Keywords: neural network, cohesion, friction angle, compaction, moisture content.

The determination of the shear strength parameters from laboratory tests is relatively time-consuming and prone to errors as a result of the complexity of the process involved. In this paper, an alternative approach, an artificial neural network (ANN) model, is proposed to estimate the shear strength parameters of a soil from compaction parameters. Soil samples were collected from an existing borrow pit in Kano state Nigeria and their basic properties were determined. The soil samples were compacted using nine different energy levels and dry densities for a wide range of moisture content were determined. Undrained triaxial tests were then conducted on the soil samples at various moisture contents and compaction energy. In all, about 120 sets of triaxial test were conducted. An ANN model was then developed using the back propagation algorithm. Out of the 120 set of results, 95 sets were used to train and validate the model while 25 sets were used to test the trained and validated model. The result of the ANN model shows a good correlation between the compaction and shear strength parameters with a training and validation correlation coefficient of 0.9926 and 0.9671 respectively. It was also found that the ANN model is quite efficient in predicting the shear strength parameters of a soil.

purchase the full-text of this paper (price £20)

go to the previous paper
return to the table of contents
return to the book description
purchase this book (price £42 +P&P)