The Finite Element And Genetic Algorithm Used For Seepage Inverse Problem Of High Dams

The seepage problem of the high dam project is very complicated. Obtaining the seepage parameter is the key of resolving the seepage field. At present, in test measure, either the inside experimentation or original location test is limit to get the exact seepage parameter which accords with fact. So it is very necessary to adopt the method of combining the experimentation with computation to get the seepage parameter. Combining the Genetic Algorithm with the finite element method and using the method of straight back analysis to compute the seepage coefficient. Major findings are as follows:1. In simple genetic algorithm, it is very fussy to ensure the crossover ratio P_c and the mutation ratio P_m by repeating experimentation for the different condition. Moreover, it is difficult to obtain the best value which is fit to the every problem. Aiming at the problem, the adaptive genetic algorithm program is compiled. The P_c and P_m may change with changing of the fitness in the program. The best P_c and P_m is gained to the solution and the computational convergence of genetic algorithm is assured while the variety of population is retained.2. At the same population size and outside conditions, the convergence speed of adaptive genetic algorithm is faster than the simple genetic algorithm for the process of computation. At same time, the simple genetic algorithm is easier to obtain the solution prematurely than adaptive genetic algorithm and can not get the best solution. So the adaptive genetic algorithm is better than simple genetic algorithm.3. According to the calculation case, the simple genetic algorithm hardly obtain the best solution in 2-D and 3-D engineering example while the population size is small. Commonly, the algorithm is improved while the population size is added. However, in the most condition, the simple genetic is not as good as the adaptive genetic algorithm while the Genetic Algorithm's population size is little bigger than adaptive Genetic Algorithm. At same time, the population size added must increase the computation time which adds more difficulties to practice application. More over, the computation result may not be obtained4. At the same population size and other condition, the error of the real heads and the computation heads is very small while the K_x and K_y are obtained by using the triangle element and quadrangle element method, and the error of the K_x and K_y is also small. The fact shows that the triangle element and quadrangle element method are feasible to back analyze the seepage coefficient in the 2-D seepage inverse analysis.5. From the computation results of the engineering example in this text, the error of computation value of seepage coefficient in the same orientation is very small to the same segment of dam while using the quadrangle element method or the cube element method, and the seepage coefficient is also reasonable. However, the seepage coefficient is smaller than the middle part of ground's which is tested. The main reason is that the influence of curtain grounting is not considered. The back analysis programs are feasible and valuable from the computation results.