Predicting the Settlement of Raft Resting on Sand Reinforced with Planar and Geocell Using Generalized Regression Neural Networks (GRNN) and Back Propagated Neural Networks (BPNN)

摘要

In this study, geogrid and geocell as soil improvement methodologies had been used to improve the tensile characteristics of soil. The variation in settlement by using geogrid as reinforcing material at different relative density, different combination of reinforcement depths with different layers of reinforcement of soil with geogrid had been studied by conducting lab experiments. The results of the various experiments have been modeled using back propagated neural networks (BPNN) and generalized regression neural networks (GRNN) for predicting settlements at different combinations of placing of geogrid in soil. Both the models had been compared and it was found that in case of geogrid as a reinforcing material BPNN model with Levenberg-Marquardt algorithm gave better results compared to GRNN models. The results of experiments with geocell as a reinforcing material were also used separately with variable parameters such as relative density, depth of reinforcement, number of layers, and height of geocell for making BPNN and GRNN models.