Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Progressive Machine Learning Approaches for Predicting the Soil Compaction Parameters
https://orcid.org/http://orcid.org/0000-0003-4986-9885
Estimating soil compaction parameters is an essential point when seeking safe and economic dams, bridges abutments, highways embankments, retaining walls, and nuclear waste disposal; however, the Proctor Test employed in this forecasting is costly and time-consuming. The current study, therefore, aims at elaborating a new alternative model for predicting the compaction parameters based on eleven new progressive machine learning methods to overcome these limitations. The modeling phase was performed using a database of 147 samples collected from different studies. Additionally, six relevant factors were selected in the input layer based on the literature recommendations. Furthermore, the relevant inputs have been modeled using machine learning methods; their performance was assessed through six performance measures and the K-fold cross-validation approach. The comparative study proved the effectiveness of the RF model, which displayed the highest performance in predicting soil compaction parameters. This elaborated model provided the optimal prediction, i.e., the closest to the experimental values compared to other models and formulae proposed in the literature. Finally, a reliable and easy-to-use graphical interface was generated in the current study dubbed “ComPara2021.” This latter will be very helpful for researchers and civil engineers when estimating compaction parameters with the advantage of saving time and money.
Progressive Machine Learning Approaches for Predicting the Soil Compaction Parameters
https://orcid.org/http://orcid.org/0000-0003-4986-9885
Estimating soil compaction parameters is an essential point when seeking safe and economic dams, bridges abutments, highways embankments, retaining walls, and nuclear waste disposal; however, the Proctor Test employed in this forecasting is costly and time-consuming. The current study, therefore, aims at elaborating a new alternative model for predicting the compaction parameters based on eleven new progressive machine learning methods to overcome these limitations. The modeling phase was performed using a database of 147 samples collected from different studies. Additionally, six relevant factors were selected in the input layer based on the literature recommendations. Furthermore, the relevant inputs have been modeled using machine learning methods; their performance was assessed through six performance measures and the K-fold cross-validation approach. The comparative study proved the effectiveness of the RF model, which displayed the highest performance in predicting soil compaction parameters. This elaborated model provided the optimal prediction, i.e., the closest to the experimental values compared to other models and formulae proposed in the literature. Finally, a reliable and easy-to-use graphical interface was generated in the current study dubbed “ComPara2021.” This latter will be very helpful for researchers and civil engineers when estimating compaction parameters with the advantage of saving time and money.
Progressive Machine Learning Approaches for Predicting the Soil Compaction Parameters
https://orcid.org/http://orcid.org/0000-0003-4986-9885
Estimating soil compaction parameters is an essential point when seeking safe and economic dams, bridges abutments, highways embankments, retaining walls, and nuclear waste disposal; however, the Proctor Test employed in this forecasting is costly and time-consuming. The current study, therefore, aims at elaborating a new alternative model for predicting the compaction parameters based on eleven new progressive machine learning methods to overcome these limitations. The modeling phase was performed using a database of 147 samples collected from different studies. Additionally, six relevant factors were selected in the input layer based on the literature recommendations. Furthermore, the relevant inputs have been modeled using machine learning methods; their performance was assessed through six performance measures and the K-fold cross-validation approach. The comparative study proved the effectiveness of the RF model, which displayed the highest performance in predicting soil compaction parameters. This elaborated model provided the optimal prediction, i.e., the closest to the experimental values compared to other models and formulae proposed in the literature. Finally, a reliable and easy-to-use graphical interface was generated in the current study dubbed “ComPara2021.” This latter will be very helpful for researchers and civil engineers when estimating compaction parameters with the advantage of saving time and money.
Progressive Machine Learning Approaches for Predicting the Soil Compaction Parameters
Transp. Infrastruct. Geotech.
Benbouras, Mohammed Amin (Autor:in) / Lefilef, Lina (Autor:in)
Transportation Infrastructure Geotechnology ; 10 ; 211-238
01.04.2023
28 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Predicting Compaction Parameters of Silty Soil by Nonlinear Multivariable Approach
| Springer Verlag | 2022
Predicting Complete Compaction Curves of Fine-Grained Soil
| British Library Conference Proceedings | 1998