| The ability of artificial intelligence systems to facilitate the generation of solutions for difficult problems in civil engineering that require symbolic reasoning and efficient manipulation of diverse knowledge has generated a considerable interest recently. The process of contaminant transport in the subsurface environment is a very complex problem, which can benefit from such system. Several sophisticated theoretical models have been developed to predict the process of solute transport in porous media. Unknown parameters, which are extremely difficult to determine in the laboratory, are introduced with each developed model creating an added difficulty. In this study the parameters of a complex contaminant transport model are related to a single parameter named the Equivalent Dispersion Coefficient (EDC). This study also presents the development and design of artificial neural network (ANN) model that is able to predict, the EDC for a contaminant transport model. The work consists of several tasks starting with generation of data from numerical simulations. Next the data is used to train the neural network models. Two learning algorithms, Back Propagation Algorithm and Levenberg Marquardt Algorithm are used to train the neural network models on a specific range of data. Once the training is complete these models are validated using synthetic as well as experimental data. The trained models give the EDC values when it receives the dimension less parameter data as their input. The EDC is used in linear equilibrium advective dispersive equation to predict solute concentrations. The results obtained from this new approach are in good agreement with the previous studies. In addition to this, the advantage of the current approach is the automated process of obtaining EDC values. |
