ADAPTIVE CONTROL FOR UNDERWATER VEHICLE-MANIPULATOR SYSTEM BASED ON FUZZY CMAC NEURAL NETWORKS

摘要

Underwater Vehicle-Manipulator System (UVMS) is a multi-body system with float base. It is difficult to control the vehicle for its dynamic uncertainty and the multi-joint manipulator's disturbances. Due to it is not easy to get the manipulator's hydrodynamics and vehicle's propeller model, an adaptive controller based fuzzy CMAC is proposed. The neural network's inputs are motion status of vehicle and manipulator, and its outputs are the control voltage of the vehicle's propellers. The control errors are decreasing by the controller's self-study with the disturbances of the manipulator. The paper presents the controller's design and stability analysis. To improve the robustness of the controller, input compensation is added. Experiment results demonstrate the effectiveness of the proposed controller.