数据预处理在滑坡位移相关分析中的应用

Application of the Data Preprocessing Methods to the Correlation Analysis of Landslide Displacement

滑坡位移的变化除受控于其内在的地质条件之外,也受到外在诱发因素动态作用的影响。为更好地揭示滑坡位移变化与诱发因素的综合关系,运用多种数据预处理方法对滑坡位移监测数据进行了处理分析。以巫山监测预警示范站玉皇阁崩滑体为例,分别采用t检验准则、正交多项式及三次样条函数进行数据剔除处理、数据平滑处理及数据等距化处理,在得到有效数据的基础上,对滑坡位移量与降雨量、库水位变化等诱发因素进行了相关分析。结果表明,滑坡位移量与降雨量、库水位变化等诱发因素存在较强的相关性,同时具有典型的滞后性,其中,连续降雨与库水位下降是滑坡变形主要的直接诱发因素,且连续降雨对滑坡位移影响的滞后时间为0～6d,库水位下降对滑坡位移影响的滞后时间为0～21d。

The variation of landslide displacement is not only determined by the internal geological conditions,but also dynamic functioning of the external inducing factors.In order to analyze the comprehensive relationships between these factors and the variation of landslide displacement,this paper pretreats the original monitoring data with a variety of methods.With Yuhuangge landslide in the Wushan demonstration station for monitoring and early-warning as an example,using t-Test Criteria,Orthogonal Polynomial and Cubic Spline Function are used to reject data,smoothen data and conduct data isometric treatment.Based on effective data,the correlation between landslide displacement and its inducing factors such as rainfall,reservoir level changes etc,have been investigated.The results indicate that landslide displacement is strongly related to rainfall and reservoir level changes,synchronously existed obviously hysteresis.Rainfall and the falling-down of reservoir level are the main inducing factors of landslide deformation.The displacement will happen obviously after 0-6 days of rainfall and after 0-21 days of the falling-down of reservoir level.The research reveals the significance of the correlation between the landslide displacement and the inducing factors and provides a theoretical basis to establish displacement prediction model of landslide based on influence of inducing factors.