基于神经网络的欠驱动水下机器人三维同步跟踪和镇定控制

Three-dimensional simultaneous tracking and stabilization of underactuated autonomous underwater vehicles based on neural network

研究了欠驱动水下机器人的三维同步跟踪和镇定控制问题,并考虑了模型参数不确定性、未知外界干扰和输入饱和限制的影响.针对不同类型期望轨迹的特性,构造了新的辅助虚拟信号以实现对欠驱动方向的控制.基于反步法和Lyapunov直接法,设计了一种饱和自适应统一动力学控制律,使得AUV的状态误差最终收敛至零点附近的有界区域内,其中未知模型参数和外部干扰通过基于神经网络的更新律进行估计.仿真结果表明该控制方法是有效的且具有较好的控制效果.

This paper addresses the three-dimensional simultaneous tracking and stabilization problem for autonomous underwater vehicles(AUVs),where the effects of modeling uncertainties,unknown external disturbances and input saturation constraints are all taken into consideration.Due to analyzing the characteristics of different types of desired trajectories,two novel auxiliary virtual signals are introduced for the purpose of controlling the underactuated directions.Based on the backstepping technique and Lyapunov’s direct method,a saturated adaptive dynamic controller with a unified form is designed to guarantee the error states of AUVs to ultimately converge to bounded areas centered at zero,and while,the neural networks based updating laws are also proposed to estimate the unknown modeling parameters and external disturbances.Simulation results illustrate the effectiveness and acceptable performance of the proposed controller.