定点形变数据同震响应检测深度学习模型

打开文本图片集

中图分类号：P315.61 文献标识码：A 文章编号：2096-7780（2025）09-0521-08

doi：10.19987/j.dzkxjz.2024-156

Deep learning model for co-seismic response detection of fixed-point deformation data

Peng Zhao1）， Shao Yongqian2，Li Yingnan1），Liu Wenbing1，Zhao Liming1）

1） Tianjin Earthquake Agency， Tianjin 30o201， China

2） Shanghai Earthquake Agency， Shanghai 20oo62， China

AbstractCo-seismic response identification of fixed-point deformation data currently relies on manual selection， and no automatic co-seismic response detection method has yet been applied.This study proposes the first deep learning model for theco-seismic response detectionoffixed-point deformationdata in China，which is used todetectco-seismic response signalsquicklyandaccuratelyon the seconddataset of asingle vertical pendulum broadband tiltmeter.The modelwasconstructed using the transfer leaming technique;it introduces three representative pre-trained models for earthquake detection in seismic data as feature extractors，migrates their knowledge and capabilities of earthquake detection inseismicdata to fixed-point deformation data，and then designs and adapts supporting data converters and classifiers.Tests onreal observational data showed that the modelprovided a good detection performance.The application ofcontinuous data from Jixian station proved that the model was notonlycapable ofdetecting alltheco-seismicresponse events recorded manually，but it also found events that were not recognized manualywith an accuracyrate of no less than 75% . Compared with traditional manual processing，the detection eficiency， detection capability，and consistency are greatly improved using this model.

Keywordsfixed-point deformation;co-seismic response detection; deep learning;transfer learning; vertical pendulum broadband tiltmeter

0 引言

定点形变观测在地震研究中有着广泛的应用，其连续观测形成的定点形变时序数据可应用到地震监测预报及相关地球科学领域的研究中[1-2]。（剩余11380字）