An investigation of friction angle correlation with geotechnical properties for granular soils using GMDH type neural networks

The Standard Penetration Test (SPT) is one of the most eective tests for quick and inexpensive evaluation of the mechanical properties of soil layers. Numerous studies have been conducted to evaluate correlations between SPT blow counts (NSPT ) and soil properties such as friction angle (ʹ0). In this paper, the relation between and in situ parameters of soil, including NSPT , eective stress and ne content, is investigated for granular soils. In order to demonstrate the relevancy of ʹ0 and corrected SPT blow count (N60), a new polynomial model, based on the Group Method of Data Handling (GMDH) type Neural Network (NN), was used based on 195 data sets including three soil parameters. These were recorded after two major earthquakes in Turkey and Taiwan in 1999. This study addresses the question of whether GMDH-type NN is capable of estimating ʹ0 based on specied variables. Results conrm that GMDH-type NN provide an eective way to recognize data patterns and predict performance over granular soils accurately. Finally, the eect of ne content and eective overburden stress on the correlation of N60 and ʹ0 has been studied using sensitivity analysis.