Prediction of the Seismic Effect on Liquefaction Behavior of Fine-Grained Soils Using Artificial Intelligence-Based Hybridized Modeling

• Published in: Arabian journal for science and engineering (2011) Vol. 47; no. 4; pp. 5411 - 5441
• Main Authors: Ghani, Sufyan, Kumari, Sunita, Ahmad, Shamsad
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.04.2022; Springer Nature B.V
• Subjects: Adaptive algorithms; Artificial intelligence; Earthquakes; Engineering; Evaluation; Fine-grained soils; Genetic algorithms; Heuristic methods; Humanities and Social Sciences; Liquefaction; Machine learning; Model accuracy; multidisciplinary; Optimization algorithms; Particle swarm optimization; Plasticity index; Reliability analysis; Research Article-Civil Engineering; Safety factors; Science; Seismic effects; Seismic response; Swarm intelligence; Uncertainty; Monotonicity analysis; Liquefaction; Firefly algorithm; Machine learning; Hybrid modeling; Seismic effect; GA; ANFIS; Fine-grained soils; PSO
• ISSN: 2193-567X; 1319-8025; 2191-4281
• DOI: 10.1007/s13369-022-06697-6

Researchers in the past have reported significant uncertainties involved in evaluating the risk of soil liquefaction using deterministic approaches. Therefore, to improve the accuracy and remove the uncertainties involved the present research aims to explore the possibility of using artificial intelligence (AI) to study earthquake-induced soil liquefaction. Also, this study highlights the relative significance of the plasticity index of fine-grained soil for assessing the risk against liquefaction. The possibility of the application of a hybrid method, comprising optimization algorithms and adaptive neuro-based fuzzy inference system (ANFIS) for assessing the safety factor ( F S ) against earthquake-induced liquefaction, is explored. Three metaheuristic optimization algorithms, namely the firefly algorithm (FF), genetic algorithm, and particle swarm optimization, were each hybridized with the ANFIS technique to create three alternative hybrid models for evaluating seismic response. The ANFIS-FF hybrid model is found as an effective and prominent AI-based approach with R 2  = 0.976, RMSE = 0.079 in the training phase and R 2  = 0.982, RMSE = 0.069 in the testing phase for predicting the liquefaction behavior of soil with minimal uncertainty and human involvement, therefore contributing to an enormous accomplishment in terms of resources and sustainability. Monotonicity analysis and real-life liquefaction data were adopted to validate the reliability and accuracy of the model to provide a better insight into the proposed machine learning technique. The finding of the present research would substantially contribute to the field of liquefaction studies for fine-grained soil with medium to high plasticity.