APPLICATION OF NEURAL NETWORKS IN SEISMIC SOIL SLOPE STABILITY ANALYSIS

Abstract

The advent of digital computers has seen the emergence of analytical tools in analysis and design of Civil Engineering systems, which earlier seemed too complex or rigorous. At this stage Artificial Intelligence techniques - especially Artificial Neural Nets, or Neural Nets, are beginning to dominate most of the analytical and design aspects. The basic advantage in the use of Neural Nets being the capability of handling imprecise, imperfect and incomplete data while yet producing acceptable solutions. In the present study the application of Neural Nets has been examined with specific reference to the analysis of slope of a typical Earth Dam for the following cases: 1. Determination of minimum factor of safety when conventional inputs are provided for a specific problem. 2. The inverse problem of determination of the critical failure surface when the minimum factor of safety and conventional parameters are provided as inputs. 3. Determination of minimum factor of safety and critical failure surface when conventional parameters are provided as inputs. The Neural Nets were formed to perform satisfactorily and a subsequent parametric study with regard to a. Relation between horizontal seismic coefficient and minimum factor of safety. b. Relation between pore water pressure and minimum factor of safety. c. Relation between angle of internal friction of the soil in shell portion of earth dam section and minimum factor of safety. d. Relation between angle of internal friction of the soil in core portion of earth dam section and minimum factor of safety. . indicates that the results are consistent with the underlying Physics of the problem. The potential of application of Neural Nets to augment/replace analytical procedures is amply demonstrated and scope for further studies has been indicated.