Research On Random Parameter Identification For Seepage Problem

Stochastic methods for seepage inverse problem are studied in detail in the paper. Both the simplex method and the DFP method are referred to compare the efficiency of the direct search methods with that of the gradient methods. The differential method of stochastic FEM is used to calculate the gradient of the objective function hi DFP method. Firstly, deterministic FEM for seepage is introduced systematically and stochastic FEM for seepage is presented. Secondly, in order to study the stochastic method, the deterministic methods for seepage inverse problem are studied in detail. In this chapter, the sensitivity problem of observation data and the disposal problem of measurement points are also studied. In the fourth chapter, the stochastic method for seepage inverse problem is studied in detail. The Gauss-Markov method, the Bayes method as well as the extended Bayes method are introduced in details. On the base of comparation of several methods, the viewpoints of the author are presented in the paper. In the fifth chapter, a general stochastic optimization method梥tochastic approaching methods, is studied. Comparing to the above-mentioned stochastic methods, this method is more convenient to be applied in practice and the better optimization result is expected. Finally, a research is developed. Both the deficiency of the work and the expectation hi the field are given.Supported by the Chinese National Natural Science Foundation(59909002) and Chinese National Natural Science Foundation(59809003).