Bayes-LSTM method for predicting heave movement of crude oil vessels during maritime navigation

Abstract

To enhance the prediction accuracy of vessels heave movement during maritime navigation and improve the safety of vessels during maritime navigation and operation, a 10×104 t class crude oil vessel was taken as the research object, and the numerical simulation software STAR CCM+ was employed to build the model simulating the motion of the vessel. Six sets of heave motion data were obtained from six working conditions formed by the combination of two conditions(no-liquid cargo hold and half-loaded liquid cargo hold) and three wavelengths of 0.5 λ, 1.0 λ and 1.5 λ(λ = 6.16 m), and they are divided into a training set and a test set at the ratio of 8:2. The prediction of the model heave movement was conducted through the utilization of the Bayesian Long Short-term Memory Neural Network(Bayes-LSTM) by employing a Bayesian algorithm. The predicted results were then compared with those predicted by the Long Short-term Memory Neural Network(LSTM) model. The results indicate that the prediction accuracy of Bayes-LSTM model at its best is more than 3 times higher than that of LSTM model, which demonstrates the advantage of Bayes-LSTM model in predicting the heave movement of vessels during maritime navigation and operation.

Keywords

• crude oil
• vessels
• heave movement
• star ccm+
• bayesian algorithm
• lstm
• bayes-lstm