基于小波神经网络的建筑工程沉降变形预测

发布时间：2019-02-25 17:35

【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.
[Abstract]:Deformation monitoring is an important part of safety, information engineering construction and management, which runs through the whole process of building design, construction and operation. Based on the theory of wavelet analysis, BP neural network, wavelet analysis and neural network, with the help of MATLAB programming, the improved BP neural network and auxiliary wavelet neural network are established. Three kinds of deformation prediction network models based on embedded wavelet neural network. Combined with the measured data, the three models are used to predict the cumulative settlement and the different model data of the period settlement respectively. The results show that the prediction effect of the combined wavelet neural network model is obviously better than that of the single BP neural network model, and it has higher prediction precision and faster convergence speed. Moreover, the more training samples, the higher the precision of the model. The better the prediction.
【作者单位】：北京交通大学土木建筑工程学院;北京交通大学轨道工程北京市重点实验室;
【基金】：中央高校基本科研业务费专项资金资助(2015JBM070)
【分类号】：TU433;TP183

【参考文献】