摘要
极移是研究地球自转参数的重要内容之一。为此,基于国际地球自转服务组织提供的极移序列,构建了两个模型,分别是经典的最小二乘与自回归(LS+AR)模型和BiLSTM与注意力机制相结合的神经网络预测模型,旨在实现极移的短期预测。实验采用12年的基础序列,总共进行了40期的短期预测。结果显示:神经网络模型预测PMX、PMY分别小于4.2 mas、2.3 mas。尽管在1~10 d的超短期预测中,所提出的神经网络预测模型预测效果与LS+AR模型相比,预测精度提升并不显著;但在后20天的预测中,神经网络模型表现出更好的预测精度,在极移预测上有可行性。
Polar motion is one of the important contents in the study of Earth rotation parameters.In this study,a classical LS+AR model was built according to the pole motion sequence provided by the International Earth Rotation Service,and then a neural network based on the combination of biLSTM and attention mechanism was established to predict the pole motion in the short term.Based on the polar motion time series data provided by International Earth Rotation and Reference Systems Service,this study constructed two predictive models:the classical least squares combined with autoregressive(LS+AR)model,and the neural network prediction model integrating BiLSTM with the attention mechanism,aiming to achieve short term polar motion prediction.The study utilized 12 years of basic sequence and performed a total of 40 short term predictions.The results showed that the neural network model predicted PMX and PMY to be less than 4.2 mas and 2.3 mas,respectively.While the proposed neural network prediction model did not significantly improve prediction accuracy compared to the LS+AR model in ultra short term predictions(1~10 days),it demonstrated enhanced accuracy in predicting the last 20 days,indicating its feasibility in polar motion prediction.
作者
阙海波
王乐洋
吴飞
严凯玲
QUE Hai-bo;WANG Le-yang;WU Fei;YAN Kai-ling(Key Laboratory of Mine Environmental Monitoring and Improving around Poyang Lake of Ministry of Natural Resources,East China University of Technology,Nanchang Jiangxi 330013,China;School of Surveying and Geoinformation Engineering,East China University of Technology,Nanchang Jiangxi 330013,China;Jiangxi Province Engineering Research Center of Surveying,Mapping and Geographic Information,Nanchang Jiangxi 330025,China)
出处
《地矿测绘》
2025年第2期1-5,共5页
Surveying and Mapping of Geology and Mineral Resources
