改进的遗传算法及其在渗流参数反演中的应用

Improved genetic algorithm in back analysis for seepage parameters of fissured rock masses

利用水头实测资料,以裂隙组的渗透系数比例因子为待反演的参数向量,在采用基本遗传算法进行参数反演研究的基础上,针对裂隙岩体无压渗流参数反问题计算量过大,目标参数众多以及参数可能变化范围大等特点,提出了一种混合遗传算法求解此类问题,力求克服简单遗传算法在解决此类问题时存在的局部搜索能力弱、易出现早熟收敛及计算量大等缺陷,并通过典型岩坡渗流算例进行验证,同时给出了基本遗传算法、传统单纯形算法的反演成果。计算结果表明,该方法保持了基本遗传算法优点,并有效地提高了算法的运行效率,从而为求解裂隙岩体无压渗流参数反问题等计算量大的系列问题提供了新的途径。

Considering the characteristics of parameter inverse problems on unconfined seepage flow such as high computational cost, numerous objective parameters and wide variation ranges of parameters, and based on the study of parameter inverse problems by means of common genetic algorithm, a new hybrid genetic algorithm is proposed to overcome the limitation of common genetic algorithm such as poor local search ability, premature convergent and excessive computational cost. With water heads used as measured data and seepage coefficient scale genes of different fissured groups selected as objective parameters, a typical slope case study is given. The result shows that the method is effective in identifying the seepage parameters of fissured rock masses on the supposition that the parameter bounds are given. And the method breaks through several limitations of traditional optimization method and improves the efficiency. For the sake of comparisons, the results of the same case study with traditional simplex method used are given.