Liquefaction behavior of Indo-Gangetic region using novel metaheuristic optimization algorithms coupled with artificial neural network

The present research aims to co-relate the plasticity and liquefaction response of soil as well as its significance in defining liquefaction probability. To accomplish this, metaheuristic hybrid models (ANN-PSO, ANN-GA, ANN-GWO, ANN-CA, ANN-FA, and ANN-GBO) were used to forecast the liquefaction probability (P-L) of alluvium soil deposits belonging to the Indo-Gangetic plain (Bihar region, India). The first-time application of ANN-based techniques hybridized with the Gradient-Based Optimizer (GBO) algorithm for predicting the P-L of fine-grained soils brings a certain degree of novelty. The main advantage of hybrid computational models is that they can subjectively analyze an unlimited amount of data and give reliable outcomes and assessments. The use of the plasticity index (PI) to measure the liquefaction behavior of fine-grained soil has a considerable impact in defining the liquefaction susceptibility for soil with moderate to high plasticity, and it seeks to make a significant contribution to the liquefaction studies. The ANN-GBO model has the best prediction ability, according to performance metrics. The overall analysis suggests that the application PI along with the proposed ANN-GBO model can be thought of as a novel tool to help geotechnical engineers estimate the occurrence of liquefaction during the early design stage of any engineering project. It was observed that fine-grained soil with PI > 14%, shows a lower P-L of about less than 35% and falls under safe to moderately safe zones. Soil with 9 < PI < 14 exhibited a P-L of about 65 to 35%, and soil with PI less than 8 exhibited a higher P-L of about 65% or more.