Research On Seepage Behavior Of Earth-rock Dam Based On Uncertainty Analysis

Seepage is an important factor affecting the safety and stability of earth rock dam.Scientific research on seepage behavior of earth rock dam can provide theoretical and technical support for the safety and stability of earth rock dam.The present deterministic analysis method which is commonly used cannot consider the uncertain characteristics such as randomness,fuzziness,grey and unascertainty in the research process of earth rock dam seepage behavior.It is very important to conduct uncertainty analysis based on the existing theory for improving the accuracy and reliability of seepage behavior research.In view of the deficiency of the existing research,the theory and method of seepage behavior of earth rock dam based on uncertainty analysis are proposed,which combines probability and statistics theory,random field theory,fuzzy mathematics theory and reliability theory.The uncertainties in the prediction of permeability coefficient,inversion of permeability coefficient,numerical simulation of seepage behavior and comprehensive evaluation of seepage safety are studied,and the main achievements are as follows:(1)In view of the lack of considering the influence of dam filling and rolling parameters in the current study of permeability coefficient prediction,and the failure of reliability analysis on the prediction results,a binary analysis method of coupling permeability coefficient and reliability of earth rock dam considering the influence of rolling real-time monitoring parameters is proposed.According to the real-time monitoring system of rolling quality,the rolling parameter data and material source parameter data of test points are obtained,and the prediction model of dam permeability coefficient based on the improved neural network of elite selection strategy genetic algorithm is established.A binary analysis method for the coupling permeability coefficient and reliability of dam body is proposed based on the quantitative analysis of the influence of randomness in the prediction model by the reliability theory.The results show that the variability of the prediction results of the seepage prediction model considering the influence of rolling parameters is reduced by17.2%.Compared with BP neural network,the improved neural network of elite selection strategy genetic algorithm improves the prediction ability(relative analysis error)of the prediction model by 79.47%,which improves the accuracy and reliability of the analysis of the dam permeability coefficient.(2)The inversion analysis method of permeability coefficient based on entropy blind number theory and DREAM algorithm is proposed considering the influence of grey,unascertained,random and other uncertainties on the inversion parameters.The maximum entropy criterion is used to improve the blind number theory to deal with the uncertainty characteristics of permeability coefficient,such as randomness,gray and unascertainty.The adaptive differential evolution metropolis(DREAM)algorithm is used to deduce the posterior distribution of the permeability coefficient,and the response surface model is used to replace the forward numerical simulation model to improve the accuracy and calculation efficiency of the permeability coefficient inversion.The results show that the simulation errors of seepage flow and uplift pressure are reduced from 11.39% and 12.25% to 3.08% and 3.98% respectively after the inversion of permeability coefficient is substituted into the numerical model.(3)In view of the lack of in-depth consideration of the spatial variability of permeability coefficient in the numerical simulation of seepage behavior and the study of the probability analysis of seepage failure under the condition of mixed uncertainty,the numerical simulation method of seepage behavior based on the constraint random field and the method of probability analysis of seepage failure under the condition of fuzzy and random uncertainties are proposed.The numerical model of water gas two-phase flow is established based on the finite volume method and VOF method.The probability analysis method of seepage failure under random and fuzzy mixed condition is proposed,and the influence of uncertain factors in seepage field of earth rock dam is considered quantitatively.Through the numerical simulation of the seepage behavior and the calculation of the probability of seepage failure of an earth dam,the influence of the spatial structure of the permeability coefficient on the distribution characteristics of the seepage field is analyzed.(4)Aiming at the lack of comprehensive evaluation model of seepage safety of earth rock dam considering continuity,dynamics and fuzziness,a fuzzy evaluation method of seepage safety of earth rock dam based on functional data analysis theory is proposed.The comprehensive index system of seepage safety of earth rock dam including basic index,derivative index and numerical simulation index is established.Through the analysis of the continuity and time-varying of the evaluation index data by the functional data analysis theory,a fuzzy extension evaluation model suitable for the functional data is proposed,and the weight of the index is determined by the subjective and objective comprehensive weighting method based on the combination of AHP and multi-scale fuzzy entropy.The results show that the results of this method are basically consistent with those of fuzzy evaluation and set pair analysis,and the changing trend of indicators and the continuity of evaluation scores can be considered in the evaluation process.(5)Combined with an earth rock dam project in Southwest China,the application of research on seepage behavior of earth rock dam based on uncertainty analysis is carried out.The practice shows that the theory and method proposed in this paper are scientific,effective and reliable,which can provide theoretical basis and technical support for the seepage analysis of earth rock dam engineering.