基于RBF神经网络的空间飞网机器人控制策略设计与改进

Design and improvement of control strategy for space flying net robot system based on RBF neural network

针对空间飞网机器人清理空间碎片的任务,利用RBF神经网络和滑模控制理论研究了输入受限情况下的稳定控制问题。首先基于图论和质量集中法建立了空间飞网机器人系统的动力学模型,并进行了动特性分析。接着考虑机动单元质量和干扰上界未知的情况,基于RBF神经网络方法设计了直接滑模控制器,可保证输入不受限下空间飞网机器人系统快速稳定。然后考虑执行机构物理约束,引入辅助系统,改进了基于RBF神经网络的直接滑模控制方法,证明了输入受限时系统的稳定性。最后通过数值仿真验证了所设计控制策略的有效性。

Aiming at the task of active debris removing by space flying net robot system(SFNRS),the stability control problem under input constraints is studied by using RBF neural network and sliding mode control theory.Firstly,the dynamic model of SFNRS was established based on graph theory and lumped mass method,and the dynamic characteristics were analyzed.Then,considering the unknown of maneuvering units’mass and bound upper of disturbance,a direct sliding mode controller was designed based on RBF neural network,which could ensure the fast stability of SFNRS with unrestricted input.Further,considering the physical constraints of the actuator,the auxiliary system was introduced,the direct sliding mode control method based on RBF neural network was improved and the stability was proved.Finally,the effectiveness of the designed control strategy was verified by numerical simulation.