BP神经网络在武汉地铁竖井沉降预测中的应用

利用matlab7.0神经网络工具箱为依托,将竖井前期沉降监测数据作为训练样本建立神经网络,分析各岩土参数对沉降变形的影响并结合工程实际情况确定网络输入,对训练样本之后10 d的沉降进行预测并与实测值进行对比,以武汉地铁某线施工竖井为例进行验证,结果表明:预测值与实测值基本吻合,相对误差较小,对竖井施工具有一定的指导意义。

On application of BP neural network in settlement forecasting at shaft of Wuhan Subway

Based on the adoption of the matlab7.0 neural network tool box,the paper establishes the neural network by taking the settlement inspected data at the prepared period of the shaft as the training sample,analyzes the influence of all the rock parameter on the settlement and deformation and identifies the network input by combining with the factual circumstance,compares the settlement prediction and the measures value after the 10 days of having the training sample,proves some facts by taking the shaft at some line construction in Wuhan Subway as the example,and proves by the result that the predicted value is basically consistent with the factual one and has relatively little error,so it can direct the shaft construction in future.