LAUSR

This paper describes the application of artificial neural network (ANN) and genetic programming (GP) methods in predicting the dynamic response of geogrid reinforced machine foundation bed. The dataset used in both models was determined through field bock resonance tests. A series of field tests were conducted over the unreinforced and geogrid reinforced foundation beds. The dynamic response was studied in terms of displacement—frequency variation. The response of both unreinforced and reinforced conditions was studied under six different dynamic force levels. From the experimental results, the significant improvement in the machine foundation performance was observed in the presence of geogrid reinforcement. The formulations for predicting displacement amplitude were developed using ANN and GP. In the formulation, depth of placement of geogrid, eccentric angle, the natural frequency of foundation soil system, damping ratio, shear strain, shear modulus and the operating frequency of a machine were considered as input variables. Primarily, the statistical performance of the models was compared through different performance indices. In addition, different algorithms were described to identify the ranking of influencing parameters, which affect the dynamic response of a geogrid reinforced bed. From the analysis, the resonant parameters predicted from the models have shown good agreement with the field test results. The operating frequency was found to be the most influencing parameter for determining the displacement amplitude of the machine foundation bed. Further, the performance of the GP model was found more accurate for predicting the response of a system than the ANN model.