Related references
Note: Only part of the references are listed.
Article
Computer Science, Interdisciplinary Applications
Ravi G. Patel et al.
Summary: Using a least squares space-time control volume scheme can more naturally handle regularity requirements, imposition of boundary conditions, entropy compatibility, and conservation, reducing requisite hyperparameters. This framework introduces inductive biases to discover thermodynamically consistent equations of state guaranteeing hyperbolicity, providing a method for discovering continuum shock models.
JOURNAL OF COMPUTATIONAL PHYSICS
(2022)
Article
Computer Science, Interdisciplinary Applications
Jan N. Fuhg et al.
Summary: This paper introduces the research status of deep neural networks as universal approximators of PDEs and compares the advantages and disadvantages of two different methods. An improved deep energy method is proposed to solve the fine features in solid mechanics and a numerical integration scheme is introduced to enhance the flexibility of the method.
JOURNAL OF COMPUTATIONAL PHYSICS
(2022)
Article
Multidisciplinary Sciences
Enrui Zhang et al.
Summary: In this study, a general framework based on physics-informed neural networks is proposed for identifying unknown geometric and material parameters in materials' internal structures and defects. By using a mesh-free method, the geometry of the material is parameterized, and the effectiveness of the method is validated using constitutive models. The framework can be applied to other inverse problems involving unknown material properties and deformable geometries.
Article
Engineering, Multidisciplinary
Alexander Henkes et al.
Summary: Physics informed neural networks are a method used in applied mathematics and engineering to solve partial differential equations. However, due to their global approximation approach, they face challenges in displaying localized effects and strong nonlinear solution fields. To overcome these issues, researchers have studied adaptive training strategies and domain decomposition.
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING
(2022)
Article
Engineering, Multidisciplinary
Jeremy Yu et al.
Summary: The article introduces a new method called gradient-enhanced physics-informed neural networks (gPINNs) to improve the accuracy of PINNs. gPINNs leverage the gradient information of the PDE residual to enhance the loss function. Experimental results demonstrate that gPINNs outperform traditional PINN with fewer training points.
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING
(2022)
Article
Engineering, Multidisciplinary
Somdatta Goswami et al.
Summary: This study proposes a physics-informed variational formulation of DeepONet (V-DeepONet) for brittle fracture analysis. By incorporating physics laws and labeled data in training, V-DeepONet accurately predicts key quantities in brittle fracture and has wide application potential.
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING
(2022)
Article
Computer Science, Interdisciplinary Applications
Hongwei Guo et al.
Summary: In this work, a deep collocation method (DCM) is proposed for solving three-dimensional potential problems in non-homogeneous media. The method combines a physics-informed neural network and material transfer learning to simplify the solution process. Various configurations are tested and validated, enhancing the effectiveness and robustness of the approach. The study also includes a global sensitivity analysis and a convergence proof of the DCM.
ENGINEERING WITH COMPUTERS
(2022)
Article
Chemistry, Physical
Jaber Rezaei Mianroodi et al.
Summary: This research demonstrates the seamless transition of the elastic response of materials between atomic and continuum scales using artificial intelligence. By training a convolutional neural network model, the researchers were able to accurately predict the elastic properties of nanoporous materials and achieve higher efficiency compared to traditional methods.
NPJ COMPUTATIONAL MATERIALS
(2022)
Review
Computer Science, Interdisciplinary Applications
Grace C. Y. Peng et al.
Summary: Machine learning excels in image recognition in fields like biomedicine, but struggles with sparse data. Integrating machine learning with multiscale modeling can create more efficient biological system models, benefiting both approaches.
ARCHIVES OF COMPUTATIONAL METHODS IN ENGINEERING
(2021)
Article
Jan Blechschmidt et al.
Article
Mathematics, Interdisciplinary Applications
Mauricio Fernandez et al.
Summary: This study investigates the capabilities of anisotropic theory-based, purely data-driven, and hybrid approaches to model the homogenized behavior of cubic lattice metamaterials. The results show that the data-driven models can accurately reproduce simulation data and manifest lattice instabilities, achieving superior accuracy in additional test scenarios.
COMPUTATIONAL MECHANICS
(2021)
Article
Engineering, Multidisciplinary
Ehsan Haghighat et al.
Summary: SciANN is a Python package for scientific computing and physics-informed deep learning. It utilizes TensorFlow and Keras to build deep neural networks and optimization models, allowing for solving partial differential equations with flexibility in setting complex functional forms.
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING
(2021)
Article
Materials Science, Multidisciplinary
Binbin Lin et al.
Summary: The study demonstrates the use of machine learning models to investigate the impact of micro structure features on the mechanical properties of paper materials, with findings showing that disorderness and contact area distribution play important roles in determining failure strain, effective stiffness, and failure strength of fiber networks.
MATERIALS & DESIGN
(2021)
Article
Computer Science, Interdisciplinary Applications
Kevin Linka et al.
Summary: Constitutive artificial neural networks (CANNs) are a novel machine learning architecture for data-driven modeling of the mechanical constitutive behavior of materials. By incorporating information from stress-strain data, materials theory, and additional information, CANNs can efficiently learn the constitutive behavior of complex materials with minimal training data. The ability to predict properties of new materials without existing stress-strain data makes CANNs potentially useful for in-silico material design in the future.
JOURNAL OF COMPUTATIONAL PHYSICS
(2021)
Article
Materials Science, Multidisciplinary
Filippo Masi et al.
Summary: TANNs is a physics-based neural network method for constitutive modeling, where thermodynamic principles are encoded into the network architecture using automatic differentiation. Its advantages include efficiency, robustness, accuracy in predictions, and maintaining thermodynamic consistency.
JOURNAL OF THE MECHANICS AND PHYSICS OF SOLIDS
(2021)
Article
Automation & Control Systems
Navid Zobeiry et al.
Summary: A physics-informed neural network is developed to solve conductive heat transfer PDEs with convective boundary conditions, improving the speed and accuracy of thermal analysis in manufacturing and engineering applications. By using physics-informed activation functions, heat transfer beyond training zone can be accurately predicted, making it a useful tool for real-time evaluation of thermal responses in a wide range of scenarios.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2021)
Article
Thermodynamics
Shengze Cai et al.
Summary: Physics-informed neural networks (PINNs) have gained popularity in engineering fields for their effectiveness in solving realistic problems with noisy data and partially missing physics. Through automatic differentiation to evaluate differential operators and defining a multitask learning problem, PINNs have been applied to various heat transfer problems, bridging the gap between computational and experimental heat transfer.
JOURNAL OF HEAT TRANSFER-TRANSACTIONS OF THE ASME
(2021)
Article
Multidisciplinary Sciences
Zhenze Yang et al.
Summary: An AI-based method is developed to predict physical properties of materials directly from their microstructure geometry, bridging the gap between design space and physical performance. The approach shows remarkable accuracy in predicting physical fields and derivative material properties, and offers extensibility in predicting complex materials behavior, improving the efficiency of evaluating physical properties of hierarchical materials.
Article
Engineering, Multidisciplinary
Ehsan Haghighat et al.
Summary: This study presents the application of Physics Informed Neural Networks (PINN) in solid mechanics, improving accuracy and convergence with a multi-network model and Isogeometric Analysis. The study demonstrates the importance of honoring physics in improving robustness and highlights the potential application of PINN in sensitivity analysis and surrogate modeling.
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING
(2021)
Article
Chemistry, Physical
Jaber Rezaei Mianroodi et al.
Summary: A deep neural network (DNN) is proposed as a fast surrogate model for local stress calculations in inhomogeneous non-linear materials, showing efficient performance and high accuracy in predicting stress distribution. The DNN model is capable of simulating micromechanics in elastic-plastic materials with significant acceleration compared to typical solvers.
NPJ COMPUTATIONAL MATERIALS
(2021)
Article
Mechanics
Zhizhou Zhang et al.
Summary: A physics-informed neural network (PINN) for digital materials analysis is introduced, trained without ground truth data using minimum energy criteria. The model reached similar accuracy as supervised ML models and prevented erroneous deformation gradients with hinge loss on the Jacobian. Parallel computing on GPU for strain energy calculation showed linear scalability with the number of nodes, laying the foundation for label-free learning in designing next-generation composites.
THEORETICAL AND APPLIED MECHANICS LETTERS
(2021)
Review
Physics, Applied
George Em Karniadakis et al.
Summary: Physics-informed learning seamlessly integrates data and mathematical models through neural networks or kernel-based regression networks for accurate inference of realistic and high-dimensional multiphysics problems. Challenges remain in incorporating noisy data seamlessly, complex mesh generation, and addressing high-dimensional problems.
NATURE REVIEWS PHYSICS
(2021)
Article
Engineering, Multidisciplinary
Hamid Reza Bayat et al.
INTERNATIONAL JOURNAL FOR NUMERICAL METHODS IN ENGINEERING
(2020)
Article
Engineering, Multidisciplinary
E. Samaniego et al.
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING
(2020)
Article
Engineering, Mechanical
Somdatta Goswami et al.
THEORETICAL AND APPLIED FRACTURE MECHANICS
(2020)
Article
Water Resources
QiZhi He et al.
ADVANCES IN WATER RESOURCES
(2020)
Article
Engineering, Multidisciplinary
Ameya D. Jagtap et al.
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING
(2020)
Article
Chemistry, Physical
Siddhant Kumar et al.
NPJ COMPUTATIONAL MATERIALS
(2020)
Article
Materials Science, Multidisciplinary
Yu-Chuan Hsu et al.
Article
Computer Science, Interdisciplinary Applications
M. Raissi et al.
JOURNAL OF COMPUTATIONAL PHYSICS
(2019)
Review
Materials Science, Multidisciplinary
Frederic E. Bock et al.
FRONTIERS IN MATERIALS
(2019)
Article
Multidisciplinary Sciences
M. Mozaffar et al.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2019)
Article
Engineering, Multidisciplinary
Kun Wang et al.
COMPUTER METHODS IN APPLIED MECHANICS AND ENGINEERING
(2018)
Article
Computer Science, Artificial Intelligence
Jens Berg et al.
Review
Computer Science, Artificial Intelligence
Juergen Schmidhuber
