The Use of a Relevance Vector Machine in Predicting Liquefaction Potential

Abstract

The prediction of liquefaction susceptibility of soil due to an earthquake is an imperative task in Civil Engineering. This study uses relevance vector machine (RVM) for the prediction of liquefaction susceptibility of soil based on cone penetration test from Chi-Chi earthquake, Taiwan. RVM is based on a Bayesian formulation of a linear model with an appropriate prior that results in a sparse representation. Here, RVM has been used as a classification tool. It gives output in a probabilistic form. Equations have been also developed for the determination of liquefaction susceptibility of soil based on the RVM model. This study shows that only two input parameters [cone resistance (qc) and maximum horizontal acceleration (amax) are sufficient for prediction of liquefaction susceptibility of soil. The results of RVM have been compared with the conventional methods. The developed RVM model provides a viable tool for civil engineers to determine the liquefaction susceptibility of soil.