基于BP算法的CFG桩复合地基承载力的神经网络预测

将BP神经网络模型应用于CFG桩复合地基承载力的设计计算中,通过对影响CFG桩复合地基承载力因素间的非线性关系分析,指出影响CFG桩复合地基承载力的主要因素有:桩的参数、置换率、土的物理力学特性、褥垫层厚度和施工工艺。为成功预测CFG桩复合地基承载力,收集了大量工程实测资料(包括地质资料、工程设计资料、静载荷试验资料),并按照神经网络的训练要求对工程资料进行了分析整理,建立了基于人工神经网络的CFG桩复合地基承载力预测模型。通过对网络的学习训练,得到的复合地基承载力预测值达到了预期精度。

NEURAL NET WORK PREDICTION ON BEARING CAPACITY OF CFG PILE COMPOSITE FOUNDATION BASED ON BP ALGORISM

Back propagation model is used in the calculation of bearing capacity of CFG piles composite foundation,according to analysis of the nonlinear relationship between the factors influencing the bearing capacity of CFG piles composite foundation,it is pointed out that its main factors are the parameters of CFG pile,displacement rate,cushion thickness and construction skill.In order to predict its bearing capacity,the authors collects and sorts a large quantity of engineering data,including geology data,design data and experimental data of static load.A new model based on neural network to predict the bearing capacity of CFG piles composition foundation is built.By learning and predicting test,the result attains the predicting accuracy.