RBF神经网络在天山地震带地震预报中的应用  

EARTHQUAKE PREDICTION APPLICATION OF RBF NEURAL NETWORK IN TIANSHAN EARTHQUAKE ZONE

作　　者：刘萍[1] 曲延军[1] 向元[1] LIU Ping;QU Yan-jun;XIANG Yuan(Earthquake Agency of Xinjiang Uygur Autonomous Region,Urumqi 830011,Xinjiang,China)

机构地区：[1]新疆维吾尔自治区地震局

出　　处：《内陆地震》2019年第2期113-120,共8页Inland Earthquake

基　　金：新疆地震科学基金(201803)

摘　　要：运用RBF人工神经网络模型,结合中国震例,通过对1976年以来新疆天山地震带MS≥4.7地震前异常参数研究分析,筛选出15个地震异常指标使其作为RBF神经网络的输入样本,经过31组样本集的训练和5组检验样本的检验,建立了基于RBF神经网络地震震级预测模型,通过对实际震例的检验取得了较为理想的预报效果。Based on RBF artificial neural network model and China earthquake cases,the anomaly parameters above MS4.7 earthquakes of Tianshan earthquakes zone in Xinjiang since 1976 were studied.Fifteen earthquake anomaly indices were selected as input samples of RBF neural network.After training of 31 samples sets and 5 testing sets of test samples,a prediction model of earthquake magnitude based on RBF neural network was established.The better prediction effect was obtained by testing actual earthquake cases.

关 键 词：RBF神经网络 天山地震带 前兆异常 

分 类 号：P315.7[天文地球—地震学]