Artificial Neural Network (ANN) Model for Shear Strength of Soil Prediction
- 8 September 2021
- journal article
- Published by Trans Tech Publications, Ltd. in Defect and Diffusion Forum
- Vol. 411, 157-168
- https://doi.org/10.4028/www.scientific.net/ddf.411.157
Abstract
Geotechnical structures, design of embankment, earth and rock fill dam, tunnels, and slope stability require further attention in determining the shear strength of soil and other parameters that govern the result. The shear strength of soil commonly obtained by conducting laboratory testing such as Unconfined Compression Strength (UCS) Test and Unconsolidated Undrained (UU) Test. However, random errors and systematic errors can occur during experimental works and caused the findings imprecise. Besides, the laboratory test also consuming a lot of time and some of them are quite costly. Therefore, soft computational tools are developed to improve the accuracy of the results and time effectively when compared to conventional method. In this study, Artificial Neural Network (ANN) was employed to develop a predictive model to correlate the moisture content (MC), liquid limit (LL), plastic limit (PL), and liquidity index (LI) of cohesive soil with the undrained shear strength of soil. A total of 10 databases was developed by using MATLAB 7.0 - matrix laboratory with 318 of UCS tests and 451 of UU tests which are collected from the verified site investigation (SI) report, respectively. All the SI reports collected were conducted in Sarawak, Malaysia. The datasets were split into ratio of 3:1:1 which is 60:20:20 (training: validation: testing) with one hidden layer and eight hidden neurons. The input parameter of Liquidity index (LI) has shown the highest R-value (regression coefficient) which are 0.926 and 0.904 for UCS and UU model, respectively. In addition, the predictive models were tested and compare with the predicted and observed cohesion obtained from the collected experimental results. In summary, the ANN has the feasibility to be used as a predictive tool in estimating the shear strength of the soil.Keywords
This publication has 19 references indexed in Scilit:
- Soil shear strength prediction using intelligent systems: artificial neural networks and an adaptive neuro-fuzzy inference systemSoil Science and Plant Nutrition, 2012
- Platy layer silicate minerals for controlling residual strength in landslide soils of different origins and geologyGeology, 2010
- Relationship Between the Atterberg Limits and Clay ContentSoils and Foundations, 2007
- Interpretation of direct shear box testing of sands as quasi-simple shearGéotechnique, 1997
- Fast shearing of pre-existing shear zones in soilGéotechnique, 1996
- Artificial neural networks: a tutorialComputer, 1996
- Residual strength of kaolin and bentonite: the influence of their constituent pore fluidGéotechnique, 1994
- Drained Residual Strength of Cohesive SoilsJournal of Geotechnical Engineering, 1994
- Notes on the Design of the Liquid Limit DeviceGéotechnique, 1958
- The Sensitivity of ClaysGéotechnique, 1952