ABSTRACT: Ultimate bearing capacity of soft ground reinforced with stone column was recently predicted using various artificial intelligence technologies such as artificial neural network because of all the advantages that they can offer in minimizing time, effort and cost. As well as, most of applied theories or predicted formulas deduced analytically from previous studies were feasible only for a particular testing environment and do not match other field or laboratory datasets. However, the performance of such techniques depends largely on input parameters that really affect the target output and missing of any parameter can lead to inaccurate results and give a false indicator. In the current study, data were collected from previous related literature including parameters handling the behavior of stone column and governing its bearing capacity. They included some parameters that were not considered previously; the undrained cohesion of soil, angle of internal friction and modulus of elasticity of fill material, area replacement ratio, and length to diameter ratio. The new model was generated using Neural Network Toolbox in MATLAB, all the five key parameters were treated as input data while the bearing capacity as the output data wanted to be predicted. A single hidden layer of twenty artificial neurons has been adopted in the generation of the model. The results and the regression analysis showed a high potential of using neural artificial network method in predicting the ultimate bearing capacity of soil strengthened with stone column. Thus, the study contributes in producing a reliable outcome as an alternative to using findings of costly and time consuming field or experimental tests
