自治式潜水器规避航行神经网络控制器设计  

Design of Neural Network Navigation Controller for AUV's Collision Avoidance

作　　者：胡红波[1] 马爱民[1] 

机构地区：[1]海军大连舰艇学院,辽宁大连116018

出　　处：《火力与指挥控制》2007年第4期65-68,共4页Fire Control & Command Control

摘　　要：在规避机动状态下,为了实现对自治式潜水器良好的航行控制,通过分析潜水器的运动状态,建立了运动状态方程,确立了非线性单输入/单输出系统。同时,将复杂的非线性模型通过微分同胚的方法转化为降阶的线性模型,并在此基础上提出了一种基于神经网络的控制方法:将滑模控制思想、反馈线性化理论及神经网络在线建模方法相结合,提出了基于神经网络的鲁棒自适应控制算法,并将此方法用于自治式潜水器高度保持的仿真中,从仿真的结果来看,控制器具有满意的控制效果,证实了该方法的有效性,能为自治式潜水器航行控制提供一定的参考。In order to have excellent control on AUV during maneuverable navigation the paper erected dynamical equation and established a nonlinear single input/single output system through analyzing AUV＇ s motion state, at the same time transformed the complicated nonlinear model into reduction linear one by diffeomorphism and put forth a control method based on NN： using sliding mode ideology, feedback linearization theory and on-line modeling by NN together put forth robust adaptive control arithmetic based on NN, finally applied in the simulation of keeping same distance with seabed for AUV. The controller has satisfying control effect and the method is effective according as simulation result, it is of reference value for AUV＇s navigation control.

关 键 词：自治式潜水器 航行控制系统 神经网络 滑模控制 反馈线性化 鲁棒性 

分 类 号：TP273.2[自动化与计算机技术—检测技术与自动化装置]