Sequential Encoding of Soil Profiles: A Large Language Model-Inspired Framework for Liquefaction Assessment

Abstract

This study presents a novel Self-attention-based deep-learning model for predicting soil liquefaction potential. The proposed architecture processes three distinct data streams: spectral seismic encoding, soil stratigraphy tokenization, and site-specific features. The architecture processes data from 165 case histories across 11 major earthquakes, employing Fast Fourier Transform for seismic waveform encoding and transformer architectures for soil layer tokenization. The model achieves 93.75% prediction accuracy on cross-regional validation sets and demonstrates robust performance through sensitivity analysis of ground motion intensity and soil resistance parameters. Notably, validation against previously unseen ground motion data from the 2024 Noto Peninsula earthquake confirms the model's generalization capabilities and practical utility. This approach establishes a new framework in geotechnical deep learning where sophisticated multi-modal analysis meets practical engineering requirements through quantitative interpretation.