基于数据关联性分析的围岩离层变形预测研究  

Deformation Prediction of Surrounding Rock Abscission Layer Based on Data Correlation Analysis

作　　者：刘立强 刘洋 李怡航 张朝鹏 张茹 艾婷 LIU Liqiang;LIU Yang;LI Yihang;ZHANG Zhaopeng;ZHANG Ru;AI Ting(Yalong River Hydropower Development Co.,Ltd.,Chengdu 610051,China;MOE Key Laboratory of Deep Earth Science and Engineering,Sichuan University,Chengdu 610065,China)

机构地区：[1]雅砻江流域水电开发有限公司,四川成都610051 [2]四川大学深地科学与工程教育部重点实验室,四川成都610065

出　　处：《昆明理工大学学报（自然科学版）》2022年第6期53-60,共8页Journal of Kunming University of Science and Technology(Natural Science)

基　　金：国家自然科学基金项目(U1965203)。

摘　　要：基于“用数据说话”的思想,结合岩土工程中的监测数据,以巷道工程中顶板围岩离层变形问题为例,建立了巷道围岩离层变形值与监测数据参数间的关联性分析模型,采用线性回归、非线性回归、主成分分析和神经网络等方法对顶板离层值进行了预测,并对预测结果进行了比较和验证,多种分析方法在工程中均有可取之处,在该围岩离层变形预测中神经网络模型区别于回归分析的机械式数据拟合,其对于预测值的走向趋势判断更为准确,可以通过跳过推导本构关系的方式去得出工程参数预测结果.Based on the idea of “Data Talk”,an analysis model of correlation between the roof abscission layer value of tunnel and monitoring data was established with the monitoring data in geotechnical engineering, and the problem of roof surrounding rock abscission layer deformation in Tunnel engineering was taken an example. Linear regression, nonlinear regression, principal component analysis and neural network were used to predict the roof abscission layer, and the prediction results were compared and verified. Various analytical methods had their merits in engineering. In this roof abscission layer engineering problem, the neural network model was more accurate in judging the trend of the predicted value, which was different from the previous mechanical data fitting. The prediction results of engineering parameters can be obtained by adjusting and exploring the constitutive relationship.

关 键 词：大数据 巷道工程 顶板离层值 关联性分析 

分 类 号：TD322[矿业工程—矿井建设]