基于RBF神经网络的机器人鲁棒自适应控制器设计

Robust Adaptive Controller Design for Robot Based on RBF Neural Networ

摘要:

在考虑机器人系统中存在的模型不确定性的情况下,提出了一种基于RBF神经网络和反演技术的鲁棒自适应控制器的设计方法。首先,通过状态变换将机器人的模型转换为适用于反演技术的形式；然后,利用反演技术设计了鲁棒自适应控制器,用两个RBF神经网络分别对模型的不确定性进行了处理,并用Lyapunov稳定性理论推导出RBF神经网络的权重矩阵调节律以及相关的鲁棒项,证明了系统的全局稳定性；最后,进行了相应的仿真研究,验证了设计的正确性和有效性。

Abstract:

Considering the robot system with model uncertainties, a robust adaptive controller design scheme based on RBF neural network and backstepping was proposed. Firstly, the robot model was changed to fit for the form of backstepping design through state transformation. Then the robust adaptive controller was designed by backstepping technology. The model uncertainties were dealed through two RBF neural networks. By the use of Lyapunov stability theory, weight matrix regulator of RBF neural network and correlation robust terms were derived, system global stability was proofed. In the end, simulation research of the design verified correctness and validity.