基于RBF网络的机械手迭代学习控制器

Iterative Learning Controller for Manipulators Based on the RBF Network

针对迭代学习控制用于机械手轨迹跟踪时存在的收敛速度慢的问题,提出了一种基于RBF网络的迭代学习控制器,利用先前跟踪不同期望轨迹所得的经验构造新的初始控制量以加快收敛速度.将给定的期望跟踪轨迹分解成多个查询点,然后用RBF网络对每个查询点周围最邻近的k个数据点进行拟合以建立系统的逆动力学特性估计,进而预测相应于查询点的初始控制输入.为验证所提方法的有效性,对一平面双连杆机械手进行了仿真研究.

In view of the slow convergence speed of an iterative learning controller in the trajectory tracking of manipulators, a kind of new iterative learning controller based on RBF neural network is proposed from considerations of past experience in tracking various trajectories so as to select the initial control input of iterative learning controller properly. A new desired trajectory can be decomposed into many query points at first, and the RBF network is applied to construct inverse dynamics of the manipulator by fitting the nearest k data points around each query point and then predicting the initial control input. The method of control is verified by computer simulation for a planar two-link manipulator.