Research On The Method Of Autonomous Underwater Vehicles Trajectory Tracking Control

With countries all over the world transferring its resources development goal to marine resources,autonomous underwater vehicle(AUV)plays an increasingly important role in exploration of marine resources,ocean investigation,underwater salvage,and military applications,and it has become an important tool in ocean exploration.However,because the AUV system is highly nonlinear,multi degree of freedom,strong coupling,motion model uncertainty,and its work environment is complex and unknown,the control of trajectory tracking is facing a great challenge.This paper uses deterministic learning theory and feedback linearization technology to design the controllers for the trajectory tracking in the horizontal plane,vertical plane and three-dimensional space,and achieves high-precision track tracking control.The specific research contents of this paper are as follows:1)Aiming at the spatial motion characters of AUV,the spatial motion variables and their conversion between different coordinate system are introduced in both the inertial frame and body-fixed frame.By analyzing the various forces and moments when AUV is working underwater,the kinematic model and kinetic model are established,and the accurate AUV six degrees of freedom space motion equation is obtained.2)Aiming at the problem of AUV's horizontal trajectory tracking control,and considering the factors like the uncertainty of the controlled model and ocean current interference,an adaptive neural network control method based on backstepping and deterministic learning theory is proposed.By using this method to learn from the uncertain dynamic of the closed-loop system,and designing the learning controller with learned knowledge,the retrain of neural network is avoided,the control performance of system is improved,and finally the verification is conducted by simulation.3)Aiming at the problem of vertical trajectory tracking control of underactuated AUV,the small gain theorem and input-to-state stability analysis are introduced,the input-to-state stability neural network controller is designed based on backstepping and deterministic learning.The state error subsystem and the weight error subsystem of the closed-loop system are input-to-state stability,and the locally-accurate approximation of the unknown closed-loop system dynamics is realized.The difference of control property between the neural network controller and the learning controller designed by using the learned knowledge are compared,the superiority of the learning controller is proved by simulation.4)Aiming at the three-dimensional trajectory tracking control problem of underactuated AUV,five degrees of freedom mathematical model of AUV is established.The system model is transformed into state space equation.Combining the RBF neural network and the input-output feedback linearization technology,an adaptive neural network controller is designed by introducing the filter tracking error.The controller can obtain better control effect,which is validated by simulation.The research results obtained in this paper have a good theoretical guidance and practical reference significance in improving the performance of trajectory tracking control.