Prediction of Liquefaction Behaviour of Fine-Grained Soil Using Nature-Inspired Optimized Algorithms Coupled with Neural Network

Infrastructure safety is substantially impacted by liquefaction phenomenon, a transformative process whereby fine-grained soils temporarily changes from a solid state to a liquid-like condition in response to cyclic forces, particularly in earthquake-prone regions. The present study examines the crucial geotechnical liquefaction aspects of fine-grained soil deposits and its effects on civil engineering and geotechnical structures. Using statistical performance indices, the current study assesses the prediction capacities of four machine learning models in determining the factor of safety against liquefaction. From the set of models examined, the artificial neural network coupled with the firefly algorithm (ANN-FA) is a definite standout, exhibiting excellent precision, reliability, and prediction accuracy. Moreover, the research goes a step further by incorporating real-life liquefaction data to validate the model’s reliability and accuracy, providing deeper insights into the proposed machine learning technique. The results underscore the importance of understanding liquefaction in fine-grained soils and demonstrates the practical utility of the ANN-FA model in geotechnical analysis and risk reduction within the realms of geotechnical engineering. By emphasizing the practical applicability of machine learning techniques for addressing geotechnical challenges, it provides a foundation for informed decision-making by engineers, designers, and stakeholders involved in infrastructure development and disaster risk mitigation.