改进规则下自适应神经网络的UUV水平面路径跟随控制

Improved path-following in horizontal plane for UUV using adaptive neural network controller

为实现无人水下航行器(UUV)在水动力参数变化和外界不确定干扰下的水平面路径跟随控制,提出一种基于Lyapunov方法的自适应神经网络控制方法。针对直线段航路交接处减速拐弯时位置控制精度较差的问题,提出利用圆弧连接的圆弧段路径的航向制导(LOS)来改进单纯直线段连接的航向制导,降低拐点处的跟踪误差。引入RBF神经网络来估计误差和海流干扰,设计自适应学习律来保证神经网络权值的最优估计,保证系统的位置误差和艏向误差收敛到零。仿真试验结果表明:设计的控制器在路径跟随过程中可有效抑制UUV载体的模型不确定性,对外界海流干扰有较好的抑制作用,且控制参数易于调节。

In order to deal with the parameter variations and uncertainties during path following control with unmanned underwater vehicles, an adaptive neural network controller was designed using Lyapunov stability analysis. In order to eliminate the cross track error near the intersection which is a connection of two straight lines, an improved LOS law was proposed using straight lines linked with arcs. For the estimation error and current disturbance, the RBF neural network（NN） was introduced to estimate unknown terms where an adaptive law was chosen to guarantee optimal estimation of the weight of NN, when meanwhile a virtual control input was introduced to ensure that the system error, including position error and heading error, can be converged to 0. Simulation results demonstrate that the proposed controller, whose parameters can be set easily, is effective to eliminate the disturbances caused by vehicle＇s nonlinear and model uncertainty, and can overcome current.