Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Surrogate Model-Based Prediction of Settlement in Foundation Over Cavity for Reliability Analysis
The stability of the foundations of a building may be seriously harmed by excavations (e.g., tunneling) or cavities (e.g., rock dissolution) underneath. It may be difficult and time-consuming to examine the bearing capacity and settlement of foundations over these cavities using traditional techniques, particularly while dealing with a variety of circumstances. Machine learning (ML) approaches provide good alternatives in such circumstances. Moreover, when these surrogate models are used in reliability analysis, computational efficiency, the capacity to handle huge numbers of samples, the ability to quantify uncertainty, the assistance for optimization and design exploration, and model calibration and validation are all can be made possible without compromising the computational accuracy. A shallow neural network has been used to construct a useful surrogate model that can be used in reliability analysis to assess the stability of foundations above cavities. Using PLAXIS 2D, a dataset was produced for this research with eight parametric changes to determine the output variable, viz., the settlement of a shallow foundation over a circular cavity. With a coefficient of determination (R2) of 0.956, a mean absolute error (MAE) of 9.854 and a mean squared error (MSE) of 28.24, the findings obtained show exceptional performance. These measurements confirm the precision and efficacy associated with the developed surrogate model in predicting foundation settlement over cavities.
Surrogate Model-Based Prediction of Settlement in Foundation Over Cavity for Reliability Analysis
The stability of the foundations of a building may be seriously harmed by excavations (e.g., tunneling) or cavities (e.g., rock dissolution) underneath. It may be difficult and time-consuming to examine the bearing capacity and settlement of foundations over these cavities using traditional techniques, particularly while dealing with a variety of circumstances. Machine learning (ML) approaches provide good alternatives in such circumstances. Moreover, when these surrogate models are used in reliability analysis, computational efficiency, the capacity to handle huge numbers of samples, the ability to quantify uncertainty, the assistance for optimization and design exploration, and model calibration and validation are all can be made possible without compromising the computational accuracy. A shallow neural network has been used to construct a useful surrogate model that can be used in reliability analysis to assess the stability of foundations above cavities. Using PLAXIS 2D, a dataset was produced for this research with eight parametric changes to determine the output variable, viz., the settlement of a shallow foundation over a circular cavity. With a coefficient of determination (R2) of 0.956, a mean absolute error (MAE) of 9.854 and a mean squared error (MSE) of 28.24, the findings obtained show exceptional performance. These measurements confirm the precision and efficacy associated with the developed surrogate model in predicting foundation settlement over cavities.
Surrogate Model-Based Prediction of Settlement in Foundation Over Cavity for Reliability Analysis
Transp. Infrastruct. Geotech.
Shubham, Kumar (Autor:in) / Metya, Subhadeep (Autor:in) / Sinha, Abdhesh Kumar (Autor:in)
Transportation Infrastructure Geotechnology ; 11 ; 1294-1320
01.06.2024
27 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Surrogate Model-Based Prediction of Settlement in Foundation Over Cavity for Reliability Analysis
Springer Verlag | 2024
|Reliability Analysis of Settlement of a Foundation Resting Over a Circular Void
Springer Verlag | 2022
|Settlement prediction model for pile foundation based on field observation
British Library Conference Proceedings | 2008
|Prediction of Foundation Load/Settlement Characteristics for Settlement '94
British Library Conference Proceedings | 1994
|