中文摘要：

反问题的求解常常需要转化为非线性优化问题，其目标函数定义为观测数据与模型数据之间的残差平方和。地下水模型参数识别最常用的优化方法都是基于梯度搜索，其缺陷在于对模型参数初始估计比较敏感和局部极小问题。与传统的基于梯度搜索的优化方法相比，模拟退火算法具有良好的全局收敛特性。把含水层参数识别反问题转化为组合优化问题，提出模拟退火算法识别二维、非稳态地下水流动模型的渗透系数和储水系数的策略。反问题的不适定性由解的不唯一性和不稳定性来表征，模拟退火算法具有解决这一问题的能力。通过与梯度搜索算法相对比，数值模拟计算结果显示所提出反演方法的有效性和适用性。

英文摘要：

The solution of inverse problem usually requires nonlinear optimization of an objective function describing the difference between measured and simulated data. Most optimization algorithms used for parameter estimation in groundwater hydrology are gradient-type methods that have the disadvantages of being very sensitive to the initial guesses of parameters and being prone to converge to local minima. Compared with traditional optimization algorithms, simulated annealing algorithm is recognized to have better capability to find the global optimal solution. The inverse problem of identifying aquifer parameters is treated as a combinational optimization problem. The simulated annealing is presented to identify the transmissivity and storage coefficient for a two-dimensional unsteady state groundwater flow model. The ill-posedness of the inverse problem as characterized by instability and non-uniqueness is overcome by using simulated annealing algorithm. The effectiveness and flexibility of presented inversion technique are evaluated and compared with descent search methods.