地震地磁观测与研究2025，Vol.46Issue(1)：49-56,8.DOI:10.3969/j.issn.1003-3246.2025.01.007
基于支持向量机和BP神经网络法的地震与爆破事件的自动识别
Automatic identification of earthquake and blasting events based on support vector machines and BP neural network methods
摘要
Abstract
This paper aims to identify,compare,and analyze the natural earthquake and artificial blasting events recorded by three seismic stations of the Uga River,Uligi,and Ceke under the jurisdiction of the Bayannur Earthquake Monitoring Center Station.Firstly,the selected seismic records of earthquake and blasting events are decomposed using the wavelet base functions of sym6,db7,and rbio1.5 by SWT,DWT,and WPT,and the corresponding approximation coefficients and detail coefficients are obtained after decomposition.Then,the energy ratio,energy entropy,and Shannon entropy of each layer are extracted and used as characteristic parameters individually,in pairs,and combination.Finally,the BP neural network and support vector machine(SVM)are used to train the feature parameters,and the recognition model suitable for the central station is determined by training and comparing the recognition rate of the two methods.The analysis results show that the support vector machine is more suitable for the Bayannur station,and its recognition rate reaches 95%.关键词
支持向量机/BP神经网络/小波变换/事件自动识别Key words
support vector machines/BP neural networks/wavelet transform/automatic identification of events引用本文复制引用
贾昊东,王禄军,王耀临,胡玮,冯雪东,石伟,于建明,周煊超..基于支持向量机和BP神经网络法的地震与爆破事件的自动识别[J].地震地磁观测与研究,2025,46(1):49-56,8.基金项目
中国地震局监测、预报、科研三结合课题(项目编号:3JH-202301009) （项目编号:3JH-202301009）
