基于深度学习的有义波高预报  

Forecast of Significant Wave Height Based on Deep Learning

作　　者：孔圆 王先洲[1,2] KONG Yuan;WANG Xianzhou(School of Naval Architecture&Ocean Engineering,Huazhong University of Science and Technology,Wuhan 430074,China;Hubei Key Laboratory of Hydrodynamics of Ships and Marine Engineering,Wuhan 430074,China)

机构地区：[1]华中科技大学船舶与海洋工程学院,武汉430074 [2]船舶和海洋水动力湖北省重点实验室,武汉430074

出　　处：《中国造船》2024年第4期263-272,共10页Shipbuilding of China

摘　　要：有义波高是海浪气候预报的一个关键参数,传统的有义波高预报方法存在计算成本高的问题。论文构建了一个两阶段深度学习模型,利用卷积神经网络和长短期记忆网络,分阶段学习风和浪的时空关系,实现了有义波高的高精度预报。Wave climate forecasting plays a crucial role in marine-related engineering,with significant wave height being one of the key parameters.In this study,a two-stage deep learning model is constructed,utilizing convolutional neural networks and long short-term memory networks to learn the spatio-temporal relationships between wind and waves in a phased manner,thus achieving a high-precision forecast of significant wave height.

关 键 词：有义波高 海浪预报 深度学习 卷积神经网络 长短期记忆网络 

分 类 号：P731.22[天文地球—海洋科学]