基于递归模糊神经网络的机器人鲁棒H_∞跟踪控制

利用递归模糊神经网络来逼近机器人系统中的非线性函数,提出了一种具有自适应能力的H∞控制策略.该控制策略能够减弱机器人系统的外扰,并把模糊神经网络的重构误差对系统的影响控制在指定的范围内.同时又能保证闭环系统的所有信号都是有界的.为了验证基于递归模糊神经网络的H∞控制策略的有效性,将其与计算力矩控制方法进行比较,仿真结果表明,在存在外扰的情况下,所提出的控制策略具有比计算力矩控制方法更好的跟踪性能.

Robust H-infinity tracking-control for robotic system based on recurrent fuzzy-neural-networks

Using recurrent fuzzy-neural-networks(RFNN) to approximate the nonlinear functions in a robotic manipulator system, we develop an adaptive H-infinity controller. The proposed controller can attenuate the effect of external disturbance and reduce the reconstruction-error of the recurrent fuzzy neural network to a prescribed level. Meanwhile, it also ensures all signals in the closed-loop system to be bounded. Simulation experiments of this control strategy are performed; the results show that this control strategy has better tracking-performance than the computed-torque-control method under external disturbances.