information and environment information by combining macro and micro risk of collision.Second,MIV feature selection algorithm was utilized to extract main factors from the initial index system.Finally,aiming at the shortcomings of the existing single classifier forecasting models,RF combined classifier was introduced to identify and forewarn the vessel collision risk situation at different encounter situation.Experimental results show that RF model has higher prediction accuracy,stability and generalization ability compared with BP neural network,support vector machine(SVM),classification and regression tree(CART)classifier model.

关键词： average impact value feature selection random forest combined classifier Anti-collision Forewarning

作者: Yishan Li Zhiqiang Guo Jie Yang Hui Fang

作者单位: Key Laboratory of Fiber Optic Sensing Technology and Information Processing,School of Information En School of Information Engineering,Wuhan University of Technology Wuhan,Hubei Province,China

会议类型: 国际会议

会议名称: 2017 IEEE 2nd Advanced Information Technology,Electronic and Automation Control Conference(IAEAC 2017)(2017 IEEE 第2届先进信息技术、电子与自动化控制国际会议)

会议地点: 重庆

会议语种:英文

页码: 346-350

在线出版日期: 2017-03-25（万方平台首次上网日期，不代表论文的发表时间）