有效波高(significant wave height,SWH)是海洋的重要参数之一,对其的精确预测对渔业发展、海上交通和海洋生态系统具有重要意义.为了提高有效波高的预测精度,本文提出了一种基于卷积神经网络-时空长短时记忆神经网络-卷积神经网络(convolutional neural network-spatiotemporal long short-term memory-convolutional neural network,CNN-STLSTM-CNN)的有效波高预测模型.该模型由编码器(Encoder)、解释器(Translator)和解码器(Decoder)构成.Encoder通过卷积神经网络提取SWH数据的空间特征,Translator通过时空长短时记忆神经网络(spatiotemporal long short-term memory,STLSTM)提取SWH数据的空间特征在时间上的变化特性,Decoder通过卷积神经网络的转置卷积模块重建预测结果.对东海和南海海域的二维有效波高数据进行建模,实验结果表明CNN-STLSTM-CNN模型的均方根误差(root mean squared error,RMSE)、平均绝对误差(mean absolute error,MAE)、均方根误差均值(mean of root mean squared error,M_RMSE)和平均绝对误差均值(mean of mean absolute error,M_MAE)等指标值均低于已有的方法,验证了CNN-STLSTM-CNN模型的有效性.
Prediction of significant wave height based on CNN-STLSTM-CNN model
Significant wave height(SWH)is one of the important parameters of the ocean,and its accurate prediction is of great significance to fishery development,maritime traffic and the marine ecosystem.In order to improve the prediction accuracy of significant wave height,this paper proposes a significant wave height prediction model based on convolutional neural network-spatiotemporal long short-term memory-convolutional neural network(CNN-STLSTM-CNN).The model consists of an Encoder,a Translator and a Decoder.The Encoder extracts the spatial features of the SWH data through the convolutional neural network,the Translator extracts the temporal variation characteristics of the spatial features of the SWH data through the spatiotemporal long short-term memory neural network(STLSTM),and the Decoder reconstructs the prediction results through the transposed convolution module of convolutional neural network.Modeling the two-dimensional significant wave height data in the East China Sea and South China Sea,the experimental results show that RMSE,MAE,M_RMSE and M_MAE values of the CNN-STLSTM-CNN model are lower than the existing methods,which verifies the effectiveness of the CNN-STLSTM-CNN model on SWH prediction.
NETL
NSTL
万方数据
