大连理工大学学报2025，Vol.65Issue(4)：369-375,7.DOI:10.7511/dllgxb202504006
基于LSTM神经网络的船舶油耗模型研究
Research on ship fuel consumption model based on LSTM neural network
摘要
Abstract
In response to the demand for energy saving and emission reduction of ships and improvement of economic benefits,an accurate ship fuel consumption model is established,which provides a decision-making basis for ships to take optimization measures for various sailing strategies.Based on the measured operational data of a Danish passenger ro-ro ship,after data preprocessing and feature selection,fuel consumption models of the case ship are established using the LSTM neural network and various machine learning algorithms.The prediction values of each model for the test set and the additional time series test set are compared with the real values respectively,and the results show that the prediction errors of the LSTM model for both test sets are lower than 1.30％,and the prediction accuracy does not fluctuate greatly.Whereas,the prediction performance of the other models for the additional time series test set decreases,and the stability and prediction accuracy are not as good as that of the LSTM model.Considering the prediction performance of the fuel consumption model and the practical application scenarios,the fuel consumption model based on LSTM neural network has a greater advantage,which is of great significance for the subsequent prediction of fuel consumption rate of ships and the optimization of sailing strategies.关键词
油耗率预测/黑箱模型方法/数据预处理/LSTM神经网络Key words
fuel consumption rate prediction/black box modeling method/data preprocessing/LSTM neural network分类
交通工程引用本文复制引用
李智东,易文欣,陆丛红,周波..基于LSTM神经网络的船舶油耗模型研究[J].大连理工大学学报,2025,65(4):369-375,7.基金项目
国家自然科学基金资助项目(52071059). （52071059）
