基于神经网络方法对多层砖房震害预测  

Earthquake Damage Prediction of Multistory Masonry Building Base on Neural Network

作　　者：马令勇[1] 刘功良[1] 姜伟[2] 

机构地区：[1]大庆石油学院土木建筑工程学院,大庆163318 [2]黑龙江八一农垦大学工程学院

出　　处：《黑龙江八一农垦大学学报》2010年第5期85-87,共3页journal of heilongjiang bayi agricultural university

基　　金：黑龙江省教育厅科学技术研究项目(11513015)

摘　　要：强烈的地震给人们生命财产带来巨大损失,为了能够在地震之前预测出建筑物震害,提出一多层砖房为例。利用MATLAB神经网络工具箱,建立一种基于贝叶斯正则算法的BP神经网络模型,并以过去发生地震地区的多层砖房调查数据为震害因子的震害预测方法。结果表明:对多层砖房的震害样本的预测达到理想效果。A strong earthquake causes huge losses for people’s life and property. In order to predict the building earthquake damage before earthquake,we take the multistory masonry building as the sample to propose by MATLAB neural network toolbox,we should establish one kind of BP neural network model based on Bayesian method and take the survey data of multistory masonry building as the earthquake damage prediction method of earthquake damage factor. The result indicates that the method can achieve an ideal effect for predicting earthquake damage of the multistory masonry building .

关 键 词：人工神经网络 震害预测 贝叶斯正则化算法 多层砖房 

分 类 号：TU362[建筑科学—结构工程]