Determination of Compacted Clay Permeability by Artificial Neural Networks

In many civil engineering practices like design of landfills, earth dams, pavements and agricultural issues, it is necessary to know the permeability coefficient of soils. In situ tests are frequently used to predict permeability of the soil. On the other hand, high ability of Artificial Neural Networks (ANNs) in prediction of nonlinear behavior has attracted the attention of many researchers. In present research a prior attempt to predict permeability of fine soils using Artificial Neural Networks is reviewed. In the prior research both soil physical properties and also compaction properties are used as ANN inputs. In present research, it is tried to show that if the dry density value after compaction be used as an input of model, other compaction properties can be omitted without any significant influence on accuracy of the prediction, and also it is tried to decrease the number of inputs in soil property group while the improved accuracy is obtained.