Research On Ship Wave Load Forecast Method Based On Long Short Term Memory Network Algorithm

During the navigation of the ship,the harsh marine environment will pose a greater threat to the safety of the hull structure.In order to ensure the safety of the hull structure in the complex marine environment and grasp the real-time status of the hull structure,the current commonly used method is to use the structural health monitoring system to monitor the hull motion,load and structural response in real time.Considering that the mechanical sensor may fail under severe sea conditions,which will cause the interruption of monitoring information acquisition,and fail to provide real-time load data for the ship control center,which affects the ship’s navigation decision.To solve this problem,this paper proposes a wave load calculation and prediction method based on long and short-term memory neural network(LSTM)algorithm to build the relationship between motion and load,and achieve the purpose of direct calculation and prediction of load based on ship motion.The main research contents of this paper are as follows:1)Based on the LSTM algorithm,an artificial intelligence calculation model for hull motion and wave load is established to realize the calculation function of ship motion to motion prediction and motion to load.The influence of the hyperparameters involved in the model establishment process on the model calculation results is analyzed.2)Based on the numerical calculation results of hull motion and wave load response in regular waves and model test data,the established motion prediction model and load calculation model are verified,and the model calculation results are in good agreement with the test data;on this basis In the above,the two models are used jointly to further verify the applicability of the forecast method to regular wave conditions.3)Based on the numerical calculation results of the ship’s nonlinear motion and wave load response in irregular waves,the motion prediction model,load calculation model and their combined model based on the LSTM algorithm proposed in this paper are verified.