摘要
针对飞行器在执行机构故障条件下配平能力受限的问题,本文提出了一种基于深度学习的故障情况下可配平能力快速预示方法。首先,建立飞行器气动力矩和执行机构故障模型,并给出飞行器旋转配平条件。其次,在不同执行机构故障情况下,采用基于二次规划的可配平能力求解方法,在迎角/侧滑角二维平面内进行遍历求解,得到当前故障情况下的可配平能力剖面,并采用8个特征点进行包络,同时为所提方法提供样本。再次,利用深度神经网络强大的拟合能力,从样本中提取故障和气动力矩信息作为网络输入,特征点的迎角和侧滑角的值作为网络输出,离线训练深度神经网络。利用训练好的深度神经网络根据当前故障信息实时计算可配平能力剖面。最后,通过仿真验证了所提方法的有效性和实时性。
Aiming at the problem of limited trimming ability of axisymmetric aircraft with X rudder under the condition of actuator faults,a method based on deep learning to quickly predict the trim capability under actuator faults is proposed.Firstly,the aerodynamic torque and actuator faults models of the aircraft are established,and the aircraft rotation and trim conditions are given.Then,in the case of different actuator failures,the method of solving the trimming ability based on quadratic programming is used to traverse the two-dimensional plane of the angle of attack/sideslip angle to obtain the trimming command profile under the current fault situation,adopt eight feature points for envelope,and at the same time,samples are provided for the proposed method.Next,using the powerful fitting ability of the deep neural network,the fault and aerodynamic moment information are extracted from the samples as the network input,and the angle of attack and sideslip angle of the feature points are used as the network output to train the deep neural network offline.After that,the deep neural network trained is used to calculate the trimming command profile in real time based on the current fault information.Finally,simulation results verify the effectiveness and realtime performance of the proposed method.
作者
武天才
王宏伦
刘一恒
杨志远
余跃
Wu Tiancai;Wang Honglun;Liu Yiheng;Yang Zhiyuan;Yu Yue(Beihang University,Beijing 100191,China;Beijing Aerospace Automatic Control Institute,Beijing 100854,China)
出处
《航空科学技术》
2023年第2期72-77,共6页
Aeronautical Science & Technology
基金
航空科学基金(2018ZC51031)
北京航空航天大学未来空天技术学院/高等理工学院卓越研究基金(230121205)。
