Prediction of Soil-Water Characteristic Curve using optimised machine learning approaches
Prediction of Soil-Water Characteristic Curve using optimised machine learning approaches
Soil-water characteristic curves (SWCC), also known as soil-water retention curves, are used to describe the mechanical behaviour of unsaturated soils. In many geotechnical problems, the relationship between the water content of the soil and its matric suction is obtained by SWCC. In this study, alternative machine learning methods including multilayer perceptron artificial neural network, extreme gradient boosting and random forest are used to predict the SWCC of unsaturated soil through influential variables. In addition, particle swarm optimisation algorithm is employed to optimate the hyper-parameters of ML algorithms. A dataset currently available in literature is employed in this study and the R-squared value, root mean square error, and variance accounted for are used as performance measures to compare the performance of the optimised machine learning methods. The results of the optimised methods are then compared with the results of linear regression. The results reveal that the optimised machine learning methods are potentially excellent candidates for predicting the soil-water characteristic curves and the particle swarm optimisation is a powerful tool in the hyper-parameter tuning.
M. Nazem; Navid Kardani; Sara Moridpour; Annan Zhou
10th European Conference on Numerical Methods in Geotechnical Engineering (NUMGE2023)
6. Machine learning and artificial intelligence
https://doi.org/10.53243/NUMGE2023-312
Contact Information
- Secretary General - Dr A M McNamara
- Email: [email protected]
- Phone: +44 20 7040 8154
- City, University of London
- Northampton Square
- London EC1V 0HB. United Kingdom