Research On Automatic Berthing Technology Of Unmanned Ship

Berthing large ships is considered to be one of the most difficult ship maneuvers.Because artificial neural networks have strong nonlinear mapping capabilities and can learn from the operating experience of experienced captains,artificial neural networks are used to learn and train manual berthing data to complete the autonomous berthing of ships.In this paper,the container ship “ Yinhe ” is selected as the simulation object in the navigation simulator,and the designed neural network controller is used to complete the berthing task;the teaching and training ship “ Yukun” of Dalian Maritime University is the research object to establish a shore-based Monitoring center,and completed shore-based remote control test on the small boat “ Zhi Long No.1 ” in the Navigation Dynamic Simulation and Control Laboratory.The main work of the paper is as follows:1.In view of the traditional neural network auto-berthing controller ignores the mapping relationship between engine order and propeller speed,rudder order and actual rudder angle,the output of the controller is changed from propeller speed and rudder angle to engine order and rudder order.In order to solve the problem that large ship heading deviation after berthing is completed,the berthing training data in multiple initial states is changed to a single initial state,and instead of selecting part of the data during the berthing process,extracting all the information during the berthing process as the controller input to improve the training effect of the controller.2.Aiming at the problem that the Neural Network Autonomous Docking Controllers can only complete the docking of a specific port,but cannot extend to other ports without training data,a coordinate system(Berth coordinates with the berth vertex as the origin and the shoreline as the Y Axis)is proposed.Use this relative position to train the controller.In the V.Dragon-5000 navigation simulator,the container ship“ Yinhe ” was selected.After docking training at Dalian Port,the docking controller was successfully extended to Shenzhen She Kou Port and Singapore Changi Port without training cata.The simulation has verified that the application of the newly established coordinate system can greatly expand the application range of the Neural Network Autonomous Docking Controller and reduce the cost required for training data.3.In response to the requirements for the Smart ship shore-based Monitoring center issued by the China Classification Society in 2020,the Smart Ship Shore-based Monitoring Center was established in order to be able to remotely and real-time obtain the movement status of the intelligent ship.The research object of “ Yunkun” equipped with multi-sensors uses 4G communication technology to monitor and record the ship’s movement status and navigation environment information in real time,and at the same time uses virtual reality technology to realistically restore the ship’s navigation status in the ocean,which is intelligent solutions for ship’s big data applications.