| In recent years,smart ship has become the development trend of shipping industry.Ship motion control is the key of intelligent navigation,which need to consider the nonlinearity,uncertainty and characteristic of power system.The influence of power system and the underactuation of ship motion are two key factors in the coupling of ship motion control.Therefore,it is of great significance to study the influence of underactuation characteristics and uncertainty of power system on ship path following,which can lay a foundation for further semi-autonomous and autonomous navigation of ships.In this paper,for the requirements of intelligent control of large ships,the characteristics of the ship’s power system are comprehensively described.Based on the basic dynamic characteristics of underactuated ships,a path following control method is established to replace human operation.The main work is as follows:1.A response model considering the hysteresis effect of ship’s engine and rudder is established.A three degree of freedom（dof）MMG model was established based on a 7-meters scaled model of KVLCC2.The forces and torques in all directions are analyzed,and the hysteresis effect of ship’s engine and rudder is modeled.The simulation of the moving model is carried out in python environment,and the maneuverability of the model is verified by the full rotation experiment,and the results are compared with the experimental data of the real ship.The results show that the simulation model is in good agreement with the experimental data of the real ship,and the accuracy of the mathematical model is verified.2.A ship course controller based on DDPG algorithm is designed.Based on the analysis of the characteristics and application scenarios of the value-based and the policy-based reinforcement learning algorithm,this paper discusses the advantages and disadvantages of the most representative algorithms such as Q-learning,DQN and DDPG,and selects DDPG algorithm for the course control.DDPG algorithm is used to design the ship’s course controller,and the simulation environment of course control task is established.The controller can adapt to the dynamic environment and optimize the policy in real time by interacting with the environment.The results show that the ship course controller based on DDPG algorithm has better adaptability and stability than the traditional fuzzy PID control.3.A path following control method considering the response characteristics of the power system is established.Aiming at the characteristics of large inertia,strong delay and uncertainty of underactuated ship,the state space of ship path following intelligent control is constructed.According to the path following control requirements and the characteristics of DDPG algorithm,the neural network structure is designed,and the input and output of different networks are customized.A reward function based on ship manoeuvring characteristics is proposed,and the ship manoeuvring characteristics are normalized as reward or punishment,so as to guide the path following the behavior of control agents.Aiming at the shortcomings of the original DDPG algorithm,which has too much overparameters,random initialization and is sensitive to the environment,an improved DDPG algorithm with delay update function is proposed.Experimental results show that the improved DDPG algorithm has better stability and reduces the possibility of non-convergence during training.The experimental results of DDPGbased method and the traditional fuzzy PID control method shows that the DDPGbased path following control method has better performance.It can respond to the path deviation quickly and is more suitable for the interference of the environment.The control process of DDPG-based controller is more stable,which conforms to the ship manoeuvring characteristics.This paper explores an intelligent control method of ship path following considering the influence of ship dynamic system.The research results are of great value in the development of autonomous navigation. |
PDF Full Text Request