基于主成分分析与BP神经网络的深基坑支护方案优选  

Optimization of supporting plan for deep foundation pit based on principal component analysis and BP neural network

作　　者：贺永俊 李超 

机构地区：[1]山西省勘察设计研究院,山西太原030013

出　　处：《四川建筑科学研究》2015年第4期32-35,39,共5页Sichuan Building Science

基　　金：国家自然科学基金项目(40972193);山西省科委自然科学基金项目(20110617)

摘　　要：为了对深基坑支护方案进行科学可靠的选择,利用BP神经网络强大的非线性处理性能对深基坑支护方案进行预测。由于影响深基坑支护的因素较多,输入层节点过多会降低神经网络的性能,利用主成分分析对深基坑支护的影响因素进行分析,减少BP神经网络输入层的节点数,提高神经网络性能。将建立的支护方案优选模型应用于32处实际深基坑工程,研究结果表明,主成分分析能将10项影响因素减少至5项,利用BP神经网络能够准确地对深基坑支护方案进行选择。In order to select supporting plan for deep foundation pit in a scientific and reliable way,use BP neural network which has a powerful non-linear processing performance ：to predict supporting plan for deep foundation pit. As the impact factors of supporting plan for deep foundation pit are numerous, .numerous nodes in the input layer will degrade the perfornumce of neural network. This paper would use principal component analysis to analyze the impact factors of supporting plan for deep foundation pit,and reduce the number of nodes in the BP neural input layer to improve the neural network performance. The optimization of supporting plan for deep foundation pit model was applied to 32 the acttud deep foundation pit projects,the results showed that principal component analysis reduced the impact factors from 10 to 5,BP neural network was able to select plans for deep foundation pit accurately.

关 键 词：主成分分析 BP神经网络 深基坑 支护方案 

分 类 号：TU317[建筑科学—结构工程]