Study On The Seepage Flow Prediction Of Earth-Rock DAM

Seepage problem has important influence for earth-rock dam safety, and it is on the dam seepage flow situation prediction analysis,can effectively avoid accidents and reduce harm,and provide the scientific basis of the treatment of dam seepage path,in order to ensure the safe operation of the dam premise, improve the engineering benefit fully.On the previous research, and on the basis of this paper main work are as in the following aspects:(1) With two-dimensional seepage as the research object and the basic equation of saturated seepage and unsaturated seepage flow derivation method,the initial flow method is derived the finite element equations involving the free surface seepage problem,which facilitate using the finite element method to implement the earth dam seepage prediction.(2) In the comprehensive analysis advantages and disadvantages of artificial fish algorithm and particle swarm optimization,and on the basis of the characteristics of the combination of two algorithms new algorithm in stages is introduced, AFPSO algorithm,which can improve the search ability and the accuracy of convergence, and at the same time,avoid the easy to fall into local optimum defects.(3) In order to realize prediction of the earth dam seepage,seepage flow prediction model is established with support vector machine for regression(SVR),which is greatly affected by the support vector machine (SVM) method of model parameters c and g,this paper with the AFPSO algorithm can achieve great scope of the above parameters for fast,which can improve the prediction ability.(4) With finite element method for the study of a deterministic model is applied to predict the earth-rock dam seepage,the height of the earth-rock dam two-dimensional seepage models are built with the finite element software,the selection of no rain and rain more two calculation scheme,the calculation results show that the predicted values and the measured value is large, and set a model and its computing solution for a longer period of time,so for this project is not fit for giving quantitative prediction of piezometric level with the finite element model set up in this paper.(5) Seepage flow prediction model of reservoir water level,rainfall, piezometric level three related is established by this paper with the improvement of the support vector machine (SVM) method,which predict piezometric level of the height of the earth-rock dam of Naban reservoir of seven piezometric tube,and using BP neural network,and the finite element model of the development of this paper from the prediction result do a comparative analysis.The results show that the proposed improved support vector machine (SVM) method in the earth-rock dam seepage flow prediction has certain advantages.