为实现自主水下机器人(AUV)的高精度变深控制,基于AUV 垂直面的运动学和非线性动力学模型,提出了神经网络自适应迭代反步控制方法,设计了运动学和动力学控制器. 文中首先考虑AUV 非线性模型的攻角和水动力阻尼系数的不确定性,设计神经网络控制器来对纵倾运动中的非线性水动力阻尼项和外界海流干扰作用进行在线估计,并基于Lyapunov 稳定性理论设计神经网络权值的自适应律,保证系统闭环信号的一致最终有界. 最后通过两组仿真实验,比较了所设计的控制器在设定控制器增益参数下的系统响应和在扰动作用下的变深控制性能,结果表明,所设计的控制器具有较小的稳态误差和较高的跟踪精度.

In order to implement precise diving control of the autonomous underwater vehicle (AUV), according tothe kinematic and nonlinear dynamic model of AUV, an adaptive iterative backstepping method based on neuralnetwork is proposed, and a kinematic and dynamic controller is designed. In the investigation, considering theexistence of attack angle and the uncertainties of hydrodynamic damping parameters of the nonlinear model of AUV,a neural network-based controller is designed to on-line estimate the nonlinear hydrodynamic damping terms existingin the pitch motion together with external ocean current disturbances. Then, the adaptive law of the network weightsis presented based on the Lyapunov stability theory to guarantee the uniform ultimate bounding of all signals in theclosed-loop system. Finally, two groups of simulation experiments are carried out to compare the system response ofthe designed controller at a certain control gain and the diving control performance in the presence of disturbances.The results show that the designed controller is of smaller static error and higher tracking precision.