基于DSC的欠驱动船舶路径跟踪神经滑模控制

为进一步提高船舶路径跟踪的准确性和稳定性,针对欠驱动船舶易受外界环境的干扰和模型不确定等问题,提出一种基于动态面技术的神经网络滑模控制策略。在虚拟船逻辑制导算法的基础上,结合反步法对虚拟控制律进行设计,引入动态面技术对虚拟控制律的导数进行估计,避免因直接求导而增加计算负担。在动力学回路设计中,将径向基神经网络与滑模控制相结合,设计神经网络权重的自适应律,实现对系统非线性项的在线估计和对实际控制律的设计。采用Lyapunov直接法证明闭环系统的稳定性,并对一艘长为32m的单体船进行仿真试验。结果表明,该控制策略可行,能实现对欠驱动船舶路径的有效跟踪。

The neural sliding mode path following control for underctuated ship based on DSC

For the external environmental disturbance and model uncertainty of underctuated ships, to further realize the accuracy and stability of ship path tracking. A dynamic surface based neural sliding mode control strategy is developed. In the guidance of a virtual ship, the virtual control law is devised by backstepping method. The dynamic surface is used to estimate its derivative, so as to reduce the computational burden caused by direct differentiating. In the design of the dynamic loop, the radial basis function neural network is combined with sliding mode. Adaptive law of weights of radial basis function neural networks is designed, in addition the online to approximate the nonlinear terms and the devise of the actual control law are realized. The direct Lyapunov technique is used to analyze the stability of the closed-loop system. Finally, the simulation on 32m mono hull ship is conducted to testify the path following effectiveness of the control scheme.