Artifical neural networks approach in classification of adapazari clays

In geotechnical engineering applications, soil classification system is one of the communication language among engineers. Engineers have discovered to transfer their experiences through this way. By only knowing soil class does not eliminate necessary detailed soil investigations and other laboratory tests which measure engineering properties. However, if an engineer knows soil class, one can have an idea about how and what kind of behavior soil will indicate under structural loads in application steps.Clay samples frequently faced in soil mechanics problems have an inevitable wide range of engineering characteristics. In this study, clay samples classification obtained by Standard Penetration Test (SPT) is researched by a new approach with ANN in Adapazari region which is located on northwest of Turkey.The ANN model is trained using liquid limit and plasticity index values of clay samples obtained from drilling studies for the definition of general ground condition of Adapazari region and also performance of the model is tested. The obtained results presented in graphical forms are evaluated. In addition, together with a study based on statistics the outputs of the ANN model are compared with the experimental data.