基于神经网络的唐山市震害预测  

Earthquake response spectrum micro-zoning in Tangshan city based on ANN

作　　者：陈静[1] 王绍杰[1] 朱庆杰[1] 

机构地区：[1]河北理工大学建筑工程学院,河北唐山063009

出　　处：《世界地震工程》2004年第4期127-130,共4页World Earthquake Engineering

基　　金：河北省科学技术研究项目(03276901)资助项目

摘　　要：地震反应谱是城市地震小区划的关键,也是房屋抗震设计和减灾的基础。反应谱最大值,即地震影响系数最大值是地震反应谱的最重要参数之一,由于地震影响系数最大值受到场地条件、基岩条件和特殊地质现象等诸多因素的共同影响,相互关系复杂且不能用经验公式简单表示,故地震影响系数最大值的精确计算是困扰科技工作者多年的难题。地震影响系数最大值分布的精细预测,关键在于如何选择基岩指标,以及如何确定基岩条件、场地条件和特殊地质现象等指标的分布与地震影响系数最大值分布之间的关系。根据唐山市具体情况,应用人工神经网络方法建立预测模型,探讨较为准确的地震影响系数最大值分布计算方法,为确定唐山市地震反应谱小区划奠定了基础。Earthquake response spectra are the key of earthquake micro-zoning, which is the basis of seismic designs of building and disaster mitigation, and the maximum value of earthquake influence coefficient is one of the most important factors of earthquake response spectra. Because the influence of site condition, basement rock condition, and special geological condition, on the earthquake influence coefficient is complicated, which can't be expressed by experiential equations. So, accurately calculation of the earthquake influence coefficient is a difficult problem in earthquake micro-zoning. The distributive prediction foundation of the earthquake influence coefficient is how to select the basement rock factors and how to analyze the relation among earthquake influence coefficient and the site condition, the basement rock condition and the special geological condition. According to the condition of Tangshan city, prediction model is constructed by artificial neural network method and the distribution of the maximum value of the earthquake influence coefficient in Tangshan city is calculated, which is important for the construction of earthquake response spectra.

关 键 词：地震影响系数 地震反应谱 小区 抗震设计 特殊地质 房屋 基础 震害预测 场地条件 基岩 

分 类 号：P315[天文地球—地震学] TU352[天文地球—固体地球物理学]