Inversion And Law Study Of Permeability Coefficient Of Earth-rock Dam Based On Long-term Monitoring Data

Grasping the variation of permeability coefficient in different periods is an important guarantee to accurately analyze the seepage field and the evolution of its operating state.Based on long-term monitoring data,the long running soils study the change rule of permeability coefficient of rockfill dam,the seepage finite element model is established,combined with the numerical calculation and parameter inversion,calculation to obtain the change of permeability coefficient of earth-rockfill dam,based on the data obtained from inversion of law of permeability coefficient prediction model is established,the analysis of the late change.Firstly,the finite element method was used to calculate the seepage of samples with different permeability coefficient combinations,and the corresponding permeability pressure values were obtained to form the initial sample group of "permeability pressurepermeability coefficient".In this process,in order to study the change of the permeability coefficient in all directions and the overall dynamics of the earth-rock dam,two analysis cases of the anisotropic permeability coefficient and the equivalent average permeability coefficient are considered in this paper,and the initial sample group is obtained by calculating respectively.The RBF neural network inversion model was constructed,and the initial sample groups in the two analysis cases were used to train the model respectively.After the "input-output" mapping relationship was formed,the long-term measured data of the permeability pressure were input to get the permeability coefficients in different periods.On this basis,based on the mathematical statistics theory,the prediction model of the permeability coefficient sequence is established in two analysis cases,and the variation law of the permeability coefficient of the dam body is analyzed.Taking a homogeneous dam as an example,the realization process of the research method is discussed in detail.The case analysis shows that the training errors of the RBF neural network model can converge quickly in both cases,and the permeability coefficient obtained from the inversion is substituted into the finite element calculation of the seepage field,and the calculated seepage pressure and the measured seepage pressure maintain a small error,indicating that the inversion results can accurately calculate the state of the seepage field.In the case of anisotropic permeability coefficient analysis,the horizontal permeability coefficient of the dam changes slightly,while the vertical permeability coefficient increases and tends to be stable in the normal range.Under the condition of the equivalent average permeability coefficient analysis,the equivalent average permeability coefficient of the dam increases year by year and tends to be stable.According to the application of the permeability coefficient prediction model,the fitting errors of the two prediction models are small.When the predicted permeability coefficient is substituted into the finite element calculation,the difference between the obtained permeability pressure and the measured permeability pressure is very small,which indicates that the model can accurately predict the variation of the permeability coefficient of the dam body.The results show that the method is reasonable and feasible,the inversion and prediction effect are ideal,and the variation law of anisotropic permeability coefficient and equivalent average permeability coefficient of earth-rock dam can be effectively understood.