自适应遗传算法及其在渗流参数反演中的应用

利用水头实测资料,以渗透系数为待反演的参数,在采用基本遗传算法进行参数反演研究的基础上,针对简单遗传算法难以确定交叉率和变异率的最佳值及计算量较大、易早熟等缺点,提出以自适应遗传算法来解决工程中的这类反演问题;为力求使改进的遗传算法计算量更小,收敛性更强,同时结合简单的二稳定渗流有限元算例,在相同的情况下分别用简单遗传算法和自适应遗传算法进行了反演计算。结果表明,自适应遗传算法在保持简单遗传算法优点的同时,有效地提高了算法的收敛性,并在一定程度上克服了简单遗传算法的早熟问题。因此,自适应遗传算法为渗流领域求解反演问题提供了新的途径。

Adaptive genetic algorithm in back analysis for seepage parameter of earth-rockfill dam

Seepage coefficient of earth-rockfill dam is analyzed as the back analysis parameters by the real water head. Considering the parameter of inverse problems on seepage flow such as high computational cost, the optimization probability of crossover and mutation determinated difficultly,premature convergent, adaptive genetic algorithm is proposed to overcome the limitation of simple genetic algorithm. At the same time ,simple genetic algorithm and adaptive genetic algorithm are used to analyze the parameter of the simple two dimension fracture network under the same condition. The result shows that the method is effective in improving the computational convergence and overcoming the premature of simple genetic algorithm partly. So adaptive genetic algorithm is efficient and feasible in back analysis for seepage coefficient of earth-rockfill dam.