Displacement back analysis for a steep slope at the Three Gorges Project site

This paper contributes to two aspects of displacement back analysis of rock slopes. First, a novel method for displacement back analysis is presented, based on error back-propagation neural network and genetic algorithm (GA). The BP network replaces the time-consuming finite element method, thus enhancing the efficiency of the analysis. The GA is used as an optimization method, whose global search strategy can effectively improve the reliability of the analysis, thereby, making the back-analyzed results independent of the initial values. Application of this methodology is illustrated with a numerical example. Secondly, the deformation mechanism of rock mass is duly taken into account in the displacement back analysis of Profile 17-17 of the permanent shiplock slope at the Three Gorges Project site, by treating the disturbed zones as weakened media, yielding reasonable results. (C) 2001 Elsevier Science Ltd. All rights reserved.