基于神经网络观测器的船舶轨迹跟踪递归滑模 动态面输出反馈控制

针对三自由度全驱动船舶速度向量不可测问题,考虑船舶模型参数和外部环境扰动均未知的情况,提出一种基于神经网络观测器的船舶轨迹跟踪递归滑模动态面输出反馈控制方法.该方法设计神经网络自适应观测器估计船舶速度向量,且利用神经网络逼近模型参数不确定项,综合考虑船舶位置和速度误差之间关系构造递归滑模面,再采用动态面控制技术设计轨迹跟踪控制律和参数自适应律,并引入低频增益学习方法消除外界扰动导致的高频振荡控制信号.选取李雅普诺夫函数证明了该控制律能够保证轨迹跟踪闭环系统内所有信号的一致最终有界性.最后,基于一艘供给船进行仿真验证,结果表明,船舶轨迹跟踪响应速度快,所设计控制器对系统模型参数摄动及外界扰动具有较强的鲁棒性.

Recursive sliding mode dynamic surface output feedback control for ship trajectory tracking based on neural network observer

Considering unknown ship model parameters and external environmental disturbances, trajectory tracking problem of 3 DOF full actuated ship with immeasurable velocity vectors is analyzed. A neural network observer based adaptive recursive sliding-mode dynamic surface output feedback control method with low frequency learning is proposed for ship trajectory tracking. The adaptive neural network observer is constructed to estimate ship velocity vectors, and a neural network is introduced to provide estimation of the model uncertainty. Combined with ship position and velocity error, a recursive sliding mode function is constructed. Moreover, the trajectory tracking control law and the parameter adaptive law are designed by dynamic surface control technique, and the low frequency learning method is introduced to eliminate the high frequency oscillation control signal caused by external disturbance. The application of a new Lyapunov function proves that all signals in the closed-loop trajectory tracking system can be guaranteed the uniformly ultimate boundedness by using the proposed control law. Simulation results show fast tracking responding speed and high accuracy, and the proposed controller has strong robustness against model parameters uncertainty and unknown external environmental disturbances.