基于多Agent仿真的冲突场景识别

Identification of conflict scenarios based on multi-agent simulation

构建了基于多Agent的空管运行仿真系统的基本框架,设计了航空器Agent、空管自动化Agent和管制员Agent.利用JADE平台,Java编程语言开发了仿真原型系统,该系统采用分布式结构,并导入全天的飞行计划数据,实现对全国交通运行态势的仿真.为了提高轨迹预测的准确性,考虑了风的影响.通过对管制员Agent记录的全天冲突数据的分析,可以得到易冲突扇区主要集中在我国的东部以及中南部区域,R343和A461为易冲突航路等.冲突数据的分析结果对于航路网络的规划、扇区划分以及飞行计划的优化有重要的实践意义.

This paper described the approach for creating a multi-agent based simulation system using JADE platform and Java language which can simulate whole day air traffic operation in China.Three kinds of agents were created to support the air traffic simulation.They are aircraft agent,ATC automation systems agent and air traffic controller agent.Wind data were modeled in the system result in accurate trajectory predictions.Through data mining the conflict data of whole day operation recorded by the air traffic controller agent,it is obvious that the vulnerable sectors mainly located in the east and south-central of China.Route R343 and A461 were easy to spell conflicts during the operation.The analysis result can be used to support the air route network planning,sector division and optimization of flight plan,which has a very practical significance and necessity.