Inversion Identification of Groundwater DNAPLs Contamination Sources

The inversion identification of groundwater DNAPLs contamination sources were studied by the combination of practical case study and theoretical analysis in this paper. Firstly, based on the field investigation, dynamic monitoring, data analysis and professional knowledge, the groundwater DNAPLs contamination multiphase flow numerical simulation model was established preliminarily. The particle swarm optimization adaptive wavelet threshold method was used to improve the denoising effect of the field dynamic monitoring data. The kernel extreme learning machine (KELM) method was applied to establish the surrogate model of the multiphase flow numerical simulation model, which improved the fitting ability of nonlinear complex mapping relations of the simulation model. A mixed integer nonlinear programming optimization model for identification of groundwater DNAPLs contamination sources was established to identify discrete integer variables (the number of contamination sources) and continuous variables (the location, release history of contamination sources and simulation model parameters) simultaneously. The hybrid homotopy-particle swarm optimization algorithm was applied to solve the nonlinear programming optimization model, and the global optimal solution independent of the initial value was obtained. By designing the feedback correction iteration process, the characteristics of contamination sources and simulation model parameters were identified respectively and corrected mutually through information feedback. Finally, the joint identification of the characteristics of contamination sources and simulation model parameters was achieved. The identification accuracy of groundwater contamination sources was improved, which provided necessary preconditions for reasonable design of groundwater contamination remediation plan. The theories and methods in this paper were applied to a case study of a chemical industrial contamination site in Jilin City for identification of groundwater DNAPLs contamination sources, and the identification results were suitable to practical situations.