Modeling non-linear displacement time series of geo-materials using evolutionary support vector machines

Evaluation of the non-linear deformation behavior of geo-materials is an important aspect of the safety assessment for geotechnical engineering in complex conditions. This paper presents a novel machine learning method, termed support vector machine (SVM), to obtain a global optimization model in conditions of large project dimensions, small sample sizes and nonlinearity. A new idea is put forward to combine the SVM with a genetic algorithm. The method has been used in the analysis of the high rock slope of the permanent shiplock of the Three Gorges Project and the horizontal deformation at depth in the Bachimen landslide in Fujian Province, China. The 92 non-linear SVMs in total were constructed with their kernel functions and the parameters were recognized using a genetic algorithm. The results indicate that the established SVMs can appropriately describe the evolutionary law of deformation of geo-materials at depth and provide predictions for the future 6-10 time steps with acceptable accuracy and confidence. (C) 2004 Elsevier Ltd. All rights reserved.