Study of some applications of the technique of artificial neural networks (ANNs) in geotechnical engineering

Resumo

The application of Artificial Neural Networks (ANNs) in geotechnical engineering has grown significantly since the early 1990s. ANNs have been used to address various geotechnical engineering problems (e.g., geotechnical classification, geotechnical properties prediction, site characterization, earth retaining structures, settlement of structures, foundation and slope stability) and have shown a certain degree of success. This article presents an overview of some ANN case applications for solving specific geotechnical engineering problems and discusses the use of ANNs for the indirect determination of geotechnical properties based on known index properties. The use of ANNs is evaluated for soils from different regions to predict one-dimensional compressibility of soft soils and permeability of saturated soils. The article also explores the use of ANNs in predicting the consistency limits of fine-grained soils and the calibration required in the Time Domain Reflectometry (TDR) technique used to measure soil moisture. The studies reviewed in this article indicate that ANNs have potential as an alternative to empirical correlations for predicting soil properties based on their index properties, for soils from various regions of Brazil and other countries.