Modeling and adaptive controlling of cable‐drogue docking system for autonomous underwater vehicles

Abstract

In this article, a novel cable-drogue docking system is proposed between an autonomous underwater vehicle (AUV) and one mobile underwater platform, to increase the docking safety and reduce the negative influences of turbulences around the platform shell on the AUV. The mathematical model of the cable-drogue system and an radial basis function neural network (RBFNN)-based Q-learning proportional-integral-differential (PID) controller are developed for the system characteristics analysis and the stabilized control. First, the cable-drogue's mathematical model is established as a link-joint-connected system, whose kinematic and dynamic equations are derived subjected to the hydrodynamic forces on system. One numerical solution procedure is also presented to obtain the motion state of the cable-drogue system. Next, considering the current interferences on the cable-drogue system, a cross rudder is added on the drogue to enhance the robustness. An RBFNN-based Q-learning PID controller is proposed to improve the control performances, in which the control parameters are adaptively optimized by an Q-learning neural network. Finally, numerical simulations are made to study the steady-state and dynamic characteristics of the cable-drogue system in ocean environment, and to investigate the performances of both the traditional PID controller and the proposed RBFNN-based Q-learning PID controller in unknown dynamical system. Simulation results show the effectiveness of the kinematic and dynamic equations of the cable-drogue system and the proposed controller in this article.