Data: 17/09/2014 à 19/09/2014
Local: São Paulo - Brazil
Assessment of Genetic Algorithms in Calibration of An Unsaturated Soil Water Flow Model Applied to a Vertical Infiltration Experiment of Two Layers of Soil (PAP014901)
Land use and Floods, landslides and erosion
This paper aims at assessing the genetic algorithms for automatic calibration applied to a model that simulates an experimental study on water infiltration in two soil layers. There is the problem of interface between layers of soil. The numerical model of Koide (1990) was used. With the computational progress, the researchers chose to use the inverse model for calibration models. That inverse model was utilized in connection with Genetic Algorithms (GA) that is a mathematical technique that optimizes an objective function (OF). The use of a single FO drifts problems of uniqueness. This article uses the colected data from the phenomenon of infiltration of two layers of sandy soils accomplished by Elmaloglou et al. (1980). The results show that the simulation of the experiment, including the interface between layers, is reasonable and there are multiple sets of parameters as responses that calibrate the model.