摘要
海平面不断上升威胁人类的生命安全,高精度的海平面预测对人类预防水文灾害具有重要意义。现有的预测方法因验潮站数据为单一时间序列而难以进行高精度预测。针对此问题,提出一种融合变分模态分解(VMD)和极度梯度提升算法(XGBoost)的变分模态分解-极度梯度提升预测模型,简称VMD-XGBoost模型。与XGBoost模型、卷积神经网络与长短期记忆神经网络混合模型(CNN-LSTM)、变分模态-卷积神经网络与长短期记忆神经网络混合模型(VMD-CNN-LSTM)对比,对荷兰沿岸海平面验潮站时间序列进行预测。验潮站预测结果分析表明:相较于XGBoost模型,VMD-XGBoost模型预测结果的均方根误差平均降低65.43%,平均绝对误差平均降低63.79%,平均绝对百分比误差平均降低63.44%,且相较于VMD-CNN-LSTM模型,VMD-XGBoost模型在验潮站海面高序列预测上具有更高预测精度,可实现高精度验潮站时间序列预测。
The rising sea level threatens the safety of human life.High precision sea level prediction is of great significance for human to prevent hydrological disasters.The existing prediction methods are difficult to carry out high-precision prediction because the tide gauge data is a single time series.To solve this problem,this paper proposes a variational mode decomposition-extreme gradient boosting,which combines variational mode decomposition(VMD)and extreme gradient boosting(XGBoost).Compared with XGBoost model,convolutional neural network and Long Short-Term Memory model(CNN-LSTM),variational mode convolutional neural network and Long Short-Term Memory model(VMD-CNN-LSTM),the time series of tidal stations along the Netherlands coast are predicted.The analysis of the prediction results of the tide gauge station shows that,Compared with XGBoost model,the root mean square error of VMD-XGBoost model prediction results decreases by 65.43%on average,the average absolute error decreases by 63.79%on average,and the average absolute percentage error decreases by 63.44%on average.Compared with the VMD-CNNLSTM model,the VMD-XGBoost model has higher prediction accuracy in the high sequence prediction of the sea surface of the tide gauge station,which can realize the high precision time series prediction of the tide gauge station.
作者
陈红康
鲁铁定
孙喜文
李祯
贺小星
赖小婷
CHEN Hongkang;LU Tieding;SUN Xiwen;LI Zhen;HE Xiaoxing;LAI Xiaoting(School of Surveying and mapping engineering,East China University of Technology,Nanchang 330013,China;School of Civil and Mapping Engineering,Jiangxi University of Science and Technology,Ganzhou 341000,China)
出处
《海洋测绘》
CSCD
北大核心
2023年第5期17-21,共5页
Hydrographic Surveying and Charting
基金
国家自然科学基金(42061077,42104023)
江西理工大学高层次人才科研启动项目(205200100588,205200100564)
江西理工大学大学生创新创业训练项目(202210407032)
江西省主要学科学术和技术带头人培养计划(20225BCJ23014)。