复杂水域船舶拟人智能避碰避险决策关键技术研究

Maritime accidents like vessel collision and grounding still frequently occurred in the water area administrated by vessel traffic service (short form for VTS), which take disastrous consequences to the Marine environment and life and property. To make the ships sail more safely and the sea more clean, based on the early research foundation of personifying intelligent decision making for vessel collision avoidance (short form for PIDVCA) in open water, the mechanism of personifying intelligent decision-making for vessel collision and obstacle avoidance (short for PIDVCOA )under complex Waters is studied, using the thinking and methods of artificial intelligence and the ship handling and Collision Prevention knowledge, which is realized by applying Research method combining subjective and objective. A dynamic Evaluation system for predicting the risk of collision and grounding of ship and the effect of avoiding collision and avoiding risks is rreated by associated collision avoidance behavior. Early warning model is established, and PIDVCOA algorithm is also designed, revealing rules of relative movement of ship encountered situation. Creating simulation test platform based on the remote real-time traffic flow, setting test method of algorithm and appling analogy traffic simulation and real traffic flow simulation test，rationality of the intelligent model will be verified and completeness of PIDVCOA algorithm with conditions will be proved. The fruits of this research can promote the intelligence of the domestic VTS system and autonomous ship collision avoidance, achieve the original source of the innovation, further expand the scope of intelligent ship collision avoidance decision making technology and enhance the risk warning and decision support function for the piloting system. The value of this research is of great importance and significance and has a good application prospect.

船舶碰撞、搁浅和触礁事故在VTS中心管辖水域仍时有发生，给海洋环境与生命财产造成了灾难性后果。为使船舶航行更安全、海洋更清洁,在开阔水域船舶拟人智能避碰决策（PIDVCA）基础上，运用人工智能、船舶操纵与避碰领域知识，采用主观与客观相结合的研究方法，分析复杂水域船舶避碰避险决策（PIDVCOA）机理，揭示船舶会遇相对运动内在规律，创建关联避碰行为的船舶碰撞搁浅触礁危险度及避碰避险行为效果预测动态评价体系，建立预警模型，设计PIDVCOA算法；创建基于远程实时交通流的仿真测试平台，采用模拟与真实交通流相结合的仿真试验，开展智能预警模型合理性验证以及有约束条件下PIDVCOA算法的完备性证明。研究成果对促进国产VTS设备的智能化和自主船自动避碰技术取得源头创新、拓展船舶智能避碰辅助决策技术适用范围、增强引航系统的危险预警及辅助决策功能都具有重要意义和研究价值，项目研究成果具有良好的应用前景。

船舶碰撞、搁浅和触礁事故在VTS中心管辖水域仍时有发生，给海洋环境与生命财产造成了灾难性后果。为使船舶航行更安全、海洋更清洁,在开阔水域船舶拟人智能避碰决策（PIDVC A）理论基础上，研究复杂水域船舶避碰避险决策模式，运用人工智能、船舶操纵与避碰领域知识，采用理论分析、专家访谈、咨询与问卷调查、数学建模、模拟与真实交通流相结合的仿真试验研究方法，揭示船舶会遇相对运动几何内在规律，建立船舶碰撞及搁浅触礁危险及其危险度评判阈值模型，构建智能预警模型，设计复杂受限水域PIDVCA算法；创建基于远程实时交通流的仿真测试平台，以上海港和镇扬气度水域为复杂水域研究对象，开展智能预警模型合理性验证以及PIDVCA算法的完备性证明。.通过本项目理论、模型及算法研究与仿真验证与第三方自测试研究结果形成以下重要结论：（1）分别创建了关联避碰行为的船舶碰撞、搁浅危险及危险度评估体系，能实现至少3~4级预警，示范应用项目基于船岸协同的镇扬汽渡渡轮航行避碰智能预警系统远程录屏测试分析结果表明，预警准确率达到99.5%；（2）创建船舶拟人智能避碰避险决策理论，提出航道交汇区船舶避让责任划分模型，构建港口水域不同区域、不同危险等级初始避碰决策库并设计PIDVCOA算法模块，通过大量仿真实验验证了基于理论成果研发的PIDVCA软件，已能处理开阔水域12海里海域内多达20艘船舶交通密度下，仅用2ms自动生成避碰决策实施方案，可安全避让其中2～3艘危险目标船，显示了研究方法和研究成果有了质的突破；（3）基于PIDVCA算法，与企业共同开发的船舶智能航行避碰辅助决策系统成功应用于工信部智能船1.0专项的4艘超大型远洋商船，首次在远洋商船实现智能预警及避碰辅助决策功能。研究成果对促进国 产VTS设备的智能化技术和船舶智能化航行技术取得源头创新、拓展船舶智能避碰辅助决策技 术适用范围、增强引航系统的危险预警及辅助决策功能都具有重要意义和研究价值，项目研究具有广阔的应用前景。