摘要
基于神经网络自适应控制方法,研究了变体跨介质航行器俯冲过程航迹角控制问题。针对航迹角动态方程中出现的纯反馈形式,采用微分滤波的方法克服"代数环"问题;采用最小学习参数方法减少神经网络参数在线更新个数,降低计算载荷;采用动态面方法克服反步法虚拟控制的"微分爆炸"问题。仿真结果表明,所设计的控制器能够实现航行器在俯冲过程中对目标航迹角的跟踪,跟踪误差小于2°。
Based on the neural network adaptive control method,the problem of flight path angle control during water-entry diving process was designed for the trans-media aerial underwater vehicle.In view of the pure feedback form in the dynamic equation of flight path angle,the filter technique was used to overcome the ‘close loop’problem.The dynamic surface control( DSC) and minimal learning parameters(MLP) techniques were employed to solve the problems of ‘explosion of complexity’in the backstepping method and the online updated parameters being too much.The simulation results show that the designed controller can track the desired flight path angle well during diving process.The tracking error is less than 2°.
作者
谭骏怡
胡俊华
马宗成
陈国明
冯金富
TAN Jun-yi;HU Jun-hua;MA Zong-cheng;CHEN Guo-ming;FENG Jin-fu(School of Aeronautics Engineering,Air Force Engineering University,Xi'an 710038,China)
出处
《飞行力学》
CSCD
北大核心
2019年第1期34-38,49,共6页
Flight Dynamics
基金
国家自然科学基金资助(51779263)
关键词
最小学习参数
动态面
自适应神经网络控制
变后掠翼
minimal learning parameter
dynamic surface control
adaptive NN control
variable sweep wing