柔性机械臂自适应神经网络反演控制

Adaptive Control of Flexible Manipulators Based on Neural Network and Backstepping

柔性机械臂系统是一类具有强耦合性的高阶非线性系统,动力学模型中通常包含一定的结构不确定性。针对结构不确定性所导致的轨迹跟踪控制复杂问题,提出一种基于RBF神经网络的自适应反演控制方法。首先分离出系统的已知信息部分和未知信息部分;而后通过反演法设计系统的轨迹跟踪控制器;再采用RBF神经网络对系统模型中的未知信息和虚拟控制量导数进行近似,设计基于RBF神经网络的自适应控制律,并通过Lyapunov理论证明了系统的稳定性。最后通过数值仿并与传统PD控制进行对比,结果表明该方法轨迹跟踪性能更优,跟踪精度提高了80%以上,证明了控制方法的有效性。

Flexible manipulators system is a kind of high-order nonlinear system with strong coupling.The dynamic model usual-ly contains structural uncertainties.Aiming at the complex problem of trajectory tracking control caused by structural uncertain-ty,an adaptive backstepping control based on RBF neural network is proposed.Firstly,the known information part and unknown information part of the system are separated;Then,the trajectory tracking controller of the system is designed by backstepping;Next,the RBF neural network is used to approximate the unknown information and virtual control derivative in the system mod-el,an adaptive control law based on RBF neural network is designed and the stability of the system is proved by Lyapunov theory.Finally,the results show that the trajectory tracking performance of this method is better,and the tracking accuracy is improved by more than 80%,which proves the effectiveness of the control method by numerical simulation and comparison with the tradi-tional PD control.