Research On Reliability Of Open-Pit Mine Slope Based On Spatial Variation From The Perspective Of Unsupervised Learning

Open-pit mining is one of the main ways to efficiently develop resources worldwide.In the process of production and construction,slope is the most common engineering form.With the influence of natural environment and human activities,the stability problem has gradually become prominent.For a long time,a certain safety factor has been used in engineering to evaluate the stability of slopes,often ignoring the existence of objective uncertainty and spatial variability of rock and soil,and overestimating or underestimating the stability of slopes,which is difficult to achieve safety.and economical balance.Especially in the environment of open-pit mines,slope stability is controlled by both structure and rock-soil strength,and the variability of characterization parameters has a significant impact on slope stability.In this thesis,the rock slopes of copper open-pit mines controlled by structure and the soft rock slopes of open-pit coal mines controlled by strength are taken as the research objects,and the research methods of site investigation,laboratory test,data analysis,numerical simulation and other research methods are integrated to realize the spatial analysis of rock and soil mass.Variability characterization,stability research and optimization design based on reliability theory,in order to achieve a win-win situation of safety and economy.The main research contents and results are as follows:(1)Combining site survey and 3D point cloud technology with MATLAB machine learning algorithm model to identify,extract and statistically analyze the occurrence characteristics of open-pit mine slope rock and soil spatial structural plane;Samples of soil strength characteristic parameters and construct a statistical model of parameter distribution;with the help of MATLAB platform,Bootstrap-ML method is used to optimize the distribution statistics of high variability parameters that characterize structural surface occurrence and strength characteristics,and obtain reliable variability characterization.(2)Using the MATLAB platform,a Hierarchical clustering algorithm based on machine learning was developed to analyze the variation degree of the representation parameters.Variation degree analysis includes: cohesion,internal friction angle,bulk density,inclination,inclination angle,structural surface friction angle and the correlation between parameters,and select parameters with high degree of variation as random variables for reliability research.The reliability index of open-pit mine slope with multiple factors such as slope engineering safety grade,slope service life and slope failure consequences is constructed to provide a reference for reliability and stability evaluation.(3)On the basis of spatial variation characterization,kinematic reliability analysis of S-Kmeans algorithm and reliability analysis based on spatial variation limit equilibrium,as well as finite difference three-dimensional numerical simulation and Box-Behnken stochastic response surface collaboration are used for different engineering scenarios Reliability analysis method combined with Monet-Carlo and Latin-Hypercube technology to achieve reliability evaluation;and through ANOVA variance analysis to realize the significance analysis of parameters on slope stability and the influence of combination effect on slope stability.(4)According to the dual indicators of slope safety factor and reliability,the south side slope is optimized.The calculation results show that for every 1° slope angle reduction,the safety factor only increases by 4.73%,while the reliability increases by 23.77%.Accordingly,two kinds of slope optimization design schemes,economical and conservative,are proposed,and targeted optimization and reinforcement measures are put forward according to the significant research results of the influence of parameters on slope stability.There are 64 figures,28 tables and 167 references in the thesis.