Prediction model for stability of high loess slopes

The landslides of loess districts restrict the local economic development seriously. In order to reduce the loss caused by landslides, it is necessary to reinforce the analysis and prediction on loess slopes. The stability of high loess slopes is controlled by water content of slopes because of the water sensitivity of loess. In view of that, based on various influencing factors of loess slopes, a model is presented to predict the stability of loess slopes by simulating water content. Firstly, based on the graphic method, an automatic table lookup program is developed to generate numerous representative slope data, by using these data, a predicting model is established based on the modified genetic Neural Networks. Then the model is validated to be of high precision by simulating the training data and test data. Finally, taking two engineering cases for example, the application of the model in predicting the stability of slopes with different initial states is introduced. By comparing with the values of graphic method, the results show that the predicted results agree with the expected results, indicating that the model has wide applicability in Guanzhong area.