BP网络在高填方地基沉降预测中的应用  

Prediction of the Settlement of High Embankments by Back Propagation Neural Network

作　　者：马丽慧[1] 韩文喜[1] 李阳[1] 

机构地区：[1]成都理工大学地质工程,成都610059

出　　处：《土工基础》2013年第1期14-16,共3页Soil Engineering and Foundation

摘　　要：采用多层神经网络(两个隐层)对昆明新机场高填方地基T1-T5线路工后1年内各点的沉降值进行预测研究,沉降均由快速沉降期过渡到稳定沉降期,且沉降量的最大值几乎都发生在填筑体较厚处。T1线路的最大沉降预测值为52.7mm,监测沉降值为55.6mm,表明两个隐层的BP模型对沉降预测具有较高的精度。The post construction settlement of high embankments of T1 to T5 alignments in Kunming new airport project site is predicted using a multilayer （two hide layers） neural network method. The observed settlement values indicate that the maxi- mum post construction settlements are related to areas with highest fills. The settlement pattern indicates that the post con- struction settlements have a rapid settlement rate initially and then gradually change to the steady settlement. It is predicted that the maximum post construction settlement at T1 is 52.7 mm and the observed value in August 2011 was 55.6 mm. It is concluded that the two hide layer Back Propagation （BP） neural network model can have an accurate prediction on the post con- struction settlement.

关 键 词：高填方地基 工后沉降 BP网络 预测 

分 类 号：TU433[建筑科学—岩土工程]