扇区空中交通风险态势网络建模与演化特征

为了准确感知空域内空中交通运行态势，提升飞行运行效能，研究了扇区空中交通风险态势的网络建模方法和演化特征，基于实测数据验证了方法的可行性与有效性；在扇区内以活跃航空器为网络节点，依据航空器位置偏差下的冲突关系构建连边，建立空中交通风险态势网络，利用连通分量的概念定义并识别网络内的航空器集群；基于航空器集群特征构建状态向量，进一步划分空中交通风险态势模式；以分类后的态势等级时序为基础，针对各模式的样本进行持续时间分析及建模，讨论了单一态势模式的生存特性与多模式间转移行为的偏好水平；为了验证方法的有效性，以广州管制AR05号扇区的实际数据为样本展开分析。研究结果表明：低风险至高风险模式的生存特性差异明显，平均生存周期分别为82.49、118.11、75.77、90.51 s，中等风险模式是估计生命周期最长且生存率最高的模式；在模式迁移的过程中，集群数目与规模相对简单的风险模式主要表现为前向可达性较高，而回迁及跃迁概率低于0.05；复杂程度较高的风险模式则表现出了较为明显的回迁行为，且概率会随着时间步长的增加(30、60、90、120 s)逐渐趋于稳定。可见，建立的空中交通风险态势网络模型能够较好反映空中交通风险态势信息，提出的演化分析方法可以为空中交通运行提供有益参考，从而发掘空中交通演变的科学规律。

Network modeling and evolution characteristics for air traffic risk situation in sectors

To accurately capture the air traffic operation situation in the airspace and improve the operation efficiency of flight, the methods of network modeling for air traffic risk situation in sectors and its evolution characteristics were studied. The feasibility and effectiveness of the methods were verified based on the measured data. In the sectors, active aircrafts were abstracted as nodes, and the edges were established based on the conflict relationship under their position deviations to help build the air traffic situation network. The aircraft clusters were defined and detected by using the concept of connected components within the network. The state vector was established based on the characteristics of aircraft clusters to further classify the risk patterns of air traffic. On the basis of the classified time sequence of situation class, the duration sample for each pattern was analyzed and modelled. The survival characteristics for a single pattern and the preference level of transition among multiple patterns were discussed. To verify the validity of the proposed method, the data about Guangzhou control sector AR05 were used as the sample for analysis. Analysis results show that the survival characteristics of the low risk to high risk pattern differ significantly with mean life cycles of 82.49, 118.11, 75.77 and 90.51 s, respectively. Among them the medium risk pattern is the one with the longest estimated life cycle and the highest survival rate. In the process of pattern transition, the risk patterns with relatively simple and smaller size of clusters mainly show a higher forward accessibility, while the backward and leap probabilities are lower than 0.05. The more complex ones show more obvious backward transition behavior, and the probabilities gradually stabilize with increasing time steps (30, 60, 90 and 120 s). So, the established network model can well reflect the information on the risk situation of air traffic, and the proposed evolution analysis method can give valuable insights into air traffic operation, thus identifying the scientific laws of air traffic evolution.