Evaluation of lateral spreading using artificial neural networks

Liquefaction-induced lateral spreading has been a very damaging type of ground failure during past strong earthquakes. Although the occurrence of liquefaction and lateral spreading at a given site can be predicted, the methods to estimate the magnitude of resulting deformations is still the focus of many researches. In this study, using professional software called STATISTICA, a neural network model is developed to predict the horizontal ground displacement in both ground slope and free face conditions due to liquefaction-induced lateral spreading. The database, implemented in this work, is the one compiled by Youd and his colleagues in their revised MLR model. The influence of seismological, topographical and geotechnical parameters on resulting deformations and their degrees of importance are investigated. The results indicate that the model presented in this research serves as a reliable tool to predict horizontal ground displacement. The correlation factors and the root mean square errors obtained in this model show the superiority of the Neural Network approach over the traditional regression analysis.