基于EMD-LSTM神经网络模型的气垫船操纵运动预报研究  

Research on Motion Prediction of Air Cushion Vehicle Based on EMD-LSTM Neural Network

作　　者：吴成东 詹可 朱仁传[1] Wu Chengdong;Zhan Ke;Zhu Renchuan(State Key Laboratory of Ocean Engineering,School of Naval Architecture,Ocean and Civil Engineering,Shanghai Jiao Tong University,Shanghai 200240,China)

机构地区：[1]上海交通大学船舶海洋与建筑工程学院海洋工程国家重点实验室,上海200240

出　　处：《水动力学研究与进展（A辑）》2023年第5期794-801,共8页Chinese Journal of Hydrodynamics

摘　　要：海上作业时,准确高效的操纵运动极短期预报对于气垫船的作业效率与安全具有重要意义,但目前对于气垫船操纵运动预报的研究仍然比较欠缺。长短期记忆(Long Short-Term Memory, LSTM)深度学习网络对于时间序列分析上的预报能力极强,是船舶操纵运动极短期预报的有力手段。基于此,该文采用经验模态分解(EMD)方法对气垫船自航模试验数据进行预处理,建立EMD-LSTM神经网络模型预报气垫船操纵运动。经对比验证可知,EMD-LSTM神经网络模型对于气垫船操纵运动预报效果良好,明显优于LSTM神经网络模型,适用于气垫船操纵运动预报研究。Accurate and efficient very short-term forecasting of manoeuvring movements during hovercraft operations at sea is of great importance for the efficiency and safety of hovercraft operations. However, there is still a lack of research on hovercraft manoeuvring motion forecasting. Long Short-Term Memory(LSTM) deep learning network is a powerful tool for very short-term prediction of ship manoeuvring motion because of its excellent forecasting capability in time series analysis. Based on this, in this paper an empirical modal decomposition(EMD) method is applied to pre-process the data and the EMD-LSTM neural network model is established to forecast the manoeuvring motion of hovercraft. The EMD-LSTM neural network model is significantly more effective than the LSTM neural network model in predicting the manoeuvring motion of hovercraft, and is suitable for the study of hovercraft manoeuvring motion.

关 键 词：气垫船 极短期预报 长短期记忆神经网络 经验模态分解 

分 类 号：U674.943[交通运输工程—船舶及航道工程]