Research On Reliability Analysis Method Of Slope Stability

With the flaw of single safety factor method on slope stability analysis is clearly day by day,it is extremely urgent and important for us to seek a new and more reliable mthod.Slope stability of reliability analysis method is a vilad method,which can make up the shortage of single safety factor method.Reliability analysis method considers the variability of each register which effects slope stability,and measures the safety of slope with strict probability theory,therefore,it is more realistic to analysis slope stability and appraisal by slope stability of reliability analysis method.At present,there are three kinds of normal reliability analysis methods,such as, Mean First Order Reliability Method(MFORM),First Order Reliability Method (FORM) and Response Surface Method(RSM),but MFORM's precision is lower,while FORM and Response Surface Method have a great of calculation.So,the thesis put forwards a new reliability analysis method,called Artificial neural networks method(ANN),the method adhere to the integration of BP neural network with Monte Carlo simulation method,then analysis the reliability of the slope.BP neural network which have mighty mapping ability is used to approximate the response functions,then aclculate the function by Monte Carlo method,and finally aclculate reliability rate and damage probability.The thesis indicates that the ANN method has a higher precision,just as FORM method,while the efficiency is much more higher than FORM method and RSM method.It indicates that ANN method is more superior in reliability analysis in slope engineering.The research is supported by the National Natural Science Foundation of China (No.50748033).