基于MEC-BP神经网络的群桩轴力预测  

Pile Group Axial Force Prediction Based on MEC-BP Neural Network

作　　者：王春晓[1] 陈志坚[1] 

机构地区：[1]河海大学地球科学与工程学院,南京211100

出　　处：《中国煤炭地质》2017年第3期53-57,共5页Coal Geology of China

基　　金：国家"十一五"科技支撑资助项目(2006BAG04B05);国家重点基础研究发展计划(973计划)项目(2002CB412707)

摘　　要：大型深水群桩基础易受到复杂的环境影响,其基桩轴力的变化情况与环境因素之间表现为复杂的非线性关系。综合考虑影响深水群桩基础轴力的环境因素相关参数,分析苏通大桥的原型监测数据,建立BP神经网络预测模型,并以此为基础,构建出基于思维进化算法(Mind Evolutionary Computation,MEC)的BP神经网络轴力预测模型,比较结果表明,MEC-BP神经网络预测结果在准确度和精确度上要明显高于BP神经网络,前者具有更强的可信度和泛化能力,在大型深水群桩基础轴力预测中具有一定的工程应用价值。The large sized deep water pile groups are easy to be impacted by complex environment. The foundation pile axial force varia- tion has presented complex non-linear relationship with environmental factors. Comprehensively considered related parameters of envi- ronmental factor impacting deep water pile group foundation axial force, have analyzed the prototype monitoring data of the Suzhou- Nantong Bridge over the Yangtze River, modeled BP neural network prediction model, on this basis modeled BP neural network axial force prediction model based on MEC （mind evolutionary computation）. The result comparison has shown that the predicted result from MEC-BP neural network is obviously higher than that from BP neural network on accuracy and precision. Thus the former has better re- liability and generalization capacity, provided with certain engineering reference value in large sized deep water pile group foundation axial force prediction.

关 键 词：深水群桩基础 思维进化算法 BP神经网络 轴力预测 

分 类 号：TU473[建筑科学—结构工程]