单纯形—模拟退火混合算法反求水文地质参数及其并行求解

本文利用单纯形法局部搜索速度快和模拟退火算法全局寻优的特点,同时为了克服各自算法的弊病,提出采用单纯形—模拟退火混合算法(SMSA)进行水文地质逆问题的求解。论文详细描述了SMSA算法的具体操作算子的实现,并将该算法应用于一个大型水源地的地下水模拟反演。计算结果表明,SMSA算法在水文地质参数反演计算具有求解速度快,精度高的特点,而且易于实现并行运算。

Hybrid Simplex-Simulated Annealing Method for Solving Hydrogeologic Parameters and Its Parallel Implementation

The local search methods such as Simplex method are quite efficient, but not very robust. That is, they normally converge in a small number of iterations, but often fall into local minima. On the other hand, modern heuristics such as Simulated Annealing are very robust, but unacceptably expensive. On the basis of analysis of the advantage and disadvantage of Simplex method and Simulated Annealing algorithm, we present a hybrid method named SMSA aimed at coupling the efficiency of Simplex method and the robustness of Simulated Annealing for solving the groundwater inverse problem. In this paper, we introduce the realization of the SMSA algorithm's operators and apply it to the inverse simulation of a large water source area. As shown by the calculation results, the SMSA algorithm has the strongpoint of high precision, fast speed. Furthermore, this SMSA algorithm is easy to parallelize.