微纳卫星动量轮神经网络全局快速终端滑模控制  

Global Fast Terminal Sliding Mode Control of Neural Networks for Micro-nano Satellite Momentum Wheel

作　　者：范书珲 廖文和[1] 张翔[1] 李磊磊 FAN Shuhui;LIAO Wenhe;ZHANG Xiang;LI Leilei(School of Mechanical Engineering,Nanjing University of Science and Technology,Nanjing 210094,China)

机构地区：[1]南京理工大学机械工程学院,南京210094

出　　处：《宇航学报》2023年第11期1727-1735,共9页Journal of Astronautics

基　　金：十四五民用航天技术预先研究项目。

摘　　要：针对存在参数不确定性和外界未知干扰的微纳卫星动量轮转速跟踪控制问题,建立了包含模型不确定性和外部干扰的动量轮系统动力学模型,并提出了一种自适应神经网络全局快速终端滑模控制方法。首先,在无外部干扰情况下,设计了全局快速终端滑模控制策略以加快系统响应速度,实现系统有限时间稳定。其次,利用径向基函数神经网络对系统参数不确定性和外部干扰进行估计并补偿,并利用李雅普诺夫方法推导出神经网络自适应律,通过自适应调节网络权值确保系统在存在干扰的情况下实现系统稳定。最后通过仿真对比验证,本文设计的控制器不仅响应速度快,而且具有更好的抗干扰能力和鲁棒性。Aiming at the speed tracking control problem of micro-nano satellite momentum wheel with parameter uncertainties and external disturbances,a momentum wheel system dynamics model including model uncertainties and external disturbances is established.An adaptive neural network global fast terminal sliding mode control strategy is proposed.Firstly,a global fast terminal sliding mode control strategy is designed to accelerate the system response speed and achieve finite-time stability in the absence of external interference.Then,a radial basis function neural network is used to estimate and compensate for system parameter uncertainty and external disturbances.The neural network adaptive law is derived using Lyapunov method,and adaptive adjustment of the network weights is used to ensure system stability in the presence of disturbances.Finally,the simulation comparison shows that proposed control scheme not only has faster tracking,but also has better anti-interference ability and robustness.

关 键 词：微纳卫星 动量轮 神经网络 全局快速终端滑模 

分 类 号：V448[航空宇航科学与技术—飞行器设计]