Abstract
Groundwater pollution source identification is the first step of groundwater remediation planning and an important link of groundwater pollution source identification. In order to make the inversion result more accurate, the bilevel programming is proposed to test its availability. In this paper, iterative time, optimization algorithm and noise level are used to test the accuracy of bilevel programming. The advantages and disadvantages of genetic algorithm, multi-objective genetic algorithm and simulated annealing algorithms are compared in searching optimal decision variables. The results show that the combination of multi-objective genetic algorithm and bilevel programming has better accuracy and robustness. The bilevel programming can be used to identify the groundwater pollution source and aquifer parameters.