Reliability Prediction Model Of Rainfall Type Slope Based On SVM

In people's various engineering construction activities,the problem of judging whether the slope is stable is often encountered.In the process of engineering construction in various fields,the damage caused by slope instability to people's lives and property is getting more and more serious,and people have gradually realized the importance of slope stability.According to the statistics of China Statistical Yearbook,there were 9710 geological disasters in 2016.The disasters caused by slope instability accounted for 76% of the total,and the disasters caused by slope instability accounted for a large proportion.90% of the slope failure is caused by rainfall,and the so-called "ten sliding nine water" is the reason.Therefore,the prediction of slope stability,especially the analysis and prediction of slope stability in rainy season,has important theoretical and practical significance.This article discusses the deterministic analysis method and reliability analysis method of the slope,and based on the previous research,the orthogonal test method is introduced to carry out sensitivity analysis and better combination analysis of the factors affecting the slope stability.Among the deterministic analysis methods,the advantages of support vector machines in solving small samples,non-linearity,high dimensionality,and local minimums,and the easy implementation and high accuracy of particle swarm optimization are combined.The feature extraction capability of the component analysis method(PCA)optimizes the parameters in the modeling process,and then establishes a PCA-PSO-SVM model for slope stability prediction,and verifies the accuracy of the model in conjunction with engineering examples.In the uncertainty analysis of slope,this paper introduces the merits and demerits of the accuracy and simplicity of the methods such as the center point method(MFOSM),the checking point method based on iterative algorithm(FORM),the response surface method and the Monte Carlo method.A slope reliability prediction model based on support vector machine(SVM)was established by combining FORM method with support vector machine(SVM).The reliability analysis of three sections of the same slope in rainy season is carried out by using the model.The results show that the reliability prediction model based on FORM method and support vector machine method overcomes the shortcomings of complicated calculation and low accuracy in other methods,and is greatly improved.The efficiency of slope reliability analysis is high.This model provides a new idea for the practical engineering and theoretical research of slope reliability analysis,and has a broad application prospect.Moreover,this method has a certain reference value for the application of stability analysis in other aspects.