基于Attention-LSTM网络的地震前兆研究  

Groundwater Level Prediction and Anomaly Analysis of Earthquake Precursors Based on Attention-LSTM

作　　者：陈新房[1,2] 赵晗清 杨丽佳 汪世伟 CHEN Xin-fang;ZHAO Han-qing;YANG Li-jia;WANG Shi-wei(Institute of Disaster Prevention,Langfang 065201,China;Hebei Province University Smart Emergency Application Technology Research and Development Center,Langfang 065201,China)

机构地区：[1]防灾科技学院,河北廊坊065201 [2]河北省高校智慧应急应用技术研发中心,河北廊坊065201

出　　处：《电脑与电信》2024年第9期23-27,共5页Computer & Telecommunication

基　　金：廊坊市科学技术研究与发展计划自筹经费项目,项目编号:2024011018;中央高校基本科研业务费青年教师资助计划项目,项目编号:2024011018;河北省教育厅科学研究项目,项目编号:ZC2023108。

摘　　要：以2015年4月13日10时28分在建水县发生的4.7级地震事件为例,分析地下水水位变化在地震前兆方面的潜力。将地震周边通海高大井与建水井的地下水水位数据集划分为地震活跃期和非活跃期,设计了Attention-LSTM模型对划分数据进行训练。实验结果表明,该模型不仅能够捕捉到地震引起的地下水水位异常变化而且在同震不同井上的效果理想,模型具有迁移能力,可以泛用在同震的不同井,为利用地下水水位作为地震前兆指标提供了启示。这种方法对于分析地下水水位变化检测地震前兆的研究方法具有一定的科学价值和现实意义,有望为地震预警和防灾减灾工作提供新的思路。Using the 4.7-magnitude earthquake that occurred at 10:28 on April 13,2015 in Jianshui County as an example to analyze the potential of groundwater level changes as earthquake precursors.The groundwater level data of the Tonghai high well and the Jianshui well around the earthquake are divided into earthquake active and inactive periods,and an Attention-LSTM model is designed to train the divided data.The experimental results show that the model can not only capture the abnormal changes in groundwater level caused by the earthquake,but also has ideal effects on different wells during the same earthquake,indicating that the model has transferability and can be widely used in different wells during the same earthquake,providing insights for using groundwater level as an indicator of earthquake precursors.This approach has certain scientific value and practical significance for the research method of analyzing groundwater level changes to detect earthquake precursors,and is expected to provide new ideas for earthquake early warning and disaster prevention and mitigation work.

关 键 词：Attention-LSTM模型 时间序列 地下水水位 地震前兆异常 

分 类 号：P641.8[天文地球—地质矿产勘探] TP181[天文地球—地质学]