In recent years, the landslide disasters in China become more and more serious sincethe frequent earthquakes and rainstorms. If the reasonable judgment of slope stability isquantified, meanwhile, the prevention and consolidation measures are proposed in time,losses caused by landslide disasters could be significantly reduced. For the randomfactors including load environment, construction conditions and material properties inslope engineering, a random reliability method which could consider these randomfactors is used in this thesis for the stability study of soil slope.Firstly, basic on the combining theories of random analysis, finite elementmethod(FEM) and structural reliability analysis, local average method, FEMdisplacement analysis and Monte Carlo(MC) method are adopted to study the randomreliability of soil slope. Combining with the project example of soil slope, simulatedmodel is built, and the random reliability index and failure probability are obtained. Thusthe stable state of slope is judged.Secondly, taking the project example as the research object, effects of slope stabilityparameters such as slope model size, correlation lengths of soil, the variability andcorrelation of soil properties on the random reliability of soil slope are analyzed by MCrandom reliability analysis method based on FEM. Meanwhile the laws of the influenceson the random reliability index and failure probability from these factors aresummarized.Finally, radial basis function neural network is introduced to overcome thedisadvantages of MC random method, and an improved method for random reliabilityanalysis of soil slope is proposed. The result of random reliability obtained from thismethod is compared with the MC method to demonstrate its feasibility in the analysis ofthe project example. And the reasons for the errors are also discussed in this thesis. |