Motion Perception of Ship-Ship Unmanned System Based on LSTM Neural Network

Due to the frequent wave impact and the ship-ship interference, the hydrodynamic characteristics in the ship-ship situations are complex. This hydrodynamics always creates difficulties and instability for ship-ship operations based on unmanned systems. To effectively guide the operation of multiple ships, the ship motion perception for unmanned systems is studied in detail. In the research, the hull motions of twin-ship under different regular waves are simulated using computational fluid dynamics technology. Then, a ship motion perception model based on a Long Short-Term Memory neural network is established and optimized. According to the simulated wave data, the real-time perception of ship heave and pitch is achieved finally. The influence of ship speed and wave monitoring locations on the accuracy of ship motion perception are also discussed, and the priority in the wave monitoring locations is identified. This technology application can improve the accuracy of multi-ship operations and avoid some accidents in ship-ship parallel sailing.