Modeling multidimensional flow in wettable and water-repellent soils using artificial neural networks

摘要: Summary This study examined the use of three different classes artificial neural networks for modeling water flow in wettable and water-repellent soils, using both synthetic numerical data experimentally measured data. The 1D self-organizing maps (SOM) successfully rendered moisture contour transition zone wetting plumes all soil types at rates. Due to SOMs inability generate external output data, multilayer perceptrons (MLP) modular (MNN), respectively, were combined with SOM predict soils. dimensionality reduction, failed capture high content soils anomalous patterns, whereas spatial moment analysis succeeded providing an accurate, albeit indirect, description. Hence, MLP MNN applied moments. comparison between predicted experimental measures demonstrated capability Comparison two indicated no significant difference their results.