The Study On The Method Of Predicting The Failure Time And Early Warning Threshold Of The Creep Landslides

China is one of the countries with the most serious landslide disasters in the world.The landslides have caused great damage to our natural environment as well as people's safety.However,the prediction of landslide is a most effective way to reduce the loss caused by landslides.In particular,the prediction of time of failure of slope and threshold setting are the key.In order to achieve accurate prediction of creep landslides,a new method for predicting the failure time of landslides and a new method for setting threshold are proposed which are based on theoretical derivation,statistical analysis and field engineering experience,and the accuracy and reliability of the two methods are illustrated by theoretical analysis and practical analysis.At the same time,the corresponding landslides early warning system(LEWS)is also constructed.Based on the phenomenological theory put forward by Voight(1988)which describes the function relationship of velocity and acceleration when the materials reach failure,the theoretical formula of inverse-square-root-acceleration(INSRA)method is deduced.The INSRA method innovatively takes the acceleration which has better time-sensibility as the direct parameter to calculate the predicted failure time of the slope,and the method of ordinary least squares(OLS)is used to achieve the INSAR-time scatter fitting diagram,while the mechanical analysis method is applied to calculate the acceleration of slope at the time of failure.Five engineering cases are analyzed with monitoring deformation data,including four landslides and one slope with significant deformation which did not reach failure.The results show that:(1)The applicability,security and accuracy of the INSRA method are preliminarily proved by comparing the case application of the INV method and the INSRA method.(2)In principle:On one hand,the accuracy is promoted by calculating the acceleration at the time of failure with mechanical analysis.On the other hand,the wrong predictions can be reduced because of the lag between the velocity and the acceleration.(3)The case study shows that the longer length of the data time window used for fitting,the more stable the prediction result is,and the closer it is to the real result.Threshold setting can trigger the alarm of slope failure,but the applicability and accuracy of the existing threshold criterion for slope failure are limited.In order to solve that problem,three dimensionless parameters are proposed in this paper which are based on the slope displacement,velocity and acceleration.And a statistical analysis on the slope deformation monitoring data of 80 slopes from 58 literatures.By using probability statistics,the two levels dimensionless failure threshold is obtained.The dimensionless threshold is successfully used for the early warning of #2 landslide in a copper mine in Jiangxi province,the statistical and analytical results show that:(1)“Two-sample heteroscedasticity t-test” method is used to prove that the dimensionless parameters in the damage cases are significantly greater than the corresponding dimensionless parameters in the cases which did not reach failure.(2)The influence of the direction of the vector parameter can be eliminated,thus the dimensionless threshold is universal in a physical sense.(3)The applicability and accuracy of failure threshold are improved by the application of dimensionless failure threshold to a certain extent,which is demonstrated by "Receiver operating characteristic curve(ROC)" method,and it is explained theoretically that the prediction accuracy of dimensionless acceleration threshold is the highest.(4)The warning time of the three dimensionless thresholds in the case is not the same,and the warning time of the dimensionless second-level acceleration threshold is earliest.Finally,the framework of Landslides Early Warning system(LEWS)is constructed by combining the slope failure time prediction method-INSRA method and the threshold setting method-dimensionless threshold,which provides theoretical basis and technical guidance for the prevention and control of landslides.By the analysis of a landslide case,it is proved that the warning system proposed in this paper can achieve accurate and safe landslide prediction: the failure time of slope is predicted accurately,and the safety early warning of slope failure is realized.