论文部分内容阅读
在机械手鲁棒控制的基础上，讨论了神经网络逼近误差界未知情形下机械手的神经网络直接自适应控制方法，这里神经网络用于补偿系统的不确定性，提高整个系统的跟随性能。提出设计方法的主要特点是神经网络控制器设计采用机械手待跟随的理想关节信号代替实际的机械手关节角、关节速度和关节角加速度作为神经网络的输入，此外神经网络的逼近误差界假设是未知的。给出了具体的系统设计算法，并证明了神经网络学习算法的收敛性和整个系统的全局稳定性。最后，一两连杆机械手的控制器设计仿真实例验证了提出算法的有效性。
Based on robust control of manipulator, neural network approach to direct adaptive control of manipulator neural network is proposed, which is used to compensate the uncertainty of the system and improve the follow-up performance of the whole system. The main features of the proposed design method are that the neural network controller design uses the ideal joint signal to be followed by robot to replace the actual manipulator joint angle, joint velocity and joint angular acceleration as the input of neural network. Moreover, the assumption of approximation error of neural network is unknown . The concrete system design algorithm is given, and the convergence of the neural network learning algorithm and the global stability of the whole system are proved. Finally, the simulation of one or two-link manipulator controller verifies the effectiveness of the proposed algorithm.