航空学报 > 2021, Vol. 42 Issue (9): 324726-324726   doi: 10.7527/S1000-6893.2021.24726

管制-飞行状态相依网络演化过程

李昂1, 聂党民1, 温祥西1, 韩宝华1, 曾裕景2   

  1. 1. 空军工程大学 空管领航学院, 西安 710051;
    2. 中国人民解放军 95429部队, 昆明 650000
  • 收稿日期:2020-09-07 修回日期:2020-10-06 发布日期:2021-09-29
  • 通讯作者: 温祥西 E-mail:wxxajy@163.com
  • 基金资助:
    国家自然科学基金(71801221)

Evolution process of control-aircraft state interdependent network

LI Ang1, NIE Dangmin1, WEN Xiangxi1, HAN Baohua1, ZENG Yujing2   

  1. 1. Air Traffic Control and Navigation College, Air Force Engineering University, Xi'an 710051, China;
    2. Unit 95429 of the PLA, Kunming 650000, China
  • Received:2020-09-07 Revised:2020-10-06 Published:2021-09-29
  • Supported by:
    National Natural Science Foundation of China (71801221)

摘要: 动态、准确的管制系统运行态势预测是航空运输系统各相关单位开展协同决策的关键基础。基于航空器间的冲突情况、管制员对航空器的管控情况以及管制移交情况构建管制-飞行状态相依网络,探究、分析其演化规律,采用相关性分析和主成分分析证明了所选5项指标的合理性。设置自由飞行和固定航线飞行两种仿真场景,通过计算平均节点度、平均点强等拓扑指标的最大李雅普诺夫指数证明各时间序列均具有混沌特性,选择长短期记忆(LSTM)人工神经网络方法对各时间序列的演化规律进行预测,并与其他预测算法进行对比。仿真结果表明LSTM算法能对管制系统的演化过程进行有效的预测,且预测精度高于贝叶斯算法和支持向量机算法;在自由飞行条件下,5项指标的预测误差绝大部分在20%以内,固定航线飞行的预测效果优于自由飞行。

关键词: 相依网络, 预测, 演化过程, 李雅普诺夫指数, 长短期记忆(LSTM), 人工神经网络

Abstract: Prediction of dynamic and accurate operation situation of the control system is essential for collaborative decision-making among relevant units of the air transport system. In this paper, a model for interdependent network of aircraft control/flight status is established based on the conflict of aircrafts, control of aircrafts by controllers, and transfer of controllers. The evolution law of the interdependent network is explored. Correlation analysis and principal component analysis are used to prove the rationality of the five indicators selected. Two simulation scenes of free flight and flight on route are set up. It is proved that each time series is chaotic by calculating the maximum Lyapunov exponent of topological indexes such as average node degree and average point weight. The Long Short-Term Memory (LSTM) method is employed to predict the evolution law of each time series, and is compared with other prediction algorithms. The simulation results show that the LSTM algorithm can predict the evolution process of the control system effectively, and the prediction accuracy is higher than that of the Bayes algorithm. The LSTM algorithm also supports the vector machine algorithm. The prediction error of the five indexes with the LSTM algorithm is mostly within 20% under the condition of free flight, and the prediction effect for flight on route is better than that for free flight.

Key words: interdependent network, prediction, evolution process, Lyapunov exponent, Long Short-Term Memory (LSTM), artificial neural network

中图分类号: