Search results

This paper aims to investigate the approach combine the deep learning (DL) and finite element method for the magneto-thermal coupled problem.

To achieve the DL of electrical device with the hypothesis of a small dataset, with ground truth data obtained from the FEM analysis, U-net, a highly efficient convolutional neural network (CNN) is used to extract hidden features and trained in a supervised manner to predict the magneto-thermal coupled analysis results for different topologies. Using part of the FEM results as training samples, the DL model obtained from effective off-line training can be used to predict the distribution of the magnetic field and temperature field of other cases.

The possibility and feasibility of the proposed approach are investigated by discussing the influence of various network parameters, in particular, the four most important factors are training sample size, learning rate, batch size and optimization algorithm respectively. It is shown that DL based on U-net can be used as an efficiency tool in multi-physics analysis and achieve good performance with only small datasets.

It is shown that DL based on U-net can be used as an efficiency tool in multi-physics analysis and achieve good performance with only small datasets.

The study aims to investigate the effects of pre-loading histories (pre-shearing and pre-consolidation) on the liquefaction behaviour of saturated loose sand via discrete element method (DEM) simulations.

The pre-shearing history is mimicked under drained conditions (triaxial compression) with different pre-shearing strain levels ranging from 0% to 2%. The pre-consolidation history is mimicked by increasing the isotropic compression to different levels ranging from 100 kPa to 300 kPa. The macroscopic and microscopic behaviours are analysed and compared.

Temporary liquefaction, or quasi-steady state (QSS), is observed in most samples. A higher pre-shearing or pre-consolidation level can provide higher liquefaction resistance. The ultimate state line is found to be unique and independent of the pre-loading histories in stress space. The Lade instability line prematurely predicts the onset of liquefaction for all samples, both with and without pre-loading histories. The redundancy index is an effective microscopic indicator to monitor liquefaction, and the onset of the liquefaction corresponds to the phase transition state where the value of redundancy index is one, which is true for all cases irrespective of the proportions of sliding contacts.

The liquefaction behaviour of granular materials still remains elusive, especially concerning the effects of pre-loading histories on soils. Furthermore, the investigation of the effects of pre-consolidation histories on undrained behaviour and its comparison to pre-sheared samples is rarely reported in the DEM literature.