Deep Learning Based Moored Ship Movement Prediction to Determine Berthing Position

Automation of berthing manoeuvers in ships is a critical topic since the berthing manoeuver is a unique and demanding activity that sailors must do. Following a predefined trajectory or path often solves berthing control issues. Between 2015 and 2020, 46 ships were monitored in the Punta Langosteira Outer Port (A Coruña, Spain) using a ship movement dataset. This research uses this data to train an improved neural network that forecasts a moored vessel’s six degrees of freedom using ship characteristics as well as ocean-meteorological data. The pitch, surge, yaw, sway, heave, along with roll movements, make up six degrees. This work determines the RMSE, or root mean square error, for each of these degrees. By combining these models with data on climate along with sea state predictions, ship attributes, as well as berthing location, also can predict ship movements multiple times during development. These findings, when compared to the restrictive sign measures for safe ship unloading operations, provide sufficient confidence in the models’ predictions. This will help us pinpoint the optimal operating location and stop operations at the right times, reducing the financial effect of cargo ships that are powerless to work.