基于神经网络模型的基坑支护结构变形预测

Deformation Prediction of Foundation Pit Supporting Structure Based on Neural Network Model

摘要: 随着基坑工程往深大方向发展，基坑带来的安全隐患也越来越多。支护结构对基坑整体的安全性至关重要，因此对支护结构的变形进行预测具有深远的意义。本文基于BP神经网络模型对湖南某变电站基坑的支护结构墙顶水平位移与墙顶竖向沉降进行了预测，选取了80天的监测数据作为数据样本，模型训练采用可变学习率法对进行优化，训练得到的BP神经网络模型对支护结构的变化趋势实现了较为准确的预测，在此基础上通过额外加入前序50天的监测数据扩大了训练集，实现了更高的预测精度，说明BP神经网络在基坑支护预测方面具有良好的适用性。

Abstract: With the development of urban construction, foundation pits are going deeper and larger. As a result, there are increasing potential dangers caused by foundation pit. The supporting structure is important to the overall safety of the foundation pit, so it is of far-reaching significance to pre-dict the deformation of the supporting structure. Based on the BP neural network model, this study predicts the horizontal displacement and the vertical settlement of the supporting structure of a foundation pit for substation in Hunan. 80 days of monitoring data are selected as the data set and the model is trained using a variable learning rate. The trained BP neural network achieved an accurate prediction. On the basis, the training set is extended by adding antecedent 50 days of monitoring data and higher prediction accuracy is achieved. The results show that the BP neural network has favorable applicability in the field of foundation pit prediction.