Using the Shuffled Complex Evolution Global Optimization Method to Solve Groundwater Management Models

摘要

The shuffled complex evolution method developed at the University of Arizona (SCE-UA) is applied to optimize the management models of groundwater in the paper. It is the first time to use this method in hydrogeology field. Different from traditional gradient-based optimization methods, such as Difference Dynamic Programming (DDP), Sequential Quadratic Programming (SQP), etc., the SCE-UA algorithm is capable of finding global optimum and it does not rely on the availability of an explicit expression for the objective function or the derivatives. Making use of virtues of two types of non-numerical approaches: search along definite direction and search randomly, and introducing the new concept "complex shuffling", the SCE-UA is a very effective and efficient global optimization method and can be used to handle nonlinear problem with high-parameter dimensionality. Two management models of groundwater resources are built in an unconfined aquifer: linear model of the maximum pumping and nonlinear model of minimum pumping cost. The SCE-UA method and some other optimization methods are used to solve these two models at the same time. Comparison of the results shows that the SCE-UA method can solve the groundwater model successfully and effectively. It is obvious that this method can be used widely to optimize management models in hydrogeology field, such as configuration of groundwater resources, prevention and manage of groundwater contamination, etc.