An artificial intelligence optimization method of back analysis of unsteady-steady seepage field for the dam site under complex geological condition.

The investigation of seepage situation of a reservoir under complex geological conditions is greatly significant and is necessary to be determined by geologic survey or numerical analysis. Based on previous geological survey data, a 3-dimensional seepage back analysis was conducted. In this study, two hydraulic conditions of unsteady and steady seepage were considered, and the corresponding back analysis came down to a multi-objective decision-making problem. The GRNN model was trained by PSO algorithm for obtaining the relationship between permeability coefficients and monitoring data, and by combining the NSGA-II algorithm, the best unbiased solution of permeability coefficients for the multi-objective function established by monitoring data of seepage discharge and pressure head was searched via iteration calculation. On this basis, the seepage safety of the reservoir was evaluated. Through taking the seepage discharge of the whole reservoir into account, the problem in the traditional seepage field back analysis that the foundation strata seepage parameters are insensitive to the pressure head inside the dam was solved. And the inversion results can provide a supplement for the geological survey under complex geological conditions. The method adopted in this paper may provide significant references for the anti-seepage design and reinforcement of hydraulic structures with complex geological conditions. [ABSTRACT FROM AUTHOR]