基于两种耦合方法的模拟-优化模型在地下水污染源识别中的对比

Simulation-optimization model of identification of groundwater pollution sources based on two coupling method

采用一种基于模拟-优化模型的方法对地下水污染源进行识别研究.模拟-优化模型分别采用响应矩阵法和状态转移方程法进行耦合,并利用遗传算法求解,经过多次迭代,使得模拟结果与观测数据的误差达到最小.最后,通过一个假想例子评估模拟-优化模型的性能,同时比较应用不同耦合方法的计算结果.研究表明：应用响应矩阵法耦合模拟-优化模型所得结果的绝对误差范围为0.1～1.6g/L,应用状态转移方程法时,绝对误差范围为0～5.2g/L,因此,采用遗传算法求解模拟-优化模型能够有效且准确地得到地下水污染源的释放量,可以应用于地下水污染源识别问题,并且采用响应矩阵法耦合模拟-优化模型优于状态转移方程法。

This paper presents a study of groundwater pollution sources identification using simulation-optimization model based on the Genetic Algorithm. The simulation-optimization model was coupled with response matrix approach and state transition equation method, which was solved with genetic algorithm, and the minimum error between the simulated results and the observed data was obtained after multiple iteration. The performance of the model was evaluated with a hypothetical example. The simulated results indicate that the absolute error of response matrix approach and state transition equation method was 0.1-1.6g/L and was 0-5.2g/L, respectively. It proves that the simulation-optimization model based on the genetic algorithm for identifying single groundwater pollution source was feasible. Also, response matrix approach was superior to the state transition equation method.