Soft Computing Based Prediction of Deviator Stress of Waste Plastic Reinforced Sand

Rakesh Kumar Dutta; Viswas Nandakishor Khatri; Tammineni Gnananandarao

doi:10.3759/joge.v6i3.2751

Abstract

In the recent past, the soft computing techniques have received a significant attention for solution of the geotechnical stability problems. Following the trend, the present study tries to explore the use of different soft computing techniques such as random forest regression, support vector machines (SVM) RBF kernel, SVM poly kernel and M5P model tree for the prediction of deviator stress of sand reinforced with waste plastic strips. The deviator stress (s) was assumed to be dependent on strip content (SC), strip elongation at failure (ets), aspect ratio (AR) of strip, confining pressure (CP), and strain at failure of the composite specimen (e). The performance of each soft computing technique was assessed with the help of various standard statistical parameters. The findings of the present study revealed that the SVM RBF kernel approach was quick and accurate tool for predicting the deviator stress of sand reinforced with waste plastic strips. Further the sensitivity analysis conducted for the SVM RBF kernel indicated that the strip content (SC) affected the prediction of deviator stress of sand quite significantly.