基于BP神经网络的滑坡稳定性分析  

Analysis of landslide stability based on BP neural network

作　　者：李朋丽[1] 田伟平[1] 李家春[1] 

机构地区：[1]长安大学特殊地区公路工程教育部重点实验室,陕西西安710064

出　　处：《广西大学学报（自然科学版）》2013年第4期905-911,共7页Journal of Guangxi University（Natural Science Edition）

基　　金：西部交通建设科技项目(2006 318 000 07)

摘　　要：为了对滑坡的稳定性进行分析,文中采用BP神经网络的方法,通过收集滑坡数据作为研究样本,选取平均坡度、前缘临空坡度、滑面坡度、多年平均最大日降雨量、近期滑动情况、人类工程活动6个因素作为滑坡的影响因素。并应用Matlab工具箱对神经网络进行了设计、权值初始化、数据归一化及网络训练,提出了滑坡稳定性分析模型,对误差进行分析。其结果表明样本训练结果和实际情况基本吻合,证明该模型能够符合工程实际。Through collecting the landslide data as the research sample and selecting average slope, slope of the invading front, slope of the sliding surface, the maximum annual average rainfall, the recent slide situation and human activities as the influence factors of landslide, BP neural network method is used to analyze the stability of the landslide. Applying Matlab toolbox to design neural networks, weight initialization, data normalization and network training, a landslide stability analysis model is put forward, and the error analysis is conducted. The results show that sample training re- suits agree with the actual situation basically, demonstrating that this model tally with the project re- ality.

关 键 词：滑坡 BP神经网络 非线性 稳定性分析 

分 类 号：U416.14[交通运输工程—道路与铁道工程]