%0 Journal Article %T 基于历史数据的空中改航算法研究 %A 陈正茂 %A 林毅 %A 杨波 %J 工程科学与技术 %D 2018 %R 10.15961/j.jsuese.201700538 %X 中文摘要: 改航是航班运行过程中应对恶劣天气、军事活动、流量管控等突发情况的重要策略。作者提出一种挖掘历史雷达数据特征的改航算法。算法首先综合考虑改航路径的经济性、安全性、航路特征和容量影响等因素建立改航约束条件,对航班可能的改航区间的航路通路进行分析初选,确定最优的改航区间及其满足改航约束条件的一系列关键点序列。随后基于航班在历史雷达数据中的位置信息分析关键点序列航段执行情况,结合该航段计划航班数据,提出了适用于改航关键点选择的航段利用率计算方式,并在此基础上确定改航路径的最优关键点序列。以航班飞行过程中的速度矢量为建模对象,通过机器学习算法挖掘经过关键点序列航段的航班飞行运动模式。采用混合高斯模型对航班速度矢量建模和参数学习;利用马尔科夫蒙特卡洛采样算法预测航班改航过程中的速度矢量序列。最后利用匀加速运动学方程预测出完整的改航路径。通过某大型流量管理系统的应用实践证明,在考虑了改航约束条件和航段历史运行情况(利用率)等因素下的改航策略符合航班的实际运行情况。同时算法还能以高精度预测航班的改航轨迹,为改航路径相关区域的流量管理提供数据支撑。本文的改航算法能很好的解决航班在飞行过程中的改航问题。</br>Abstract:Rerouting is an important strategy to deal with bad weather, military activities, traffic flow control and other emergency situations in the process of flight operation. An improved rerouting algorithm based on mining features of historical radar data is proposed. Firstly, reroutingconstraints are established by taking the economy, security, route features, and the impact of traffic flow and capacity into consideration. A rough process is made to select the optimal rerouting zone and possible key-point sequences which meet those constraints for rerouting.The routes duringhistorical flight operations are analyzed by the positions of historical flights. Combining with the flights whose plan path contain the given route, a utilization representation is proposed to select the preferred key point. Then, the optimal key-point sequence is obtained by analyzing its utilization from historical radar data. Taking the velocity vector as the modeling object, the machine learning algorithm is used to mine the flight pattern of the route segment. Gaussian Mixture Model is applied to model the distribution of velocity, and Markov chain Monte Carlo is used to predict the velocity sequence during rerouting. Finally, the whole rerouting path is planned by kinematics equation with constant acceleration rules. The application of a large-scale traffic management system demonstrates the high practicability ofthe proposed algorithm with considering the actual operating conditions of flight. The proposed approach can also predict the trajectory of rerouting flights, which provides the data supportfor the air traffic flow management in the rerouting path areas. Therefore, the rerouting issue of flight can be solvedreasonably and efficiently. %K 改航 约束条件 历史雷达数据 关键点序列 机器学习< %K /br> %K reroute constraints historical radar data key point sequence machine learning %U http://jsuese.ijournals.cn/jsuese_cn/ch/reader/view_abstract.aspx?file_no=201700538&flag=1