分析抽水试验数据的单纯形差分进化混合算法

The Simple Method-Differential Evolution Combined with Algorithm for Analyzing Pumping Test Data

分别以无限含水层和有直线隔水边界含水层情况下的解析解为基础,应用提出的单纯形差分进化混合优化算法求解分析2种条件下的抽水试验数据,确定含水层参数的函数优化问题。将具有全局搜索能力强、原理简单、受控参数少、而局部搜索能力弱等特点的差分进化算法与具有局部搜索能力强、运算速度快、而对参数初值的选取依赖性较强和易于陷入局部极值等特点的单纯形优化算法进行结合,构成了一种混合优化算法,即单纯形差分混合优化算法。这种混合算法同时具有确定性运算和随机性搜索所具有的共同优点,能够较好地平衡全局搜索能力和局部搜索能力。数值实验结果表明,单纯形差分混合优化算法能够有效地应用于分析抽水试验数据,识别含水层参数;与其他方法相比较,其具有运算速度快、效率高和计算结果精度高等优点。

Based on the analytic solution of infinitely aquifer and linear impervious boundary case, the simple-differential evolution combined algorithm was put forward and applied to analyze the data of pumping tests for estimating aquifer parameters. This paper combines Differential Evolution algorithm, which has some characters, such as strong global search ability, simple principle, less controlled parameters, but has weak local search ability with Simple Method optimization algorithm which has this characters, such as strong local search ability, high speed operation, but has higher dependence on the selection of initial parameters and easily trapped into local extreme value. By doing this,it obtains a combined optimization algorithm, namely Simple Method-Difference Evo lution combined optimization algorithm The combined algorithm possesses common advantages of random operation and deterministic algorithm, and this method can balance the global and local search ability preferably. The results show that the method may be suc- cessfully applied to analyze the data of pumping tests for estimation of aquifer parameters. Comparing the other intelligence optimiza- tion algorithm, the combined algorithm has the advantage of rapid convergence rate, high optimization efficiency and high accuracy.