摘要
针对不确定性机器人系统轨迹重复跟踪问题,提出一种自适应神经网络迭代学习控制方法。将系统的不确定项描述为周期性和非周期性两部分,通过采用迭代学习算法对周期性不确定部分进行迭代学习,采用RBF神经网络对非周期性不确定部分的未知上界自适应学习,并引入低通滤波器(LPF)来消除滑模控制中出现的抖振现象。该控制方法不仅对系统的不确定性和有界外部扰动具有鲁棒性,而且使得整个系统在迭代域中是全局渐进稳定的。严格的证明和仿真结果表明了该控制策略的有效性。
Adaptive neural network iterative learning control is proposed for repetitive tracking control of robotic systems with uncertainties. The uncertain part of the system is described as periodica/and non-periodic parts. Iterative learning control algorithm is used to learn the periodical uncertainty, while RBF neural network is used to learn the unknown upper bound of system non-periodic uncertainty adaptively. Low pass filter （LPF） is utilized to eliminate the chattering of sliding mode control. The control law not only guarantees the robustness for the system uncertainties and the bounded external disturbances, but also guarantees that the system is asymptotical stable in iterative domain. The computer simulation results demonstrate that the expected control purpose can be achieved using the proposed algorithm.
出处
《武汉理工大学学报》
CAS
CSCD
北大核心
2009年第24期135-138,144,共5页
Journal of Wuhan University of Technology
