- 3751 - Modelling of Soil Shear Strength Using Neural Network Approach Sushama Kiran* and Bindhu Lal** ****Researcher, Dept. of Civil Engineering, Birla Institute of Technology , Mesra, Ranchi, India e-mail: iamsushkiran1@gmail.com) ****Associate Professor, Dept. of Civil Engineering, Birla Institute of Technology, Mesra, Ranchi, India e-mail: bindhu_63@yahoo.co.in ABSTRACT The potential of Neural network to capture the nonlinear interaction between various input and output parameters has attracted many researchers towards investigating it. This paper presents the feasibility of ANN as prediction tool for determination of internal friction angle ‘Φ’ and cohesion ‘c’. With this context, an attempt has been made to develop a neural model from the index properties of soil consisting of water content (w), Plasticity Index (PI), Bulk density (BD), Sand % (SP), Silt%(STP), and Clay% (CP) as input parameter. A comprehensive set of experimental data obtained from Bihar were used in developing models of neural network based soil models. Linear and Non-linear least square regression (LLSR & NLSR) analysis was performed using same dataset and statistical parameters to judge the efficiency of proposed model. Apart from prediction analysis input variable importance were identified using connection weight, Garson’s weight approach and Weight bias approach. According to this analysis clay% (CP) and silt% (STP) are the most significant variables contributing in estimation of strength parameter ‘c’. However plasticity Index (PI) and dry density (DD) affects considerably in prediction of internal friction angle “φ. Moreover, on the basis of the weight and biases obtained from constitutive model formulae were derived in terms of index properties soil, without the need of very sophisticated and time consuming laboratory experiments with a minimal effort. KEYWORDS: neural network, MLP, shear strength, NLSR INTRODUCTION A loaded soil mass fails when the shearing stress induced in it reaches to limiting value causing the deformation of soil. Sinking of footing, movement of wedge soil behind a retaining wall or sliding in an earth embankment are some of the forms of shear failure. An improper estimation will constitute a serious damage to both property and life. Hence prior to designing for foundation of structure or of road embankment, retaining wall etc., determination of shear strength of soil is of utmost importance. In addition to this it has been enlisted in most complex properties of soil, as obtaining undisturbed soil sample from the field and conducting no. of triaxial shear test in laboratory is both time consuming and need careful supervision. It has created a persistent demand and interest among researchers for deriving shear strength from simple index properties of soil. Consequently there is now a tendency in countries all over the world towards building up correlation equation between the soil properties and index propertiesof soil (Obasi and Anyaegbunam, 2005).Many empirical and hyperbolic functions have already been employed for estimating shear strength parameters. (Bolton,