一种基于神经网络的自适应非线性自动舵  

A Nonlinear Adaptive Autopilot Based on Neural Network

作　　者：武慧勇 唐匀龙 

机构地区：[1]中国电子科技集团公司第三十八研究所,安徽合肥230088 [2]孔径阵列与空间探测安徽省重点实验室,安徽合肥230088

出　　处：《舰船电子对抗》2017年第2期92-98,共7页Shipboard Electronic Countermeasure

摘　　要：针对仅航向角可测量的船舶航向控制系统,提出一种基于神经网络的自适应非线性航向控制器。首先采用神经网络在线逼近系统中的未知项,并设计滑模观测器在线估计艏摇角速度;然后基于动态面控制思想设计非线性观测器-控制器。利用Lyapunov方法证明了误差变量是一致最终有界的。以某自航船模为例仿真,结果表明所提方法鲁棒性好,且操舵合理。Aiming at ship course control system in which only the course angle can be measured, this paper proposes a nonlinear adaptive course controller based on neural network. Firstly the neural network is used to online approach the unknown terms in the system,and a sliding mode observer is designed to estimate the yaw angular velocity online; then a nonlinear observer-controller based on the dynamic surface control idea is designed. According to the Lyapunov approach, all the error variables are proved to be uniformly ultimately bounded. Finally, simulation of a self-propelled ship model is performed, and the result demonstrates that the designed controller has good robustness and the steering gear operation is reasonable.

关 键 词：航向控制器 非线性 神经网络 观测器-控制器 动态面控制 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]