Serret-Frenet框架下的神经滑模船舶路径跟踪控制

Neural Sliding Mode Control for Tracking of Ships under Serret-renet Frame

针对欠驱动船舶在恒定速度航行下的路径跟踪问题,提出了一种在Serret-Frenet框架下,基于输入输出线性化的神经滑模控制算法.该算法利用Serret-Frenet框架下船舶运动方程的推导形式,将其转换为类似于直线航迹控制的问题,采用神经网络对基于趋近律的滑模控制进行优化,解决了趋近律滑模控制对系统模型的依赖性,提高了控制器的鲁棒性,并设计了状态观测器对控制对象状态进行重构,以解决系统状态量测量误差对控制效果的影响.在无干扰和存在干扰及参数摄动的条件下分别进行了仿真,结果表明该控制律具有良好的跟踪性能.

Aiming at the problem of path following of underacuated ships at a constant speed,a neural sliding mode algorithm based on input-output linearization under serret-frenet frame was designed.This algorithm uses the deduction form of ship motion equation under serret-frenet frame to converts the problem above into a case similar to straight line path following and employs neural network to optimize approaching law sliding mode control（ALSMC）.It solves the dependency on system model of ALSMC and then improves the robust of controller.To eliminate the influence on control effect brought by measurement error of system state variables,a state observer was designed to reconstruct states of the controlled plant.Simulations on the condition without disturbance and with disturbance also with parameters perturbation indicate that this control law has a good performance index.