广义Bayes法在裂隙岩体渗透系数随机反演中的应用

工程中由于存在各种随机因素,采用确定性的渗流分析方法进行渗透系数的反演必然会导致结果的不确定性.本文基于渗流场的随机有限元分析方法,结合变尺度优化算法和广义Bayes法,建立了一种渗透系数的随机反演方法,推导了详细的计算公式.该方法不仅考虑量测水头、渗透系数的随机性,还考虑了边界水头的随机性,不仅可以获得渗透系数的均值反演结果,还可以得到标准差的反演结果.最后将该法应用于重力坝坝基渗流算例分析中,以渗流有限元正分析计算结果作为"假想"的实测点水头值,通过随机反演,同时获得渗透系数均值与标准差的反演结果,将输入信息与反演结果对比分析,验证了渗透系数和标准差反演结果的正确性.

Application of generalized Bayesian method to stochastic analysis of fissured rock permeability coefficient

The random factors in engineering practice, such as randomness of water head on boundary, error of monitored water head and other random factors always result in the uncertainty of permeability obtained from reverse analysis using deterministic seepage analysis method. In this paper, a method for stochastic inverse analysis of permeability coefficient based on the stochastic finite element analysis method and applying variable metric algorithm and generalized Bayesian method is developed. By using this method not only the randomness of the monitored water head and permeability but also the randomness of boundary water head can be considered. The inversion results include the mean permeability coefficient and standard deviation of permeability coefficient. The validity of the proposed method is verified by the seepage analysis result of a gravity dam foundation. The results obtained from stochastic finite element analysis of seepage field are regarded as the assumed measured water head. The mean permeability coefficient and standard deviation are obtained from inverse analysis according to this assumed water head. The inversion results are in good agreement with the input data.