面向飞行安全的低空救援空中交通辅助决策系统关键问题研究

As the most effective means widely used in many countries around the world, aeronautical rescue has lots of advantages, such as rapidness, efficiency and less geographical space constraints, and so on. However, due to the interference of the uncertainties (terrain environment, intensive flights. etc), the difficulty of rescue is increasing. Therefore, we tend to bungle rescue opportunity and even lead to secondary accidents. Low-altitude rescue air traffic assistant decision-making system can effectively improve the level of security and management capabilities of the rescue. Faced with the three scientific issues consists of the time variability of conflict detection, the reliability of independent hedge and the dynamic nature of resource scheduling exist in this system. This duty aim at the changeable security posture of the aircraft running status, use for reference molecular motion ideas, based on the space-time parameters and performance indicators of aircrafts, we establish the conflict detection model of the multi-aircraft variable height maneuvers; for the complexity of human-machine coordinated reliable operation, we introduce the theory of situated cognition and integrated use of the voronoi diagram and multi-agent method to determine multi-aircraft intensive flight reliable independent hedge model; due to dynamic change of demand, supply and distribution of the resource, we take the limitations of space-time mapping functions and the airborne fuel as the constraint as well as the resource demand information as the search criteria. In the mean time, we use the means of controlling data mining to find a reasonable resource dynamic scheduling rules. The study will lay the theoretical and technical basis to establish an impeccable low-altitude rescue air traffic assistant decision-making system.

航空救援具有快速、高效、受地理空间限制少等优势，是世界许多国家普遍采用的最有效手段。但航空器由于受地形环境、密集飞行等不确定因素的干扰，施救难度增大，往往贻误施救时机，甚至导致二次事故。低空救援空中交通辅助决策系统能有效提高救援的安全水平和管理能力。面对该系统存在冲突探测的时变性、自主避险的可靠性、资源调度的动态性三个科学问题，本项目针对航空器运行状态时变的安全态势，借鉴分子运动思想，基于航空器时空参量和性能指标，建立多机变高度机动飞行的冲突探测模型；针对复杂人机协同可靠运行，引入情境认知理论，综合运用Voronoi图和多Agent方法，确定多机密集飞行可靠自主避险方法；针对资源需求、供给和分布状况等信息动态变化，以时空映射的限制函数和携带油量为约束，以资源需求信息为搜寻条件，采用控制挖掘手段，找到合理的资源动态调度规则。该研究将为建立完善的低空救援空中交通辅助决策系统奠定理论与技术基础。

我国是世界上发生自然灾害以及各类公共安全事故灾害最为严重的国家之一，而航空救援具有快速、高效、受地理空间限制少等优势，已然成为社会各界高度关注和重视的对象。本课题将低空救援航空器航迹规划问题划分为预战术和战术两个层面进行研究。预战术层面通过结合实际的低空救援飞行环境，先针对单个航空器航迹规划问题，在三维空域网格划分的基础上，基于改进的算法得到最优的初始飞行计划航迹。在单个航空器初始飞行航迹基础上，结合时间窗原理，提出了两种方法解决了多航空器无冲突飞行计划航迹规划问题，并通过计算机仿真对两种方法进行了对比分析。战术层面则在预战术航迹规划的基础上，结合已划设的空域网格，提出了基于航空器飞行状态模式估计和飞行意图推理的航迹预测算法。该算法结合了交互式多模型算法和改进的意图推理算法的优点，摒弃了两种算法单独使用时的不足之处，进而能够确保预测得到的航迹实时准确。基于航迹预测的冲突解脱算法在探测到飞行冲突时，为了实现航空器自主避让飞行冲突，将航空器当作独立自主的智能体Agent，依据救援目视飞行规则，在Airspace模型上，构建Agent模型。基于优先级准则，进行非协作式航向调整冲突解脱策略，并提出了协作式航向调整冲突解脱策略思想，确保航空器能够得到最优无冲突飞行航迹。在低空救援的智能调度方面，将航空救援调度问题划分为战略和战术两个层面。战略层面通过结合实际的航空器救援调度实施场景，针对耐用品对应急时间要求高及快速消费品连续消耗的不同特性，给出多出救点多受灾点组合调度模型，采用汶川地震救灾数据，结合基本的和改进的遗传算法对模型进行求解，并对二者的性能进行对比。战术层面则通过采用不同缓冲需求处理方法解决了快速消耗品和耐用品需求差异问题，运用模糊聚类方法对灾区进行分组并确定各受灾点调度优先级，结合汶川地震区县级救援案例，通过构建实时调度规划模型算法解决了航空器路径规划和配送方案生成的问题。