摘要
针对航天器编队通信受限和空间受摄下的构形维持控制问题,研究了一种基于学习切比雪夫神经网络(Chebyshev Neural Network,CNN)和动态事件触发(Dynamic Event Triggering,DET)机制的控制方法。为了补偿空间干扰的影响,研究了一种基于迭代学习算法的学习CNN模型,与传统CNN模型相比,该模型计算资源占用更少。然后,基于滑模控制理论和图论知识设计了一种终端滑模控制器。最后,为所提控制器设计了一种DET机制。仿真结果表明,所提出的控制方法能实现高精度的构形维持,并且有效减少编队集群间的通信频率超过92.01%,同时避免芝诺现象。
To address the issue of spacecraft formation maintenance under communication constraints and space disturbances,this paper investigates a control method based on the learning Chebyshev Neural Network(CNN)and a Dynamic Event Trigger(DET)mechanism.To compensate for the effects of space disturbances,a learning CNN model based on an iterative learning algorithm is proposed,which requires fewer computational resources compared to traditional CNN models.Subsequently,a terminal sliding mode controller is designed based on sliding mode control theory and graph theory.Finally,a DET mechanism is designed for the proposed controller.Simulation results demonstrate that the proposed control method achieves high-precision formation maintenance,effectively reduces inter-formation communication frequency by over 92.01%,and avoids the Zeno phenomenon.
作者
李庚欢
贾庆贤
宋婷
孙秀芹
LI Genghuan;JIA Qingxian;SONG Ting;SUN Xiuqin(College of Astronautics,Nanjing University of Aeronautics and Astronautics,Nanjing 210016;School of Astronautics,Northwestern Polytechnical University,Xi’an 710000;Shanghai Aerospace Control Technology Institute,Shanghai 201109;Shanghai Key Laboratory of Space Intelligent Control Technology,Shanghai 201109;Maanshan College,Maanshan 243100)
出处
《飞控与探测》
2025年第6期37-47,共11页
Flight Control & Detection
基金
微小型航天器快速设计与智能集群全国重点实验室基金(MS01240109)
南京航空航天大学科研与实践创新计划(xcxjh20241504)。
关键词
航天器编队
动态事件触发
神经网络
迭代学习算法
滑模控制
spacecraft formation
dynamic event trigger
neural network
iterative learning algorithm
sliding mode control
