基于神经网络的鲁棒自适应舵减摇控制

针对欠驱动船舶在未知海浪扰动下的舵减摇控制问题，设计一种基于有限时间收敛的滤波反步滑模自适应控制器。

首先，针对常规反步法中微分爆炸问题，引入一阶滤波器，以避免对虚拟控制律的求导；然后，根据有限时间控制理论，通过滤波反步法与滑模自适应控制方法，设计有限时间收敛的舵减摇控制律；最后，通过李雅普诺夫稳定性理论，证明航向子系统和减摇子系统的稳定性。

仿真结果表明，在不同海浪扰动下，所设计控制器使船舶能够在有限时间内追踪航向并实现减摇效果；与反步滑模控制相比，所设计控制器可改善船舶航向保持控制性能，减摇率提升了5%。

所提方法可为欠驱动船舶舵减摇控制问题提供参考。

Robust adaptive rudder roll stabilization control based on neural network

An adaptive finite-time filtered backstepping sliding mode controller is designed to solve the rudder roll stabilization control problem of underactuated ships in unknown wave disturbance conditions. First, in order to avoid differential operation for the virtual control law, a first-order filter is introduced to address the problem of differential explosion that affects the backstepping approach. By combining the filtered backstepping technique with adaptive sliding mode control, finite-time convergent control laws for rudder roll stabilization are designed on the basis of the finite-time control theory. Finally, the stability of the heading control subsystem and roll damping subsystem is proven by the Lyapunov stability theory. The simulation results indicate that the designed controller can realize course tracking and roll damping simultaneously in finite time under different wave disturbance conditions. Compared with backstepping sliding mode control, the designed controller can improve the course-keeping control performance, and the roll reduction rate is increased by 5%. The proposed method can provide useful references for the rudder roll stabilization control problem of underactuated ships.