自适应神经变结构的机器人轨迹跟踪控制

提出一种神经网络与变结构融合的控制策略用于非线性机器人控制,该方案利用神经网络来自适应补偿不确定模型,并通过变结构控制器消除逼近误差.考虑到局部泛化网络的不足,根据其状态空间的划分,分别对3个区间采用神经网络与变结构的分级与集成控制.该方案能在控制阶段初期及网络逼近区域外使两种控制器共同起作用以保持系统的强鲁棒性,基于Lyapunov理论证明了闭环系统的全局稳定性.仿真结果进一步表明了该方法的优越性.

Neural-variable structure-based adaptive trajectory tracking control of robot manipulators

The trajectory tracking of a class of robot manipulators with uncertainties is considered. The syncretic control
algorithm is proposed by adaptive neural network and variable structure. Neutral network is used to adaptivly learn and
compensate the unknown system, and approach error as disturbance is eliminated by using variable structure controller.
Considering the shortage of local network, based on partition of state dimensional, neural network and variable structure
separate control is applied to three sections with classification and integration. Two controllers together keep the robust of
system in control initial stages and outside of approach region. The controller can guarantee good robustness and the stability
of closed-loop system based on Lyapunov. The simulation results show the effectiveness of the presented methods.