Physics-aware deep learning framework for linear elasticity

Abstract

The paper presents an efficient and robust data-driven deep learning (DL) computational framework developed for linear continuum elasticity problems. The methodology is based on the fundamentals of the Physics Informed Neural Networks (PINNs). For an accurate representation of the field variables, a multi-objective loss function is proposed. It consists of terms corresponding to the residual of the governing partial differential equations (PDE), constitutive relations derived from the governing physics, various boundary conditions, and data-driven physical knowledge fitting terms across randomly selected collocation points in the problem domain. To this end, multiple densely connected independent artificial neural networks (ANNs), each approximating a field variable, are trained to obtain accurate solutions. Several benchmark problems including the Airy solution to elasticity and the Kirchhoff-Love plate problem are solved. Performance in terms of accuracy and robustness illustrates the superiority of the

A Computer Vision Enabled damage detection model with improved YOLOv5 based on Transformer Prediction Head

“…More recently, deep learning (DL) characterized by multilayer neural networks (NN) (LeCun et al, 2015) has shown remarkable breakthroughs in pattern recognition for various fields including image classification (Rawat and Wang, 2017;Khan et al, 2022b,a), computer vision (Voulodimos et al, 2018;Roy and Bhaduri, 2021;Roy et al, 2022c;Roy and Bhaduri, 2022;Roy et al, 2022a), object detection (Zhao et al, 2019a;Chandio et al, 2022;Roy et al, 2022b;Singh et al, 2023a), brain-computer interfaces (Roy, 2022b,a,c;Singh et al, 2023b), signal classification Roy, 2023, 2022) and across diverse scientific disciplines (Bose and Roy, 2022;Roy and Bose, 2023b;Roy and Guha, 2022;Roy and Bose, 2023a;Roy and Guha, 2023). Following the success, there is an increasing thrust of research works geared towards damage classification tasks employing DL techniques, mostly convolutional neural networks (CNN), such as ResNet (Bang et al, 2018), AlexNet (Dorafshan et al, 2018;Li et al, 2018), VGG-net (Gopalakrishnan et al, 2017;Silva and Lucena, 2018) and various others (Chow et al, 2020;Nath et al, 2022;.…”