欠驱动船舶神经网络自适应路径跟踪控制

针对模型参数未知的欠驱动船舶路径跟踪问题，将神经网络技术与反演设计法相结合，提出一种神经网络稳定自适应控制方法。首先根据运动学误差方程和线性变换确定辅助的前进速度和艏摇角，然后利用神经网络逼近技术对模型中任意不确定因素进行补偿，设计自适应控制律,使得实际的前进速度和艏摇角分别收敛到辅助值。应用Lyapunov函数证明了船舶路径跟踪闭环系统的误差信号最终一致有界。仿真结果表明，利用设计的控制律可以迫使欠驱动船舶跟踪曲线和直线路径，并且具有较强的鲁棒性。

Neural network adaptive path tracking control for underactuated ships

Considering path-following problems of underactuated ships with parameter uncertainties, the nerve network technology was combined with the backstepping method for proposing a stable nerve-network adaptive control method. Firstly, based on kinematics error equations and linear transformation, auxiliary surge velocity and heading angle were determined; then the nerve network approximation technology was utilized to compensate for any uncertainties in the model, an adaptive control law was designed, so as to make actual surge velocity and heading angle converge to the auxiliary values respectively. By using the Lyapunov function, it was proved that the ultimately uniform boundedness of the error signals in the closed-loop path following system of ship. Numerical simulation results show that, the designed law can force underactuated ship to follow curve and straight path, it has strong robustness.