Ship Motion Prediction in Irregular Waves Based on TCN-BLSTM-TPA Model
In order to improve the prediction accuracy of ship motion in irregular waves,a network structure that can extract time series features and focus on local information is established by combining the temporal convolutional neural network and the time pattern attention based bidirectional long short-term memory neural network to predict ship motion in irregular waves.An adaptive particle swarm optimization algorithm is proposed to optimize the hyper-parameters of neural network.The ship's roll motion in beam waves and heave and pitch motions in head waves at sea state 5 are selected as the test cases.The effectiveness of the proposed model is verified.It is found that the improved particle swarm optimization enhances the adaptability of the model to different advance prediction times and effectively improves the prediction accuracy.When the advance time reaches 15 seconds,compared to the attention mechanism bidirectional short-term memory neural network model,the proposed model improves the prediction accuracy for ship's roll,heave and pitch by at least 29％,31％and 44％,respectively.The research results can provide a certain reference for short-term prediction of ship motion in irregular waves.
万方数据
维普
