Studies On Structural Behavior Safety Evaluation Method In Earth-rock Dam

In the end of 1996,it has been showed statistically that earth-rock dam is more than 90 percent in 84,000 large, medium and minitype reservoirs and dams which have been built up. The dams play a important role in building of gross national economy, but, with dam number increasing, the used time going by, the dam age growing, a number of dams have some unsafe elements and potential danger of dams is on the increase. Therefore, it is very interesting research that analyzing relationship between every effect quantity and influence quantity of earth-rock dam and having reasonable evaluating for safety of earth-rock dam structure, according to prototype observed data of earth-rock dam.In this thesis, according to the characteristics of earth-rock dam real operation behavior, it is searched the new method of error of prototype observed data recognition and prototype observed analysis of earth-rock dam. The characteristic number method of the multi-pole fuzzy pattern recognition of earth-rock dam structure is presented. The above methods are proved by the real example.1. Recognition method of system error and lourse error of the prototype observed data of earth-rock dam is presented. Analyze reason of piezometric level of earth-rock dam and establish the statistics model of the observed data of piezometric level. Combining the measured data of piezometric level of LuHun dam, we can check and demonstrate the above theory and method.2. In view to the inquality flexibility and adaptive deficiency of general hyperbolaregression model, in this thesis, we put forward and create the hyperbola regression three-parameter model that enhances the fit and forecasting accuracy of the model. In the example of the observed data analysis of LuHun dam settlement, the hyperbola regression three-parameter model, the general hyperbola regression model, logarithm model and gray system model have been built, and analyzed results show that the forecast of the hyperbola regression three-parameter model is optimal.3. Establish the new information model of equal dimension of GM(1,1) of aging deformation of embankment dam. Through the research to dam, we find there lies an optimal dimension area. As can be showed from analyzing real example, the optimal dimension area, having relation between the quality of researched question and the characteristic of data sequence, can be defined by numerical experiment.4. point the problem of BP model such as slow convergence > easing emboging local minimum and quantity of manual adjustment, some improved methods are presented such as nonlinear normalization for weight value and threshold value, and automatic selective preference of parameter. It is proved by the real example that the above-improved methods on BP model have great betterment to increase the rate of convergence, avoid emboging local minimum and improve the accuracy. In this thesis, the modeling principle and method for Gray-BP model and BP-Residual model are presented too.5. Base on the unworthiness of the biggest degree of membership principle in the multi-pole fuzzy pattern recognition theory of engineering fuzzy sets, together withthe characteristics of observed behavior of earth-rock dam, create the new method that adopt characteristic number of absolute relative (or class) as the new index mark to judge or identify something. Together with LuHun dam, we even demonstrate the above theory and method. The analyzed results show that the characteristic number method of the multi-pole fuzzy pattern recognition theory of earth-rock dam is effective and feasible, and also provide new method and thinking of prototype observed behavior safety evaluation of earth-rock dam.