Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Prediction of OCR and su from PCPT Data Using Tree-Based Data Fusion Techniques
AbstractThis study examines the feasibility of using data fusion techniques to predict the overconsolidation ratio (OCR) and undrained shear strength (su) from piezocone penetration test (PCPT) measurements, in situ stresses, and additional features created from available data. Several data fusion models were developed using two feature-level fusion techniques: regression trees and model trees. Hybrid models were developed using these feature-level fusion techniques and two decision-level fusion techniques, namely, bootstrap aggregation and stacked generalization. After training and testing the data fusion models, the predicted values of OCR and su were compared to the reference values, obtained from laboratory or in situ tests, and to the values estimated using several existing interpretation methods. The model trees, which predict continuous values, performed better than the regression trees, which predict discrete values. The decision-level fusion algorithms tended to improve the results of the regression trees, but had little effect on the results of the model trees. Overall, the data fusion models were found to perform well and tended to perform better than the corresponding interpretation methods in estimating values of OCR and su.
Prediction of OCR and su from PCPT Data Using Tree-Based Data Fusion Techniques
AbstractThis study examines the feasibility of using data fusion techniques to predict the overconsolidation ratio (OCR) and undrained shear strength (su) from piezocone penetration test (PCPT) measurements, in situ stresses, and additional features created from available data. Several data fusion models were developed using two feature-level fusion techniques: regression trees and model trees. Hybrid models were developed using these feature-level fusion techniques and two decision-level fusion techniques, namely, bootstrap aggregation and stacked generalization. After training and testing the data fusion models, the predicted values of OCR and su were compared to the reference values, obtained from laboratory or in situ tests, and to the values estimated using several existing interpretation methods. The model trees, which predict continuous values, performed better than the regression trees, which predict discrete values. The decision-level fusion algorithms tended to improve the results of the regression trees, but had little effect on the results of the model trees. Overall, the data fusion models were found to perform well and tended to perform better than the corresponding interpretation methods in estimating values of OCR and su.
Prediction of OCR and su from PCPT Data Using Tree-Based Data Fusion Techniques
Griffin, Erin P (Autor:in) / Kurup, Pradeep U
2017
Aufsatz (Zeitschrift)
Englisch
BKL:
56.20
Ingenieurgeologie, Bodenmechanik
Prediction of OCR and s~u from PCPT Data Using Tree-Based Data Fusion Techniques
| British Library Online Contents | 2017
Neural Networks for Profiling Stress History of Clays from PCPT Data
| Online Contents | 2002