基于GA-SVR的渗透系数参数反演方法

Methods of estimation of hydraulic conductivity with genetic algorithm-support vector regression machine

摘要: 渗透系数参数反演的本质是优化问题求解,遗传算法是一种基于自然选择和群体遗传机理的新的全局优化求解方法,可以较好地用于求解诸如渗透系数参数反演等复杂非线性组合优化问题。基于结构风险最小化原理的支持向量机具有逼近复杂非线性系统、较强的学习泛化能力,可以用来计算渗透系数参数反演过程中的测点水头值。实验表明,基于遗传算法-支持向量回归机的地下水渗透系统参数反演拟合效果良好,能大大提升区间搜索效率,避免出现局部最优解,其参数识别精度符合实际应用要求。

Abstract: The hypostasis of estimation of hydraulic conductivity is a problem of optimization.Genetic algorithm is a new method to search solution in a whole domain based on nature choice and population heredity,which can be used to solute the complex non-linear optimization problem including estimation of hydraulic conductivity.Support vector machine based on the minimum of structured risk has some characteristics including the approach to complicated non-linear system and the strong ability to learn and...