船舶拟人智能避碰决策算法测试及应用  

Test and Application of Personifying Intelligent Decision-Making Algorithm for Vessel Collision Avoidance

作　　者：李丽娜[1] 陈国权[1] 杨凌波 许春亮 王兴华 文婷 LI Lina;CHEN Guoquan;YANG Lingbo;XU Chunliang;WANG Xinghua;WEN Ting(Navigation School,JiMei University,Xiamen 361021,China;Beijing Highlander Digital Technology Co.,Ltd.,Beijing 102200,China)

机构地区：[1]集美大学航海学院,福建厦门361021 [2]北京海兰信数据科技股份有限公司,北京102200

出　　处：《中国航海》2022年第1期1-7,共7页Navigation of China

基　　金：国家自然科学基金(51879119)。

摘　　要：论述船舶拟人智能避碰决策(Personifying Intelligent Decision-Making for Vessel Collision Avoidance, PIDVCA)的算法原理及其基础模型的构建方法;提出一套避碰算法合理性及完备性的测试方案,利用船舶智能操控仿真测试平台验证算法的“拟人”特性;简要介绍基于PIDVCA算法开发的航行避碰辅助决策系统(Navigation-Collision Avoidance Decision-Making Assistant System, N-CADAS)的主要功能及其海上试验与远洋商船示范应用情况。结果表明:该算法具备在线自主学习和“拟人化”解决问题的能力,可实现对会遇船舶的自动感知、碰撞危险与危险度的自动认知和避碰决策问题的自动求解及优化,为驾驶员提供开阔水域多目标船会遇危险预警与避碰辅助决策支持。研究成果对确保船舶航行安全、推动船舶航行智能化及无人船的发展具有重要意义。The principle of the PIDVCA(Personifying Intelligent Decision-Making for Vessel Collision Avoidance) algorithm and the construction of its basic models are introduced. The testing scheme for the rationality and completeness of the algorithm is formulated and a testing platform is developed for verifying the performance in imitating human action. The major functions of the N-CADAS(Navigation-Collision Avoidance Decision-Making Assistant System) developed according to the PIDVCA concept are introduced, which is able to perceive the traffic situation, recognize the risk of collision and propose solution schemes automatically. The results of sea trial and application of the system on an ocean-going ship are pre-sented.

关 键 词：PIDVCA 决策 预警 碰撞危险度 机器学习 

分 类 号：U675.96[交通运输工程—船舶及航道工程]