Research And Application Of Dam Foundation Uplift Pressure Model

Uplift pressure is one of the most important loads on the concrete gravity dam, its magnitude and evolution have great influence on the safe operation of the dam. It is very significant to analyze the monitoring data of dam foundation uplift pressure for checking the stabilization, monitoring the seepage feature, learning about the work efficiency of foundation curtain and drainage system, and understanding a sort of load condition that affects the dam distortion and stress. In this article, statistical model, GM (1,1) model, improved GM (1,1) model, BP and Radial basis function neural network models are set up according to the fundamental principles and methods of stepwise regression analysis, grey system, and neural network respectively. And then they are used in the project.The results indicate: (1) Statistical model can quantify each influencing factor of uplift pressure. (2) Since that the influencing factors are complex, uplift pressure can be taken as one grey system to be studied. Thus the obsession of those unsure influencing factors can be avoided. The GM (1, 1) model is applied to forecasting uplift pressure, but this model at times is poor in precision, so it is improved in initial value, background expression and residual amend. Then the improved GM (1, 1) model is used to forecast uplift pressure. And the result proves its precision is better. (3) Both the precision and calculation speed of the Radial basis function model are better than those of BP model.Through comparing these different models, statistical model can quantify the influencing factors of uplift pressure better;The hypotheses and experience in ascertaining influencing factors can be avoided when using grey model and neural network model;What's more, for forcasting, grey model is fit to the short sequence data, while neural network model is better for the long.