基于RBF神经网络的摩擦补偿建模与控制

Compensating Modeling and Control of Friction Based on RBF Neural Network

机械系统摩擦的精确数学模型很难建立,因此,尝试采用RBF神经网络系统在线逼近摩擦模型并将辨识结果作为控制算法的补偿项。在控制方法上,采用了基于RBF神经网络系统补偿的PD算法。在系统证明上,从李雅普诺夫函数中导出了自适应参数并且分析了闭环系统跟踪误差的有界性。利用Matlab对提出的方法及证明的有效性进行了验证。

An application of a radial basis functions neural networks for compensating the effects induced by the friction in mechanical system is presented. A neural networks system based on radial basis functions is employed, and a bound on the tracking error is derived from the analysis of the tracking error dynamics. The hybrid controller is a combination of a PD controller and a neural networks controller which compensates for nonlinear friction. The proposed scheme is simulated on a single link robot control system. Theory proof and simulation test both show the availability of proposed method.