Estimating Orel-consolidation Ratio and Lateral Stress Coefficient Using Neural Networks

摘要

The overconsolidation ratio (OCR) and coefficient of earth pressure at rest (K_o) of clays are important factors influencing the strength, stress-strain behavior, and the compressibility characteristics of cohesive soils. This paper evaluates the feasibility of using an artificial neural network (ANN) model for estimating the OCR and K_o of soil deposits from piezocone penetration test (PCPT) data. An ANN model was developed using the general regression neural network (GRNN) algorithm, and trained using actual PCPT records from test sites around the world, together with the corresponding oedometer OCR test results and empirically determined values of K_o. After training, the GRNN model was tested using previously unseen PCPT data, and model predictions were compared to reference oedometer OCR values and empirically determined K_o values to evaluate the network's success. Model predictions were also compared with some of the existing interpretation methods.