基于死区补偿的神经网络自适应鲁棒控制

Neural Network Adaptive Robust Control based on Dead time compensation

机器人执行机构的死区非线性对系统性能有较大影响。本文采用具有快速学习能力的RBF神经网络代替经典补偿器中的BP增广网络,设计了死区补偿的RBF网络自适应鲁棒控制,不但可以大大减少系统参数,还可以使得网络的初始化工作清晰明确。同时引入了GL矩阵和GL乘法算子的数学概念,从而在数学上严格证明了n节关节机器人系统的稳定性问题。仿真结果表明,所提方法具有良好的跟踪性能和较强的鲁棒性。

Executive body of the dead nonlinear has greater influence on the system＇s performance. In this paper, the dead zone compensation of RBF network adaptive robust control were designed by Using the RBF neural network instead of classic compensator of BP network.It can greatly reduce the system parameters and also make the network initialization work clear. GL and the GL matrix multiplication operator were introduced and thus mathematically rigorous proof of the n section joint robot system stability.The simulation results show that this method has good tracking performance and strong robustness.