改进的模拟退火遗传算法在地下水管理中的应用  

Application of improved simulated annealing genetic algorithm to groundwater management

作　　者：李平[1,2] 卢文喜[2] 靳孟贵[1] 

机构地区：[1]中国地质大学(武汉)环境学院,武汉430074 [2]吉林大学环境与资源学院,长春130026

出　　处：《水文地质工程地质》2011年第3期9-13,共5页Hydrogeology & Engineering Geology

基　　金：国家自然科学基金项目(40672157);国家"863"计划(2007AA06Z337)

摘　　要：对于高度非线性、非凸的地下水管理模型,传统优化方法难以找到全局最优解。本文采用模拟退火遗传算法求解地下水管理模型,并从三个方面对算法进行改进:引入小生境技术,采用自适应交叉和变异概率,在选择过程中采用最优保存策略,从而提高算法的全局寻优能力和收敛速度。采用惩罚函数法处理约束条件。用Fortran 90语言编制了计算程序,并通过Schaffer测试函数验证了该算法不仅具有强大的全局寻优能力和局部搜索能力,而且具有较快的收敛速度和较高的优化精度。将该算法应用到某研究区地下水管理中,取得了较好的效果。For a highly nonlinear and nonconvex groundwater management model, the traditional optimization methods are difficult to find the global optimum. Simulated annealing genetic algorithm was used to solve a groundwater management model in this paper and was improved from three aspects to enhance the ability of global optimization and convergence speed： （1） introducing the niche technology, （2） using adaptive crossover probability and mutation probability, and （3） using the elitist strategy during the selection. Penalty function approach was used to handle the constraints. The calculation process was programmed with the Fortran 90. The algorithm was validated with the Schaffer test function. The results show that the algorithm not only has the strong ability of global optimization and local search, but also has a faster convergence speed and high optimization accuracy. The algorithm was applied to groundwater management of a certain study area, and reasonable results were achieved. This paper provides a new idea and new method for solving combinatorial optimization problems such as a water resource management model. The algorithm is highly effective in solving large-scale groundwater management problems.

关 键 词：模拟退火遗传算法 地下水管理 小生境 自适应 最优保存 惩罚函数 

分 类 号：P612.2[天文地球—矿床学]