基于连续不确定区域的遗传算法求解水文地质逆问题

Sequential uncertainty domain based genetic algorithm for inverse estimation of hydrogeology parameters

提出一种新的基于连续不确定区域的遗传算法求解水文地质逆问题。该方法运用概率与统计理论 ,将参数求解空间分为连续的不确定区域 ,并分别在各个区域利用改进的遗传算法计算各组参数对应的目标函数 ,然后通过比较目标函数在容许值内的频率分布来缩小参数的识别空间 ,从而使参数逐步达到最优。与传统的试估-校正方法类似 ,该方法实质上也是一种逐步逼近的迭代过程 ,但通过概率论方法处理 ,可大大节省传统的简单遗传算法识别参数所需的时间。

A sequential uncertainty domain based genetic algorithm(SUDGA)for inverse estimation of hydrogeology parameters is presented.The domain for determining the parameters is divided into a sequential uncertainty parts and a modified genetic algorithm is applied to calculate the corresponding objective function for each part.By comparing the frequency distribution of hit scores of objective functions within a certain tolerance the solution domain can be narrowed and the parameters is optimized step by step.This method is more efficient in computer consuming than the simple binary genetic algorithm.