Artificial Neural Network Prediction Models for Maximum Dry Density and Optimum Moisture Content of Stabilized Soils

Abstract

Artificial neural network (ANN) has been applied to many geotechnical engineering problems and has demonstrated many modeling for soil permeability and hydraulic conductivity, soil compaction, and so on. Compaction is one of the most important coefficients for soil improvement. Since many prediction models were used to predicate the maximum dry density and optimum water content for soil alone and soil stabilized with lime, cement, and asphalt, this study is continued from previous studies to present the application of artificial neural network for modeling maximum dry density and optimum water content for soil stabilized with nano-materials. Feed-forward artificial neural network with back-propagation algorithm is utilized to construct comprehensive and accurate models relating the maximum dry density and optimum water content of stabilized soil to the properties of natural soil such as particle-size distribution, plasticity and the type and quantity of stabilizing additives. Two sets of separate ANN prediction models and statistical models, one for maximum dry density and the other for optimum water content are developed for three types of nano-materials, and a combined ANN model for both maximum dry density and optimum water content outputs is developed. The maximum dry density and optimum water content data were trained with the soil’s classification properties and the type and quantity of nano-material additives. A comparison with the test data indicated that the accuracy of the prediction of ANN models was more than 97%. Moreover, the results show that the ANN models are better than the statistical models.