Research On Prediction And Early Warning Of Deformation And Stability Evolution Of Slopes

A systematic study is carried out on prediction and early warning for slope by using measures of engineering analogy, monitoring feedback and numerical simulation, in which artificial intelligence, the modern mathematics theories and nonlinear theory are introduced. In order to study the prediction and early warning technology of slopes, slope stability will be divided into strongly stable, weakly stable and unstable. For stable slopes including strongly stable and weakly stable, a landslide case-based-reasoning(CBR) system is developed based on database, and with the CBR system and the study results related, a early warning indicators system of slope stability is proposed which includes displacement rate, growth rate of stress, vector angle of displacement, macro indicator, et al. Based on monitoring information a forecasting model of slope deformation based on the formation mechanism is obtained, by which the dynamic deformation of slopes can be predicted in combination with monitoring data and external conditions, and then the stability prediction and early warning would be realized. For unstable slopes, based on slice method a dynamic prediction model for evolution of slopes is developed, which can simulate the whole sliding process and predict maximum sliding speed and maximum sliding distance of landslide. Combined with practical work, the early-warning criterion of the slope at left bank of Jinping I Hydropower Station is given, and the deformation stage and development trend of the slope are distinguished according to the monitoring data. The main study works and conclusions are summarized as follow:(1) In order to study the early warning index of landslide, a landslide database composed of 31 typical rock landslides is established; and based on the database a landslide case-based-reasoning system is developed using engineering analogy and fuzzy comprehensive assessment model. Taking Jinping I Hydropower Station for example, five landslides that are most similar with the slope at left bank of Jinping I Hydropower Station are got. so in terms of four periods the early warning criterion of the slope at left bank of Jinping I Hydropower Station is given, based on deep analysis of deformation and failure process of the similar slopes. A fuzzy comprehensive assessment model is established to revise the criterion mentioned above, and exact interval of (1,1.2) mm/d is given which determines the displacement rate threshold when the slope at left bank of Jinping I Hydropower Station develops into the accelerated deformation stage.(2) Excavation unloading, blasting vibration, rock rheology and rainfall are four main factors which affect the deformation of slope during the construction period. With deep analysis of different factors on the mechanism of slope deformation, a forecasting model of slope deformation based on the deformation mechanism is obtained according to slope displacement monitoring data. The model can reflect role and relevance of each factor on the deformation of the slope, from which a more scientific and reasonable explanation for the mechanical mechanism of slope deformation is got. Then the early-warning criterions of the slope at left bank of Jinping I Hydropower Station are given, and the deformation stage and development trend of the slope are distinguished according to the monitoring data.(3) In particular, the one-state variable friction law used in the landslide analysis by Chau is extended to a two-state variable friction law. For translational slides, a system of three coupled non-linear first-order ordinary differential equations is formulated, and a bifurcation model of slope evolution is proposed based on state variable friction law. A linear stability analysis is applied to study the stability in the neighborhood of the equilibrium solution of the system, which can explain the nonlinear evolution mechanism of systems of engineering slopes, and provide a new instability criterion.(4) Based on traditional slice method and the assumption that the deformation of slice is continuous, a dynamic prediction model for evaluation of slopes is developed, which takes the changes in groundwater level due to rainfall into account and considers that the dissipation of the excess pore-fluid pressure in the saturated layer is governed by the Terzaghi's one dimensional consolidation theory. It is considered that the speed and deformation of the sliding mass are of inhomogeneity and changes in interaction forces are caused by the deformation of slice. The new model can simulate the whole sliding process and predict the maximum sliding speed and maximum sliding distance, which is proposed based on dynamics analysis of landslide using Newton's laws of motion.