航空学报 > 2019, Vol. 40 Issue (5): 122488-122488   doi: 10.7527/S1000-6893.2018.22488

基于神经网络自适应动态逆的结冰飞机飞行安全边界保护方法

魏扬1, 徐浩军1, 薛源1, 郑无计1, 李哲1, 裴彬彬2   

  1. 1. 空军工程大学 航空工程学院, 西安 710038;
    2. 空军石家庄飞行学院 第四训练旅, 保定 074212
  • 收稿日期:2018-06-27 修回日期:2018-07-18 出版日期:2019-05-15 发布日期:2018-09-19
  • 通讯作者: 薛源 E-mail:szxy1986@163.com
  • 基金资助:
    国家"973"计划(2015CB755802);国家自然科学基金(61503406);民用飞机专项研究(MJ-2015-F-019)

Aircraft flight safety envelope protection under icing conditions based on adaptive neural network dynamic inversion

WEI Yang1, XU Haojun1, XUE Yuan1, ZHENG Wuji1, LI Zhe1, PEI Binbin2   

  1. 1. Aeronautics Engineering College, Air Force Engineering University, Xi'an 710038, China;
    2. Fourth Training Brigade, Shijiazhuang Flighting College of Air Force, Baoding 074212, China
  • Received:2018-06-27 Revised:2018-07-18 Online:2019-05-15 Published:2018-09-19
  • Supported by:
    National Basic Research Program of China (2015CB755802); National Natural Science Foundation of China (61503406); Civil Aircraft Special Research (MJ-2015-F-019)

摘要: 考虑到飞机带冰飞行的安全问题,对结冰飞机进行安全边界保护成为一种有效的解决手段。基于神经网络自适应动态逆跟踪性能好、鲁棒性强的优点,提出了以安全关键飞行参数限制值作为神经网络自适应动态逆的输入,获取可用舵面偏转角的边界保护方法。建立了飞机本体动力学模型,采用高精度的数值模拟方法获得结冰数据库。设计了神经网络自适应动态逆控制律,通过在动态逆环节引入单隐层神经网络,对不确定性逆误差进行自适应补偿,增强了控制系统的鲁棒性。以俯仰姿态保持模式为例设计了结冰飞行闭环安全边界保护系统。以结冰飞机最小平飞速度的估算值作为飞机最低飞行速度,设计自动油门控制系统,实现对飞行速度的保护。通过仿真验证了设计的控制律具有较强的鲁棒性。对结冰严重程度线性增加情形下飞机状态参数的动态响应进行了分析。仿真结果表明,所设计的结冰边界保护系统,能够实现飞机在容冰飞行过程中对安全关键参数如迎角、飞行速度的实时保护。

关键词: 飞机结冰, 边界保护, 神经网络自适应动态逆, 俯仰姿态保持, 安全关键参数

Abstract: Flight safe envelope protection of iced aircraft has become an effective solution to the flight safety under icing conditions. Based on the good tracking performance and strong robustness of the adaptive neural network dynamic inversion control, an envelope protection method is proposed to obtain the available deflection angle of control surface by using the limit value of the key flight safety parameters as input of control loop. The dynamic model of the aircraft is established, and a high precision numerical simulation method is used to obtain the icing aerodynamic database. The adaptive neural network dynamic inversion control law for an aircraft is designed. By introducing the single hidden layer neural network in dynamic inversion, the uncertainty inverse error is compensated adaptively, and the robustness of the control system is enhanced. The closed loop envelope protection system in pitch attitude hold mode under icing conditions is designed. An automatic throttle control system is designed on the assumption that the minimum flying speed is the same as the estimated value of the minimum level of the flight speed and the design is used to protect the flight speed. The simulation results show that the designed control law has strong robustness. The dynamic response of the aircraft state parameters is analyzed in the case of the linear increase of icing severity. The simulation results show that the designed envelope protection system can realize the real-time protection of the key security parameters such as the angle of attack and the speed during the ice-tolerant flight.

Key words: aircraft icing, envelope protection, adaptive neural network dynamic inversion, pitch attitude hold, key security parameters

中图分类号: