林彬华^1,2,,
金星^1,2,3,,,
康兰池^1,3,
韦永祥^1,3,
李军^1,3,
张燕明¹,
陈惠芳¹,
周施文¹
1. 福建省地震局, 福州 350003
2. 福州大学, 福州 350108
3. 中国地震局厦门海洋地震研究所, 厦门 361021

基金项目: 国家重点研发项目（2018YFC1504003，2018YFC1504005），福建省地震局攻关项目（G201902），福建省地震局科技基金项目（SF202005）资助


详细信息

作者简介: 林彬华, 男, 1988年生, 博士, 工程师, 主要从事地震监测预警方面研究.E-mail: lbhfzu@sina.com

通讯作者: 金星, 男, 1960年生, 博士, 研究员, 主要从事地震监测预警、主动源探测等方面研究.E-mail: jinxing_fj@163.com

中图分类号: P315

收稿日期:2020-09-28
修回日期:2021-07-13
上线日期:2021-10-10

The research of earthquake magnitude determination based on Convolutional Neural Networks

LIN BinHua^1,2,,
JIN Xing^1,2,3,,,
KANG LanChi^1,3,
WEI YongXiang^1,3,
LI Jun^1,3,
ZHANG YanMing¹,
CHEN HuiFang¹,
ZHOU ShiWen¹
1. Earthquake Administration of Fujian Province, Fuzhou 350003, China
2. Fuzhou University, Fuzhou 350108, China
3. Xiamen Institute of Marine Seismology, China Earthquake Administration, Xiamen 361021, China


More Information

Corresponding author: JIN Xing,E-mail:jinxing_fj@163.com

MSC: P315

--> Received Date: 28 September 2020
Revised Date: 13 July 2021
Available Online: 10 October 2021


摘要
摘要:地震预警震级测定是地震预警系统最重要也是最困难的部分之一.本文提出了基于卷积神经网络的地震预警震级测定方法，将震级测定问题转化为震级分类问题，即将M_L>2.0的震级分成20个不同等级类别处理.收集了福建台网2012—2019年期间记录到福建、台湾海峡及台湾共1928个地震作为研究资料，经过台站记录截取、大震样本增强、标签制作、质量筛选等预处理共得到14644条三分向地震样本记录；构建了3 s波形输入的卷积神经网络震级预测模型，并用2012—2018年震例对模型进行训练，用2019年震例对模型进行测试.结果表明，单台震级偏差有85.6%可控制在±0.3以内，前三台平均的震级偏差有91.8%可控制在±0.3以内，其中震级较大偏差的事件多为缺乏历史样本.相较于传统方法，该模型测定的震级值更加稳定可靠，可为解决地震预警震级测定这一挑战性难题提供新的技术手段.
关键词: 卷积神经网络/
地震预警/
震级测定/
震级偏差/
深度学习

Abstract:The magnitude determination of earthquake is one of the most important and challenging parts in Early Earthquake Warning (EEW) system. In this paper, the method of determination of earthquake magnitude based on convolutional neural network (CNN) is proposed. This method transforms magnitude determination problem into a classification problem, by dividing earthquake magnitudes into 20 different categories which are greater than 2.0 (M_L>2.0). In this paper, a total of 1928 earthquakes in Fujian, Taiwan Strait and Taiwan area recorded by Fujian Seismic Network from 2012 to 2019 were collected as research data; 14644 three component seismic records were obtained by station record interception, data tagging and quality screening along with other pre-processing procedures. A convolutional neural network (CNN) model for magnitude prediction was constructed by inserting three-second data records. The model was trained with the earthquake events from 2012—2018 and tested with the earthquake events in 2019. The results showed that 85.6% of the magnitude deviation of a single station can be controlled within 0.3, and 91.8% of the average magnitude deviation of the first three stations can be controlled within 0.3. Those cases with relatively large deviation are mainly due to the lack of historical samples. Compared with the traditional methods, the magnitude determined by the CNN model is more stable and reliable, which can provide a new technical method for solving the challenging problem of EEW magnitude determination.
Key words:Convolutional Neural Networks (CNN)/
Earthquake Early Warning (EEW)/
Magnitude determination/
Magnitude deviation/
Deep learning



PDF全文下载地址:

http://www.geophy.cn/data/article/export-pdf?id=df0eb0ac-7483-4329-8ff1-db39baec8cbb