The Research About Medol And Analysis Of Landslid Stability On Intelligence Algorithms

Landslide is a kind of important geologic hazard. It is not only a threat to mankind but also destroies environment and resource. It has been one of practical research fields to evaluate and predict landslide hazards for more than 100 years.The research mades great contribution to improve the human ability to resist landslide disaster.This paper applies modern intelligent algorithm for mining law of regional landslide stability and uses these rules to forecast the regional landslide stability.According to the uncertainty factors, the evaluation indicators are usually high-dimensional non-normal and non-linear data. It is difficult to achieve the desired effect characteristics using conventional evaluation methods for high dimensional non-normal, nonlinear data. In this paper, artificial neural networks and support vector machine are used to solve the problem of stability evaluation of landslides. By landslide stability test, the results show that artificial neural networks and support vector machine are good way to solve the problem of slope stability evaluation. However, with the study of factors and landslide samples increasing we found that the training of artificial neural network was slow and the network structure was unstable; Support vector machine method also appeared slow and the classification accuracy declined rapidly with the number of categories increasing.The factors effecting on landslide stability is various and complicated, and data collected both incomplete and redundant. Firstly this paper reduces influence factors of landslide stability using rough set, and establishes the impact index system of landslide stability.With sample numbers and landslide attributes increasing ANN and SVM would take long time to train the model and prediction precision rate would decrease. Combined the theory of RS and ANN into the field of landslide prediction is a good way. In this chapter we use RS to delete landslide attribute of Xinpin landslide and use ANN to train model and predict landslide stability. The results show that average prediction accuracy about 50/71, and meanwhile we combined the theory of RS and SVM into the field of landslide prediction to predict landslide stability of Xinping landslide. The results show that average prediction accuracy about 197/213. Compared with RS-ANN ,the prediction result based on RS-SVM model is more accurate and less influenced by sample numbers. In the end we use Grey-Markov chain to discuss landslide displacement .The Grey-Markov chain can provide landslide displacement at every moment and decrease the observation expesive.