基于PCA和BP神经网络边坡稳定性分析  

Analysis on slope stability based on combination of PCA and BP neural network

作　　者：陈建宏[1] 郑荣凯 陈浩[1] 

机构地区：[1]中南大学资源与安全工程学院,湖南长沙410083

出　　处：《中国安全生产科学技术》2014年第5期142-147,共6页Journal of Safety Science and Technology

基　　金：国家自然科学基金(51374242);全国优秀博士学位论文专项资金资助(200449)

摘　　要：基于影响边坡稳定性各因素之间具有一定的相关性和边坡工程是一个非线性、不确定的动态过程等这些特征,首次应用主成分和BP神经网络的原理和方法,建立了边坡稳定性评价模型,并应用SPSS软件对影响因素进行分析并确定主成分,应用Matlab7.1神经网络工具箱对一些边坡样本进行训练仿真。对比了经过主成分分析和未经过主成分分析评价结果,结果表明,经过主成分分析的BP神经网络评价精度更高,相对误差更小。表明了建立主成分和BP神经网络评价模型具有较好的可行性和适用性。Based on the characteristics of a certain correlation between various factors affecting the slope stability and the slope engineering is a dynamic process being nonlinear and uncertain , the methods of principal component analysis and BP neural network were applied to set up the evaluation model of slope stability .The influencing fac-tors were analyzed to determine the principal component by using SPSS software , and the network was then trained and simulated based on slope samples data set by using the NN toolbox in Matlab 7.1.The results with principal component analysis and without principal component analysis were compared .It showed that the former is better , with higher accuracy and smaller relative error , the method of principal component analysis and BP neural network established in this paper has a good feasibility and applicability .

关 键 词：边坡稳定性 主成分分析 BP神经网络 

分 类 号：X935[环境科学与工程—安全科学]