机械臂协调操作柔性负载鲁棒神经网络控制

Robust neural network control of dual-manipulator cooperative system handling flexible payload

针对多变量刚柔耦合的双机械臂协调操作柔性负载系统,基于奇异摄动理论研究该系统有限元模型的分解以及轨迹跟踪控制问题.考虑动力学模型的复杂性,通过双时标变换将协调系统分解成表征系统大范围刚性运动的慢变子系统和表征系统弹性振动的快变子系统.基于反演思想在慢变子系统中设计鲁棒神经网络控制策略,实现系统轨迹跟踪性能;针对快变子系统,设计鲁棒最优控制策略抑制系统的弹性振动.仿真研究结果表明,该控制策略增强了系统的跟踪性能和鲁棒性.

Aimed at the multivariable rigid-flexible coupling system of dual-manipulator cooperative system handling a flexible payload,the decomposition and the trajectory tracking control of finite element modeling for the system were conducted based on the singular perturbation theory.Considering the complexity of dynamic model,the model was divided into two subsystems which included a slow subsystem describing the large rigid motion and a fast subsystem expressing the elastic vibration by the dual-time scale transform.A robust neural network controller was designed based on the idea of backstepping in the slow subsystem in order to complete the system trajectory tracking performance.For the fast subsystem,a robust optimal controller was presented to suppress the elastic vibration of the system.Simulation results show that the controller can strengthen the tracing ability and the robustness of the system.