中文摘要：

针对自主水下航行器平面轨迹跟踪控制的学习能力问题，采用RBF神经网络设计了一种结合反步法和确定学习理论的自适应神经网络控制器，在满足持续激励的条件下，对系统未知动态进行神经网络辨识，实现了对未知动态的局部准确逼近和部分神经网络权值的收敛，保证了系统所有信号最终一致有界和稳定。将从动态模式中学到的知识静态保存，提取存储知识设计学习控制器，实现了对参考轨迹更加快速精确地跟踪，为执行同样的任务节省了时间和能量。通过仿真验证了所设计学习控制器的有效性和优越性。

英文摘要：

An adaptive neural network controller based on backstepping and deterministic learning is proposed by using RBF neural network for the ability of learning during the trajectory tracking of autonomous underwater vehicles. All signals in the system are uniformly ultimately bounded, and the convergence of partial neural weights and locally - accurate approximation of unknown system dynamics are achieved by using neural network identification under the persistent excitation condition. The learned knowledge acquired from the dynamical pattern can store as constant neural weights,which can be used to design learning controller. The new controller achieves more quick and accurate tracking of reference trajectory, and saves time and energy for the same task. The simulation resuhs show the effectiveness and superiority of the learning controller.