A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
A data-driven method to model stress-strain behaviour of frozen soil considering uncertainty
Abstract Various experiments and computational methods have been conducted to describe the mechanical behaviours of frozen soils. However, due to high nonlinearity and uncertainty of responses, modelling the stress-strain behaviours of frozen soils remains challenging. Accordingly, we first propose a novel data-driven method based on Long Short-Term Memory (LSTM) to model the mechanical responses of frozen soil. A compiled database on the stress-strain of a frozen silty sandy soil is employed to feed into the LSTM model, where the mechanical behaviours under various temperatures and confining pressures are measured through triaxial tests. Subsequently, uncertainty of the stress-strain relations (i.e., deviatoric stress and volumetric strain to axial strain) is investigated and considered in LSTM-based modelling with Monte Carlo dropout (LSTM-MCD). Results demonstrate that the LSTM model without uncertainty can capture the stress-strain responses of the frozen soil with considerable predictive accuracy. Uncertainty analysis from LSTM-MCD reveals that the model with uncertainty can be applied to evaluate the mechanical responses of frozen soil with 95% confidence intervals. This study sheds light on the advantage of the data-driven model with uncertainty in predicting mechanical behaviours of frozen soils and provides references for permafrost construction.
Highlights A data-driven method is proposed to model the mechanical behaviour of frozen soil. LSTM with dropout is developed for considering the uncertainty in modelling. Confidence intervals are provided for reliability evaluation in practice.
A data-driven method to model stress-strain behaviour of frozen soil considering uncertainty
Abstract Various experiments and computational methods have been conducted to describe the mechanical behaviours of frozen soils. However, due to high nonlinearity and uncertainty of responses, modelling the stress-strain behaviours of frozen soils remains challenging. Accordingly, we first propose a novel data-driven method based on Long Short-Term Memory (LSTM) to model the mechanical responses of frozen soil. A compiled database on the stress-strain of a frozen silty sandy soil is employed to feed into the LSTM model, where the mechanical behaviours under various temperatures and confining pressures are measured through triaxial tests. Subsequently, uncertainty of the stress-strain relations (i.e., deviatoric stress and volumetric strain to axial strain) is investigated and considered in LSTM-based modelling with Monte Carlo dropout (LSTM-MCD). Results demonstrate that the LSTM model without uncertainty can capture the stress-strain responses of the frozen soil with considerable predictive accuracy. Uncertainty analysis from LSTM-MCD reveals that the model with uncertainty can be applied to evaluate the mechanical responses of frozen soil with 95% confidence intervals. This study sheds light on the advantage of the data-driven model with uncertainty in predicting mechanical behaviours of frozen soils and provides references for permafrost construction.
Highlights A data-driven method is proposed to model the mechanical behaviour of frozen soil. LSTM with dropout is developed for considering the uncertainty in modelling. Confidence intervals are provided for reliability evaluation in practice.
A data-driven method to model stress-strain behaviour of frozen soil considering uncertainty
Li, Kai-Qi (author) / Yin, Zhen-Yu (author) / Zhang, Ning (author) / Liu, Yong (author)
2023-05-25
Article (Journal)
Electronic Resource
English
Deep learning , Frozen soil , Constitutive modelling , Uncertainty , Dropout , Monte Carlo , ANN , Artificial neural network , <italic>b</italic> , Bias , BNN , Bayesian neural network , <italic>c</italic> <inf><italic>t</italic></inf> , Cell state , <math xmlns="http://www.w3.org/1998/Math/MathML"><msub><mover><mi>c</mi> <mo>˜</mo></mover> <mi>t</mi></msub></math> , Candidate memory cell state , <italic>f</italic> , Forget gate , <italic>i</italic> , Input gate , LSTM , Long short-term memory , m , Monte Carlo times , MAE , Mean square error , <italic>n</italic> , Number of data points , <italic>o</italic> , Output gate , <italic>q</italic> , Deviatoric stress , R<sup>2</sup> , Coefficient of determination , RMSE , Root mean squared error , RNN , Recurrent neural network , <italic>T</italic> , Temperature , <italic>W</italic> , Weight , <italic>x</italic> <inf>max</inf> , Maximum input , <italic>x</italic> <inf>min</inf> , Minimum input , <italic>x</italic> <inf>ni</inf> , Is normalised <italic>i</italic>-th input , <italic>y</italic> <inf>e</inf> , Experimental datum , <italic>y</italic> <inf>m</inf> , Mean value of experimental measurements , <italic>y</italic> <inf>p</inf> , Predicted datum , <italic>ε</italic> , Strain , <italic>ε</italic> <inf>a</inf> , Axial strain , <italic>ε</italic> <inf>v</inf> , Volumetric strain , <italic>σ</italic> , Stress , <italic>σ</italic> <inf>3</inf> , Confining pressure
Effect of consolidating process on stress-strain behaviour of artificially frozen soil
| British Library Conference Proceedings | 2000
Technical Improvements in Testing Small-Strain Deformation Behaviour of Frozen Soil
| Springer Verlag | 2020
Dynamic stress–strain behavior of frozen soil: Experiments and modeling
| Online Contents | 2014
Dynamic stress–strain behavior of frozen soil: Experiments and modeling
| Elsevier | 2014