摘要
变形监测是安全化、信息化工程建设和管理的重要内容,贯穿于建筑物设计、施工和运营整个过程.本文基于小波分析、BP神经网络、小波分析与神经网络结合的相关理论,借助MATLAB编程,建立了改进的BP神经网络、辅助式小波神经网络、嵌入式小波神经网络3种变形预测网络模型.结合工程实测数据,利用建立的3种模型,分别应用累积沉降和期间沉降不同模式数据进行预测.结果表明,两种小波神经网络组合模型的预测效果明显优于单一的BP神经网络模型,具有更高预测精度和更快的收敛速度,且训练样本数目越多,模型精度越高,预测效果越好.
Deformation monitoring is a key work for engineering safety management,which is performed throughout the entire process of building design,construction and operation. Based on the theoretic analysis of wavelet and neural network,we applied the improved BP neural network model,auxiliary wavelet neural network model,and embedded wavelet neural network model to the settlement prediction of building engineering with MATLAB software programming. The overall performances of three models were analyzed and compared. The results show that the prediction precisions of two wavelet neural network models are obviously superior to the single BP neural network model.
出处
《应用基础与工程科学学报》
EI
CSCD
北大核心
2015年第3期629-636,共8页
Journal of Basic Science and Engineering
基金
中央高校基本科研业务费专项资金资助(2015JBM070)
关键词
建筑物沉降
变形预测
BP神经网络
小波分析
小波神经网络
building settlement
deformation prediction
BP neural network
wavelet analysis
Wavelet neural network
