非线性系统神经模糊自适应控制的问题与策略

Dynamic neuro-fuzzy adaptive control of nonlinear systems

针对具有未知动力学的机械臂系统 ,提出一系列神经模糊自适应控制方法。提出神经模糊动态逆稳定自适应控制方法 ,该方法使用动态神经模糊系统逼近非线性动态系统 ,设计的动态逆控制器可以通过参数的设定保证闭环系统在初始控制段的动态性能 ,而无需事先要求机械臂状态位于某一紧集的假设。结合延时神经模糊网络 ,引入降维观测器估计输出重定义后机械手的速度矢量 ,从而建立了非线性系统的控制器观测器设计的新方法。采用了动态逆和“Back-stepping(后退 )”的技术 。

A set of neuro fuzzy adaptive control approaches was developed for robotic manipulators with poorly known dynamics. The dynamic neuro fuzzy adaptive control based on dynamic inversion approximates the dynamics of the whole nonlinear system. The dynamic inversion guarantees the dynamic system performance in the initial control stage and facilitates controller design because it does assume that the system state is within a compact set because the compact set can not be specified before the control loop is closed. A controller observer design was developed for a flexible link manipulator that combined a time delay neuro fuzzy network and a reduced order observer. The neuro fuzzy method based on dynamic inversion has been expanded to control the flexible link manipulator with actuator dynamics using the backstepping technique, which is an open control problem.