Investigation of Stress Distribution in a Railway Embankment Reinforced By Geogrid Based Weak Soil Formation Using Hybrid Rnn-Eho

Abstract

As the primary method of track support, traditional sloping embankments are typically used by railroad lines. Geosynthetically Reinforced Soil (GRS) systems, as an alternative to traditional embankments, have gained appeal, notably for high-speed lines in India. This system's reduced base area compared to traditional embankments means that less ground stabilization, improvement, and land taking is necessary. The research's findings provide intriguing strategies that may be implemented into the way tracks are designed now to accommodate faster freight trains pulling greater loads. This research explains how to anticipate the bearing capacity of weak sand supported by a method of compacted granular fill over natural clay soil using a hybrid Recurrent Neural Network (RNN) and Elephant Herding Optimization (EHO) with Georgic reinforced soil foundation. The exact prediction target for the proposed model was developed by using displacement amplitude as an output index. A number of elements influencing the foundation bed's properties, Georgic reinforcement, and dynamic excitation have been taken into account as input variables. The RNN-anticipated EHO's accuracy was compared to that of three other popular approaches, including ANN, HHO, CFA, and MOA. Strict statistical criteria and a multi-criteria approach were principally used to assess the predictive power of the developed models. The model is also examined using fresh, independent data that wasn't part of the initial dataset. The hybrid RNNEHO model performed better in predicting the displacement amplitude of footing lying on Geogrid-reinforced beds than the other benchmark models. Last but not least, the sensitivity analysis was used to highlight how input parameters might affect the estimate of displacement amplitude.