摘要
针对高填路堤地基沉降预测中影响因素众多且存在高度的非线性,难以用解析式表达等特点,提出采用基于L-M(Levenberg-Marquardt)的BP神经网络法对高填方地基沉降进行预测,并通过对工程实例的网络训练和网络检验,得出BP神经网络计算值与实测值十分接近的结论,可充分证明L-M法BP神经网络在高路堤地基沉降预测中具有很好的实用价值。
In the prediction of foundation settlement for high-filled embankment, there are many affecting factors and most of them are high nonlinear, therefore, it is difficult to express with analytical formulas. Aiming at these characteristics, BP neural network based on L-M method can be put forward to predict foundation settlement. It trains and tests the network with engineering examples, and the results show that network calculation value is close to actual survey value, which indicates that L-M based on BP neural network is practical in prediction of foundation settlement for high-filled embankment.
出处
《交通标准化》
2006年第10期167-171,共5页
Communications Standardization
基金
广西交通科技项目
关键词
高填路堤
地基沉降预测
L-M法
BP神经网络
high-filled embankment
foundation settlement prediction
Levenberg-Marquardt method
BP neural network
