基于BP神经网络的单层钢筋混凝土柱工业厂房震害预测
详细信息 本馆镜像全文    |  推荐本文 | | 获取馆网全文
摘要
将人工神经网络理论应用于等高单层钢筋混凝土柱工业厂房的震害预测.在分析震害特点的基础上,将震害影响因子分为精确性和规律性两大类,提出以地震反应指标、天窗类型、支撑情况、建筑材料作为主要的影响因子,并给出了相应的量化取值范围,然后将震害等级作为输出结果,构造了震害预测的BP人工神经网络.通过对52个实际震害实例的检验,网络的准确率超过80%.计算结果证明了该人工神经网络的有效性.
A back propagation artificial neural network is applied to predict seismic damage of single-story reinforced concrete industrial building.Based on the analysis of characteristics of seismic damage,it is found that earthquake response index,type of skylight,bracing system and building material are the main factors affecting seismic damage.The four factors can be classified into two types: precise factors and regular factors.The corresponding spans of factors are suggested and applied to engineering examples.Thus the back propagation artificial neural network is developed,with factors affecting seismic damage as input and seismic damage grade as output.Verified by 52 engineering examples,the percentage of accuracy is above 80%.It is concluded that the back propagation artificial neural network developed in this paper is applicable to predict seismic damage of single-story reinforced concrete industrial building.
引文
[1]王晓明,朱伯龙.基于二次模糊修正的钢筋砼单层工业厂房震害预测综合评判模式[J].建筑结构学报,1994,15(1):32-38.
    [2]李桂青,李正农.建筑物和构筑物震害的灰色预测方法[J].湖北工学院学报,1992,7(3/4):7-12.
    [3]徐祥文,黄崇福.结构动力反应与震害关系的模糊识别[J].地震工程与工程震动,1989,9(2):57-65.
    [4]王光远,徐祥文,周锡元,等.单层厂房自振特性及其在地震反应计算中的应用[C]//地震工程研究报告集(第一集).北京:科学出版社,1962:51-75.
    [5]汤吉庭.钢筋混凝土单层厂房动力特性的实测与分析[C]//冶金部建筑研究总院,西安冶金建筑学院.地基与工业抗震.北京:地震出版社,1984:119-137.
    [6]尹之潜,李树桢,杨淑文,等.震害与地震损失的估计方法[J].地震工程与工程震动,1990,10(1):99-107.
    [7]吴育才,黄宗明,王金海.单层厂房震例及其应用[M].济南:山东科学技术出版社,1991:152-225.

版权所有:© 2023 中国地质图书馆 中国地质调查局地学文献中心