基于神经网络补偿的滑模控制在AUV运动中的应用*

由于自主水下机器人水动力模型参数的不确定性及其强非线性，提出神经网络动态滑模面控制法。将系统分为确定与不确定部分，通过滑模控制实现对系统确定部分的控制，通过神经网络所具有的自适应调节能力实现对未知干扰与不确定部分进行补偿控制，提高系统的强鲁棒性。通过Lyapunov法验证了控制算法的收敛性；通过MATLAB仿真平台和半物理仿真平台，验证了算法的鲁棒性和抗干扰性。

NN compensation controller based on sliding mode control for AUV movement

This paper designed one algorithm of neural network (NN) dynamic feedback control coupled with sliding mode control for the movement of AUV. Decomposed the model of AUV of one known subsystem and one unknown subsystem. The fore part was controlled by sliding mode control theory which was performed by sliding mode control meanwhile the last one was actuated by ANN whose parameters were online adjustment. Proved theories and developed by Lyapunov theory which denoted that the tracking error could be converged to zero. Completed some simulation experiences by MATLAB platform and accomplished some virtual environment experiences with disturbance waves by the proposed algorithm. All these experiences achieve expected results, such as the desired depth control in the vertical plane and the ideal heading tracking in horizontal plane.