宇航学报

宇航学报 ›› 2019, Vol. 40 ›› Issue (10): 1224-1233.doi: 10.3873/j.issn.1000-1328.2019.10.014

• 制导、导航、控制与电子 • 上一篇    下一篇

远程操控飞行器自适应神经网络观测器设计

许红羊，李宏君，凡永华，闫杰   

  1. 1. 西北工业大学航天学院，西安 710072；2. 西安航天动力研究所，西安 710100
  • 收稿日期:2019-06-14 修回日期:2019-06-27 出版日期:2019-10-15 发布日期:2019-10-25

Adaptive Neural Network Observer Design for a Remotely Piloted Vehicle

XU Hong yang, LI Hong jun, FAN Yong hua, YAN Jie   

  1. 1. School of Astronautics, Northwestern Polytechnical University, Xi’an 710072, China; 2. Xi’an Aerospace Propulsion Institute, Xi’an 710100, China
  • Received:2019-06-14 Revised:2019-06-27 Online:2019-10-15 Published:2019-10-25

PDF

290

摘要：

针对无人飞行器远程操控系统设计时，由于远程操控飞行器动力学的非线性和飞行器控制系统性能的不确定性，无法精确建立远程操控飞行器控制系统模型的问题，提出了一种自适应神经网络状态观测器设计方法实现对远程操控飞行器的控制系统模型的估计。首先将飞行器的动力学环节与自动驾驶仪构成的闭环回路作为一个整体建立了远程操控飞行器控制系统的非线性模型。然后针对模型中存在未建模动态的问题，采用神经网络算法对非线性动力学模型进行在线辨识，并引入鲁棒项对附加扰动进行抑制。最后设计自适应律对神经网络的权值进行实时调整，保证了系统的稳定性，并基于Lyapunov理论证明了观测器的估计误差是最终一致有界的。仿真结果表明，所设计的观测器能够保证远程操控飞行器在存在未建模动态和附加扰动的情况下对飞行状态具有良好的估计性能。

关键词: 远程操控飞行器, 自适应律, 神经网络, 观测器

Abstract:

Aiming at the problem that it is difficult to establish the model of the control augmentation system of a remotely piloted vehicle (RPV) accurately due to the nonlinear dynamics of the RPV and the uncertainties of the performance of the RPV control augmentation system, an adaptive neural network state observer is proposed to approximate the model of the RPV control augmentation system. The closed-loop system composed of the RPV dynamics and control augmentation system is taken as a whole, and the nonlinear model of the whole system is established. To deal with the unmodeled dynamics, a neural network algorithm is proposed to identify the nonlinear dynamics model online, and a robust term is induced to suppress the disturbance. Meanwhile, to guarantee the stability of the overall observer system, an adaptive law is designed to turn the neural network weights. Moreover, the overall adaptive observer scheme is proved to be uniformly and ultimately bounded. The simulation results show the effectiveness of the adaptive neural network observer in the presence of the unmodeled dynamics and external disturbance.

Key words:  Remotely piloted vehicle (RPV), Adaptive law, Neural network, Observer

中图分类号: