基于神经网络的模糊反应谱确定方法  

DETERMINATION OF FUZZY RESPONSE SPECTRUM BASED ON NEURAL NETWORK

作　　者：王夏璐 曹双梅[2] 张世海[3] 

机构地区：[1]南阳市地震台网中心,河南南阳473000 [2]河南工业职业技术学院,河南南阳473000 [3]南阳理工学院,河南南阳473004

出　　处：《南阳理工学院学报》2012年第4期85-89,共5页Journal of Nanyang Institute of Technology

基　　金：国家自然基金项目(61040031);河南省重点科技攻关项目(092102310157)

摘　　要：首先,分析了场地、特征周期及反应谱的模糊性特征,介绍了基于二维隶属函数的模糊场地分类方法,并指出了其不足,给出了一种基于BP神经网络的模糊场地分类方法,克服了二维隶属函数在转折边界处不平缓和有突变的不足,使场地分类结果更符合工程实际;接着,建立了基于BP神经网络的模糊化特征周期的确定方法,给出了该糊化特征周期的空间分布图,实现了确定模糊化特征周期方法的智能化;最后,建立了两种基于BP神经网络的模糊抗震设计反应谱确定方法,指出了两种方法均可智能高效地确定模糊反应谱,上述方法为更客观合理地确定地震作用提供了新的途径和方法。First, this paper analyzes the fuzzily characteristics of the sit, the characteristic period and response spectrum, and also introduces two dimensions membership function-based fuzzy site classification method. A novel Back Propagation （BP） network-based site fuzzy classification method is proposed to compensated for the two dimensions membership function＇ weak of being not gradual or gentle but involving irregular or abrupt breaks at the surfaces （junctures of planes） , and the site classification results more accords with the engineering application. Then, BP-based fuzzy characteristic period method is set up, and its space distribution graphics is also given, which makes fuzzy characteristic period method intelligent. Finally, two kinds of methods to determine the fuzzy seismic design response spectrum based on BP networks are constructed. It is pointed out that both the two methods can intelligently and effectively determine the fuzzy response spectrum, which propose new methods and ways of reasonably and objectively determining the earthquake action.

关 键 词：场地模糊分类 模糊特征周期 模糊反应谱 智能化方法 神经网络 

分 类 号：P315.9[天文地球—地震学]