基于高阶递归神经网络的AUV鲁棒控制方法

目的 提出一种基于高阶递归神经网络的AUV鲁棒控制方法。方法 利用结构简单但逼近效果优越的高阶递归神经网络，对建模不确定性和外部未知干扰进行估计，并将其补偿到输入控制律中，以提高控制性能。之后，基于HJI理论和Lyapunov稳定性分析导出神经网络权重自适应更新律和AUV自适应控制律，设计反步滑模方法作为对比方法，并进行仿真实验。结果 设计的基于高阶递归神经网络的AUV鲁棒控制方法的跟踪误差、调节时间等控制指标均优于反步滑模方法。设计的鲁棒控制方法可以控制AUV精确跟踪目标轨迹，同时具有优秀的控制性能和鲁棒性。结论 这一研究为AUV轨迹跟踪控制领域提供了一种高效且有效的方法，有望在复杂、不确定的水下环境中得到应用。

A Robust Control Method for AUV Based on High Order Recurrent Neural Networks

The modeling uncertainties and external unknown disturbances, among other factors, impose higher demands on the control methods for Autonomous Underwater Vehicle (AUV) in terms of trajectory tracking. The work aims to propose an AUV robust control method based on high-order recurrent neural networks to address it. High-order recurrent neural networks with simple structure but superior approximation performance were employed to estimate modeling uncertainties and external unknown disturbances, which were then compensated for in the input control law to enhance control performance. Subsequently, the neural network weight adaptive update law and AUV adaptive control law were derived based on the HJI theory and Lyapunov stability analysis. Finally, a backstepping sliding mode method was designed as a comparative approach, and simulation experiments were conducted. The experimental results indicated that the proposed AUV robust control method based on high-order recurrent neural networks outperformed the backstepping sliding mode method in terms of tracking error, settling time, and other control metrics. Simulation experiments demonstrate that the proposed robust control method can effectively facilitate precise target trajectory tracking by AUVs, while simultaneously exhibiting excellent control performance and robustness. This research provides an efficient and effective approach for AUV trajectory tracking control, with the potential for application in complex and uncertain underwater environments.