Evaluation of liquefaction-induced lateral spreading using a new evolutionary method

Liquefaction-induced lateral spreading is defined as the finite, lateral displacement of gently sloping ground due to increasing of pore pressure in liquefaction phenomenon in shallow underlying layers of saturated sandy soils during earthquake. In this paper, we employed a new evolutionary approach, named as polynomial neural network (PNN), optimized by genetic algorithm, in order to estimate the liquefaction-induced lateral movements. The database compiled by Youd et al., containing 464 cases from 10 earthquakes in two seismo-genic zones of Japan and United States, is updated and used for modeling the new relationship. Comparison between the new model and previous developed models, also, has been done after modeling and the results showed a reasonable predictive capability of the new PNN developed model over existing models.

Keywords:

lateral spreading , polynomial neural networks , database , genetic algorithm