Prediction model of lateral spreading of liquefied soil during earthquakes based on neural network

In response to the complex issue of predicting lateral spreading of liquefied soil during earthquakes, this study establishes a comprehensive seismic database. The Newmark sliding block method is utilized to calculate lateral displacements, and a sensitivity analysis model for liquefaction-induced lateral spreading is developed by considering multiple parameters. Based on the three most influential parameters identified through sensitivity analysis, a novel liquefaction-induced lateral spreading prediction model is proposed. The reliability of this new model is then demonstrated by comparing it with existing classical prediction models, as well as by analyzing correlation coefficients and root mean square errors.