Prediction of Safe Bearing Capacity with Adaptive Neuro-Fuzzy Inference System of Fine-Grained Soils

Abstract

A lot of fieldwork is required to assess the safe bearing capacity (SBC) of fine-grained soil using IS Code, along with performing shear parameters to determine angle of internal friction and cohesion. Standard penetration tests are conducted in order to obtain N-value of soil, and evaluating atterberg limits and dry soil density. Here, it is proposed that Adaptive Neuro-Fuzzy Inference System(ANFIS) is adopted to predict fine-grained soil's safe bearing capacity. For this, input parameters considered for ANFIS system are depth of foundation, dry density, liquid limit, plasticity index, Percentage fine fraction, width/Length ratio, and N-Value. A wide range of safe bearing capacity data from various site locations was investigated and trained on. Four different models were developed with variations in membership function for each input, all the models are used with a gaussbell type of membership function. Among the four, the third model is predicting the nearest value with an R2 of 0.9738. Based on the conclusion the ANFIS model is the most reliable technique for assessing the SBC of soils. Investigation of soil properties and estimation of safe bearing capacity will be having more difficulty with respect to skilled person to investigate and time required is also more as dimension of the footing changes SBC also varies. So, to overcome this type of problems my model will give you a best suitable and reliable SBC.

Keywords

• safe bearing capacity (sbc)
• liquid limit
• adaptive neuro-fuzzy inference system (anfis)
• dry density
• plasticity index
• n-value