基于BP神经网络邻近地铁基坑土体参数反演分析

Inversion Analysis of Soil Parameters of Adjacent Subway Foundation Excavation Based on BP Neural Network

为能够更好预测邻近地铁基坑变形,基于天津某邻近地铁深大基坑,运用Plaxis3D数值分析软件和Matlab分析软件,采用BP神经网络进行位移反演分析。结果表明:BP神经网络结构4-13-5具有较好的反演分析能力,将求取的反演参数主固结割线模量导入数值模型中,反演优化后数值结果与监测值吻合度较高,误差仅为3.6%。

In order to better predict the deformation of the foundation excavation near the subway,Plaxis3 d numerical analysis software and Matlab analysis software are used,and BP neural network was used for displacement inversion analysis.The results show that BP neural network structure(4-13-5)has a good ability of inversion analysis.The secant modulus of main consolidation calculated by BP neural network is imported into the numerical model.After the optimization of inversion,the numerical results are in good agreement with the monitoring values,and the error is only 3.6%.