Reliability And Probabilistic Analyhses Of Large Deformation Failure Of Rock Slopes Considering The Spatial Variability Of Rock Mass Parameters

With the continuous expansion of the scale of basic engineering construction in my country,slope stability analysis and landslide risk control have become the focus of geotechnical engineering research.At present,a lot of useful researches have been carried out to tackle the difficulties in reliability analysis of rock slopes,but few studies which is on the characterization of the spatial variability of rock mass mechanical parameters,the reliability of rock slopes with wedge rock mass and anti-dip rock mass,and the failure probability of large deformation of the slope,lack theoretical basis.In this paper,a series of research work is carried out in the aspects of spatial variability characterization of rock slope parameters,large deformation failure mode and reliability analysis.The main contents and conclusions are as follows:(1)For rock slope models with different failure modes(arc failure,plane failure,wedge failure and reversed-dip failure),the corresponding non-Gaussian random field characterization method for the cross-correlation of rock mass and structural plane parameters is proposed,and the objective characterization of the spatial variability of mechanical parameters of rock mass and discontinuity is realized.(2)The autocorrelation distance and fluctuation range of different types of rock parameters were systematically counted.The correlation distance of the parameters of the surrounding rock of the tunnel is relatively large,ranging from 0.1 to 100 m,mainly around20 to 90 m;The correlation distance between parameters of highly weathered slope rock mass is 0.5-30 m;The relative distance of rock mass parameters in discontinuity is generally 2-30 m,and the value in horizontal direction is about 6-20 times of that in vertical direction.The exponential autocorrelation function is the most widely used theoretical autocorrelation function in the characterization of the spatial variability of rock mass parameters.(3)On the basis of using the Karhunen-Loève series expansion method to characterize the spatial variability of rock mass and structural plane parameters,the explicit functional relationship between the slope safety factor and the spatial variation parameters is constructed based on Hermite stochastic polynomial or back-propagation neural network expansion.,proposed a non-intrusive stochastic finite difference method for slope reliability analysis considering the spatial variability of rock mass parameters,and developed the FLAC3 D software interface program for slope reliability analysis.This method can not only consider the influence of the non-stationary distribution characteristics of rock mass parameters,but also can efficiently solve the reliability of rock slopes with different failure types,which provides a novel means for reliability analysis of complex rock slopes considering spatial variability of rock mass parameters.(4)By constructing a principal component analysis-support vector machine(PCA-SVM)model to judge the unstable state of the slope,a random material point method is developed for probabilistic analysis of slope large deformation the considering the spatial variability of parameters,and effectively simulate the evolution process of the large deformation failure of the landslide,the failure consequences of slope instability are quantitatively evaluated,quantitative evaluation of landslide probability and disaster risk can be quickly carried out.In the meanwhile,the evolution law of the failure modes of the shallow,middle and deep layers of the slope is revealed.