摘要
为解决大数据量下滑坡的位移数值精确预测,采用数据挖掘技术对滑坡多源监测数据进行预处理,进而采取粗糙集理论对输入变量集进行定量评价、约减并完成滑坡变形阶段预测,在此基础上利用不同算法进行滑坡变形位移数值预测。实验显示,粗糙集对滑坡变形阶段划分的准确度达到96.5%,在此基础上利用分类回归树预测滑坡位移的精度达到6.5mm。结果表明,分阶段的位移预测方法是可行的,其提供的预测精度显著优于普通方法并且达到了工程应用的需求。
Rough set theory is introduced in variables set assessment and reduction for deformation stage prediction in the Baijiabao landslide after its multi-source monitoring data are preprocessed by means of data mining technique. On that basis, serval different algorithms are utilized to predict landslide displacement quantitatively for the purpose of comparison.The tests show that the rough set theory is capable of predicting landslide deformation stage precisely.The results obtained by the rough set contribute to improve performances of numerical prediction of the landslide displacement and the stage-divided method has an advantage over other conventional algorithm.
出处
《吉林大学学报（地球科学版）》
EI
CAS
CSCD
北大核心
2014年第3期925-931,共7页
Journal of Jilin University：Earth Science Edition
基金
国土资源部三峡库区三期地质灾害防治重大科研资助项目(SXKY3-3-2)
关键词
滑坡
粗糙集
变形阶段预测
位移预测
landslide
rough set
deformation stage prediction
displacement prediction
