摘要
为了更好地解决机器人系统中存在的参数不确定和外部干扰的鲁棒控制问题,提出一种基于耗散性理论的神经网络自适应鲁棒控制器,首先应用无源性理论对名义模型设计镇定控制器,然后利用RBF神经网络自适应学习系统的不确定部分,将神经网络逼近误差作为外部干扰,基于H∞控制理论使干扰对系统输出的影响抑制到所要求的最小程度,并用Lyapunov稳定性理论推导出RBF神经网络的权重矩阵调节律以及相关的鲁棒控制器,证明了系统的全局稳定性。仿真结果表明,这种控制器对机器人系统可能受到的干扰具有较好的抑制能力,提高了系统的鲁棒性,实现了系统轨迹的快速准确跟踪,又能很好地消除控制器的抖振,进而提高机器人工作性能。
To the robust control problem with the model uncertainties and external disturbances of the robotic systems,a neural network adaptive robust control scheme is proposed.A stability controller based on the passivity theory is proposed for the nominal model.An RBF neural network is used adaptively learn the system uncertainties,taking the neural network approximate error as the external disturbances.Based on H∞ control theory,the influence of the disturbance on system output is suppressed to a small extent.By using Lyapunov stability theory,the weight matrix regulator of RBF neural network and correlation robust controller are derived.The system global stability is proved.The simulation results show that the controller subject to interference suppression improves system robustness,and achieves a fast and accurately tracking,and then enhances the robot operational performance.
出处
《控制工程》
CSCD
北大核心
2010年第6期853-855,共3页
Control Engineering of China
基金
河北省基金资助项目(F2007000223)
河北省科技厅指导性计划(072135140)
关键词
机器人
无源性
耗散性
神经网络
robot manipulators
passivity
dissipativity
neural network
