基于理想轨迹学习的机械手神经自适应控制

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。

Neural adaptive control based on desired trajectory learning for robots

Building upon the robot robust control, a direct adaptive control approach using neural networks (NN's) was developed for robots with unknown dynamics nonlinearities, where NN's were used to improve the system dynamic performance by compensating the robot dynamic nonlinearities. A key property of the proposed approach was that the actual joint angle values in the control law was replaced by the desired joint angles, angle velocities and accelerators, and the bound on the NN reconstruction error is assumed to be unknown. Concrete neuro adaptive controller design method is given, and the system stability and the convergence of the NN learning algorithm are proved. Finally, the control performance of the proposed controller is verified with simulation studies.