The purpose of this study was to improve the optimization model for predicting more parameters in more difficult conditions (more grid cell numbers and high time interval numbers) than other studies in groundwater flow modelling. Also, the model needs fewer observation numbers for estimating parameters than other studies. In the present study, an optimization model based on model calibration was developed to estimate simultaneously four groundwater flow parameters – hydraulic conductivity, transmissivity, storage coefficient and leakance. The modified clonal selection algorithm, a class of artificial immune systems, was used as a heuristic optimization method. In order to simulate the groundwater flow, MODFLOW was used in conjunction with the model in MATLAB. The input files for MODFLOW were obtained by GMS groundwater simulator. The model was applied to two different hypothetical groundwater systems (two- and three-dimensional) under transient conditions to evaluate its performance. The results showed that the model was feasible for groundwater flow modelling and it could determine the groundwater flow parameters successfully with less observations and more grid cell numbers than the other studies.

这项研究的目的是要改进优化模型，以在比地下水流建模中的其他研究更困难的条件下（更多的网格单元数和较高的时间间隔数）预测更多的参数。而且，与其他研究相比，该模型需要较少的观察数来估计参数。在本研究中，开发了基于模型校准的优化模型，以同时估算四个地下水流量参数–水力传导率，透射率，储水系数和渗漏。改进的克隆选择算法是一类人工免疫系统，被用作启发式优化方法。为了模拟地下水流量，将MODFLOW与MATLAB中的模型结合使用。MODFLOW的输入文件是通过GMS地下水模拟器获得的。该模型在瞬态条件下应用于两个不同的假设地下水系统（二维和三维），以评估其性能。结果表明，该模型对于地下水流模拟是可行的，与其他研究相比，该方法可以成功地确定地下水流参数，观测值更少，网格单元数更多。