基于RBF神经网络的作业型AUV自适应终端滑模控制方法及实验研究

Adaptive Terminal Sliding Mode Control Method Based on RBF Neural Network for Operational AUV and Its Experimental Research

研究了作业型AUV(自主水下机器人)的轨迹跟踪控制问题.实际作业中,水下机械手展开作业过程将引起AUV动力学性能变化,进而影响AUV轨迹跟踪控制;并且水流环境干扰亦将影响AUV轨迹跟踪控制.针对上述AUV轨迹跟踪控制问题,提出一种基于RBF(径向基函数)神经网络的AUV自适应终端滑模运动控制方法.该方法在李亚普诺夫稳定性理论框架下,采用RBF网络对机械手展开引起的AUV动力学性能变化和水流环境干扰进行在线逼近,并结合自适应终端滑模控制器对神经网络权值和AUV控制参数进行自适应在线调节.通过李亚普诺夫稳定性理论,证明AUV系统轨迹跟踪误差一致稳定有界.针对滑模控制项引起的控制量抖振问题,提出一种变滑模增益的饱和连续函数滑模抖振降低方法,以降低滑模控制量抖振.通过AUV实验样机的艏向和垂向的轨迹跟踪实验,验证了本文AUV系统控制方法和滑模降抖振方法的有效性.

The trajectory tracking control problem of operational AUV(autonomous underwater vehicle) is addressed. In general, the stretching and operation processes of underwater manipulator will lead to changes of AUV dynamic performance,which will affect the AUV trajectory tracking control, and so does the water current. Aiming at the trajectory tracking control problem of AUV, an adaptive terminal sliding mode control method based on RBF(radial basis function) neural network is proposed. Under the framework of Lyapunov stability theory, the RBF neural network is used to approximate the changes of AUV dynamic performance caused by the stretching of the manipulator and the disturbance of the water current online. Then combined with the adaptive terminal sliding mode controller, the weights of neural network and control parameters of AUV are adaptively adjusted online. According to the Lyapunov stability theory, it is proved that the system trajectory tracking error of AUV is uniformly stable and bounded. Aiming at the chattering problem caused by the sliding mode control items, a chattering reduction method for the saturated continuous function with variable sliding mode gain is proposed to reduce the chattering of sliding mode control variables. Experiments on heading and vertical trajectory tracking are conducted to verify the effectiveness of the AUV system control method and the sliding mode chattering reduction method.