The prediction of soil cracking caused by desiccation using artificial intelligence and statistical analysis
The prediction of soil cracking caused by desiccation using artificial intelligence and statistical analysis
Desiccation cracking affects soil hydromechanical behaviour and can cause serious hazards. Several factors affect desiccation cracking. Multiple and non-linear factors make desiccation cracking difficult to predict. Artificial intelligence (AI) has proven effective for predicting parameters based on different numbers of variables, but rarely for predicting soil cracking. Based on drying tests and generated databases, this study investigated if AI methods could predict two-dimensional soil desiccation cracking. Multiple linear regression (MLR) and support vector machines (SVM) were used to predict two outputs, including vertical shrinkage and cracks and shrinkage intensity factor (CSIF). Based on four input parameters, soil thickness, liquid limit, plasticity index, and shrinkage limit, mathematical models were built. In MLR, coefficient of determination (R2) and mean absolute error (MAE) were 0.803 and 2.396 for predicting CSIF, respectively, and 0.768 and 2.608 for predicting vertical shrinkage. CSIF was predicted with R2 of 0.963 and MAE of 1.225, and vertical shrinkage with R2 of 0.970 and MAE of 0.960 by SVM. The soil thickness was the most significant input parameter in the sensitivity analysis.
Abolfazl Baghbani; M. Yaghoubi; Firas Daghistani; Susanga M. Costa; T. Baumgartl
10th European Conference on Numerical Methods in Geotechnical Engineering (NUMGE2023)
6. Machine learning and artificial intelligence
https://doi.org/10.53243/NUMGE2023-92
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