Analyzing the impact of geosynthetic reinforcement on Sinkhole: A numerical investigation with Machine Learning approach

Abstract

The present study describes the results of numerical investigations with implementation of Machine Learning approach to prevent the event of sinkhole formation into the ground by using geosynthetic reinforcement. A series of 2D finite-element analysis were carried out with emphasis on the effects of geosynthetics reinforcement parameters such as geogrid length, stiffness and spacing between the geogrid layers to evaluate the behavior of ground over the sinkhole. Moreover, a database was also generated from the simulations for the application of Machine Learning techniques to predict the behavior as well. The hyperparameters of the Machine Learning models were optimized using Bayesian optimization to enhance the prediction accuracy. The critical values of geogrid reinforcement parameters for maximum reinforcing effect are also suggested. The results indicate the effectiveness of geosynthetic reinforcement to significantly reduce the sinkhole deformations and also the potential of Machine Learning to predict ground subsidence.