Modelling Soil Behaviour in Uniaxial Strain Conditions by Neural Networks

It is the primary goal of this research to investigate how neural networks may be used to describe soil behavior under situations of uniaxial strains. Geotechnical engineering issues have been effectively modelled using artificial neural networks (ANNs) during the last several years, particularly soil behavior under uniaxial strain situations. The purpose of artificial neural networks (ANNs), a subset of artificial intelligence, is to simulate, as closely as possible, the human brain and nervous system. Most geotechnical engineering issues may be modelled using ANNs, which have a high degree of complexity [1]. Depending on the size of the research area, it may be necessary to drill numerous boreholes and conduct a number of experiments to determine the structure of the soil layers. It is possible to better comprehend the near surface geology by learning more about the qualities of the soil layers between boreholes. ANNs training from the samples supplied to them in order to exploit the subtle appropriate data linkages, even if the fundamental links are unclear or the physical interpretation is hard to describe [1]. ANN is able to categorize the various layers at varying depths, and in order to ascertain the thickness of each layer at a specified level, multi layer neural networks are trained independently. Data from the test boreholes was fed into the neural network, and the results were compared to real site investigation data to see how well the neural network performed in predicting changes in soil behavior. The results will indicate that ANN models are quite good at making predictions