Prediction of Compression Index of Soft Soils from the Brazilian Coast Using Artificial Neural Networks and Empirical Correlations
A.G. Oliveira Filho, L.B. Totola, K.V. Bicalho, W.H. Hisatugu
Soils and Rocks, São Paulo, 43(1): 109-121, January-March, 2020 | PDF
This paper aims to explore the potential use of artificial neural networks (ANNs) to predict the compression index (CC) of soft soils from the Brazilian coast. Results from 225 standard consolidation (oedometer) tests and the corresponding soil index properties (i.e., initial void ratio, natural water content and Atterberg limits) of a wide variety of fine-grained soils reported in the literature were compiled and investigated herein. The ANN prediction performance is compared with linear empirical correlations created from the database investigated. In addition, correlations presented in the literature are also used and evaluated through different statistical techniques. Overall, for the organized dataset, the ANN outperformed the empirical correlations, highlighting the fragility and limitations of single and multiple variable linear empirical correlations.
Submitted on June 1, 2019; Final Acceptance on Januray 29, 2020; Discussion open until August 31, 2020.