Application of the optimal regression-based analysis to estimate the deformation of geogrid-reinforced soil structures

The analysis of deformation plays a crucial role in the design of Geosynthetic reinforced structure (GRS) constructions. However, the publications emphasize the potential of artificial prediction algorithms in addressing geotechnical engineering challenges. The main aim of the present study is to examine the possible employ of estimation methods in the prediction of the deformation (Dis) of GRS. The present paper shows and approve a new technique which integrates the Dwarf Mongoose Optimizer (DMO) framework with the Multi-layered perceptron (MLP) neural network and Support Vector Regression (SVR) (abbreviated as SVR_DMO and MLP_DMO). Afterwards, a whole of 166 finite element values performed in the publications were utilized so as to make the data collection. Based on the results obtained, it can be concluded that both SVR_DMO and MLP_DMO have considerable potential in properly forecasting the Dis. The R² values for SVR_DMO were 0.9835 throughout training and 0.9866 throughout testing. After careful examination of several types of performance tests and their comparison to valid publications, it has been determined which the SVR_DMO offers a more suitable framework for calculating the Dis of GRS.