针对机械臂末端载荷变化引起机械臂抖动的情况，提出了将神经网络、滑模控制和迭代学习控制相结合的控制方法。首先，利用神经网络的逼近特性，逼近建模误差和外部干扰，使逼近值作为滑模控制器中新的增益，避免滑模控制器本身的抖振。然后，引入迭代学习控制，发挥滑模控制和迭代学习控制各自的优势，得到精度高、鲁棒性强的控制方法。最后，仿真结果表明，对于械臂末端载荷变化引起的不同范围的外部干扰，所提控制方法只需较少的迭代次数就可以达到较高的轨迹跟踪精度。利用迭代过程中误差最小的控制力矩对机械臂进行控制，可以使机械臂准确到达所需位置。

A control method combining neural network, sliding mode control and iterative learning control is proposed to solve the problem of jitter caused by load change at the end of the manipulator. Firstly, the approximation property of the neural network is used to approximate the modeling error and external disturbance, and the approximate value is used as a new gain in the sliding mode controller to avoid the jitter of the sliding mode controller itself. Then, iterative learning control is introduced to give full play to the advantages of sliding mode control and iterative learning control, and a control method with high precision and strong robustness is obtained. Finally, the simulation results show that the proposed control method can achieve high trajectory tracking accuracy with fewer iterations for different range of external disturbance caused by load change at the end of the manipulator. The manipulator can reach the desired position accurately by inputting the control torque with the least error in the iteration process.