Academic literature on the topic 'Physics-Informed neural network'
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the lists of relevant articles, books, theses, conference reports, and other scholarly sources on the topic 'Physics-Informed neural network.'
Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.
You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.
Journal articles on the topic "Physics-Informed neural network"
1
Trahan, Corey, Mark Loveland, and Samuel Dent. "Quantum Physics-Informed Neural Networks." Entropy 26, no. 8 (2024): 649. http://dx.doi.org/10.3390/e26080649.
Full textAbstract:
In this study, the PennyLane quantum device simulator was used to investigate quantum and hybrid, quantum/classical physics-informed neural networks (PINNs) for solutions to both transient and steady-state, 1D and 2D partial differential equations. The comparative expressibility of the purely quantum, hybrid and classical neural networks is discussed, and hybrid configurations are explored. The results show that (1) for some applications, quantum PINNs can obtain comparable accuracy with less neural network parameters than classical PINNs, and (2) adding quantum nodes in classical PINNs can increase model accuracy with less total network parameters for noiseless models.
2
Wang, Jing, Yubo Li, Anping Wu, et al. "Multi-Step Physics-Informed Deep Operator Neural Network for Directly Solving Partial Differential Equations." Applied Sciences 14, no. 13 (2024): 5490. http://dx.doi.org/10.3390/app14135490.
Full textAbstract:
This paper establishes a method for solving partial differential equations using a multi-step physics-informed deep operator neural network. The network is trained by embedding physics-informed constraints. Different from traditional neural networks for solving partial differential equations, the proposed method uses a deep neural operator network to indirectly construct the mapping relationship between the variable functions and solution functions. This approach makes full use of the hidden information between the variable functions and independent variables. The process whereby the model captures incredibly complex and highly nonlinear relationships is simplified, thereby making network learning easier and enhancing the extraction of information about the independent variables in partial differential systems. In terms of solving partial differential equations, we verify that the multi-step physics-informed deep operator neural network markedly improves the solution accuracy compared with a traditional physics-informed deep neural operator network, especially when the problem involves complex physical phenomena with large gradient changes.
3
Hofmann, Tobias, Jacob Hamar, Marcel Rogge, Christoph Zoerr, Simon Erhard, and Jan Philipp Schmidt. "Physics-Informed Neural Networks for State of Health Estimation in Lithium-Ion Batteries." Journal of The Electrochemical Society 170, no. 9 (2023): 090524. http://dx.doi.org/10.1149/1945-7111/acf0ef.
Full textAbstract:
One of the most challenging tasks of modern battery management systems is the accurate state of health estimation. While physico-chemical models are accurate, they have high computational cost. Neural networks lack physical interpretability but are efficient. Physics-informed neural networks tackle the aforementioned shortcomings by combining the efficiency of neural networks with the accuracy of physico-chemical models. A physics-informed neural network is developed and evaluated against three different datasets: A pseudo-two-dimensional Newman model generates data at various state of health points. This dataset is fused with experimental data from laboratory measurements and vehicle field data to train a neural network in which it exploits correlation from internal modeled states to the measurable state of health. The resulting physics-informed neural network performs best with the synthetic dataset and achieves a root mean squared error below 2% at estimating the state of health. The root mean squared error stays within 3% for laboratory test data, with the lowest error observed for constant current discharge samples. The physics-informed neural network outperforms several other purely data-driven methods and proves its advantage. The inclusion of physico-chemical information from simulation increases accuracy and further enables broader application ranges.
4
Karakonstantis, Xenofon, Diego Caviedes-Nozal, Antoine Richard, and Efren Fernandez-Grande. "Room impulse response reconstruction with physics-informed deep learning." Journal of the Acoustical Society of America 155, no. 2 (2024): 1048–59. http://dx.doi.org/10.1121/10.0024750.
Full textAbstract:
A method is presented for estimating and reconstructing the sound field within a room using physics-informed neural networks. By incorporating a limited set of experimental room impulse responses as training data, this approach combines neural network processing capabilities with the underlying physics of sound propagation, as articulated by the wave equation. The network's ability to estimate particle velocity and intensity, in addition to sound pressure, demonstrates its capacity to represent the flow of acoustic energy and completely characterise the sound field with only a few measurements. Additionally, an investigation into the potential of this network as a tool for improving acoustic simulations is conducted. This is due to its proficiency in offering grid-free sound field mappings with minimal inference time. Furthermore, a study is carried out which encompasses comparative analyses against current approaches for sound field reconstruction. Specifically, the proposed approach is evaluated against both data-driven techniques and elementary wave-based regression methods. The results demonstrate that the physics-informed neural network stands out when reconstructing the early part of the room impulse response, while simultaneously allowing for complete sound field characterisation in the time domain.
5
Li, Zhenyu. "A Review of Physics-Informed Neural Networks." Applied and Computational Engineering 133, no. 1 (2025): 165–73. https://doi.org/10.54254/2755-2721/2025.20636.
Full textAbstract:
This article presents Physics-Informed Neural Networks (PINNs), which integrate physical laws into neural network training to model complex systems governed by partial differential equations (PDEs). PINNs enhance data efficiency, allowing for accurate predictions with less training data, and have applications in fields such as biomedical engineering, geophysics, and material science. Despite their advantages, PINNs face challenges like learning high-frequency components and computational overhead. Proposed solutions include causality constraints and improved boundary condition handling. A numerical experiment demonstrates the effectiveness of PINNs in solving the one-dimensional heat conduction equation, showcasing enhanced model stability and accuracy. Overall, PINNs represent a significant advancement in merging machine learning with physics.
6
Pu, Ruilong, and Xinlong Feng. "Physics-Informed Neural Networks for Solving Coupled Stokes–Darcy Equation." Entropy 24, no. 8 (2022): 1106. http://dx.doi.org/10.3390/e24081106.
Full textAbstract:
In this paper, a grid-free deep learning method based on a physics-informed neural network is proposed for solving coupled Stokes–Darcy equations with Bever–Joseph–Saffman interface conditions. This method has the advantage of avoiding grid generation and can greatly reduce the amount of computation when solving complex problems. Although original physical neural network algorithms have been used to solve many differential equations, we find that the direct use of physical neural networks to solve coupled Stokes–Darcy equations does not provide accurate solutions in some cases, such as rigid terms due to small parameters and interface discontinuity problems. In order to improve the approximation ability of a physics-informed neural network, we propose a loss-function-weighted function strategy, a parallel network structure strategy, and a local adaptive activation function strategy. In addition, the physical information neural network with an added strategy provides inspiration for solving other more complicated problems of multi-physical field coupling. Finally, the effectiveness of the proposed strategy is verified by numerical experiments.
7
Kenzhebek, Y., T. S. Imankulov, and D. Zh Akhmed-Zaki. "PREDICTION OF OIL PRODUCTION USING PHYSICS-INFORMED NEURAL NETWORKS." BULLETIN Series of Physics & Mathematical Sciences 76, no. 4 (2021): 45–50. http://dx.doi.org/10.51889/2021-4.1728-7901.06.
Full textAbstract:
In recent years, modern information technologies have been actively used in various industries. The oil industry is no exception, since high-performance computing technologies, artificial intelligence algorithms, methods of collecting, processing and storing information are actively used to solve the problems of increasing oil recovery. Deep learning has made remarkable strides in a variety of applications, but its use for solving partial differential equations has only recently emerged. In particular, you can replace traditional numerical methods with a neural network that approximates the solution to a partial differential equation. Physically Informed Neural Networks (PINNs) embed partial differential equations into the neural network loss function using automatic differentiation. A numerical algorithm and PINN have been developed for solving the one-dimensional pressure equation from the Buckley-Leverett mathematical model. The results of numerical solution and prediction of the PINN neural network for solving the pressure equation are obtained.
8
Hou, Shubo, Wenchao Wu, and Xiuhong Hao. "Physics-informed neural network for simulating magnetic field of permanent magnet." Journal of Physics: Conference Series 2853, no. 1 (2024): 012018. http://dx.doi.org/10.1088/1742-6596/2853/1/012018.
Full textAbstract:
Abstract With the rapid development of deep learning, its application in physical field simulation has been widely concerned, and it has begun to lead a new model of meshless simulation. In this paper, research based on physics-informed neural networks is carried out to solve partial differential equations related to the physical laws of electromagnetism. Then the magnetic field simulation is realized. In this method, the governing equation and the boundary conditions containing physical information are embedded into the neural network loss function as constraints, and the backpropagation of neural networks is realized based on automatic differentiation to solve partial differential equations. The high-precision simulation of tile-shaped and rectangular permanent magnet magnetic fields of permanent magnet motors based on physical information neural network is studied, and the error is within 5%. We consider the simulation of magnetic field in two coordinate systems, and realize the joint training of multiple neural networks in multiple sub-domains and different media.
9
Pan, Cunliang, Shi Feng, Shengyang Tao, Hongwu Zhang, Yonggang Zheng, and Hongfei Ye. "Physics-Informed Neural Network for Young-Laplace Equation." International Conference on Computational & Experimental Engineering and Sciences 30, no. 1 (2024): 1. http://dx.doi.org/10.32604/icces.2024.011132.
Full text
10
Yoon, Seunghyun, Yongsung Park, and Woojae Seong. "Improving mode extraction with physics-informed neural network." Journal of the Acoustical Society of America 154, no. 4_supplement (2023): A339—A340. http://dx.doi.org/10.1121/10.0023729.
Full textAbstract:
This study aims to enhance conventional mode extraction methods in ocean waveguides using a physics-informed neural network (PINN). Mode extraction involves estimating mode wavenumbers and corresponding mode depth functions. The approach considers a scenario with a single frequency source towed at a constant depth and measured from a vertical line array (VLA). Conventional mode extraction methods applied to experimental data face two problems. First, mode shape estimation is limited because the receivers only cover a partial waveguide. Second, the wavenumber spectrum is affected by issues such as Doppler shift and range errors. To address these challenges, we train the PINN with measured data, generating a densely sampled complex pressure field, including the unmeasured region above the VLA. We then apply the same mode extraction methods to both the raw data and the PINN-generated data for comparison. The proposed method is validated using data from the SWellEx-96, demonstrating improved mode extraction performance compared to using raw experimental data directly.
Dissertations / Theses on the topic "Physics-Informed neural network"
1
Mirzai, Badi. "Physics-Informed Deep Learning for System Identification of Autonomous Underwater Vehicles : A Lagrangian Neural Network Approach." Thesis, KTH, Skolan för elektroteknik och datavetenskap (EECS), 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-301626.
Full textAbstract:
In this thesis, we explore Lagrangian Neural Networks (LNNs) for system identification of Autonomous Underwater Vehicles (AUVs) with 6 degrees of freedom. One of the main challenges of AUVs is that they have limited wireless communication and navigation under water. AUVs operate under strict and uncertain conditions, where they need to be able to navigate and perform tasks in unknown ocean environments with limited and noisy sensor data. A crucial requirement for localization and adaptive control of AUVs is having an accurate and reliable model of the system’s nonlinear dynamics while taking into account the dynamic environment of the ocean. Most of these dynamics models do not incorporate data. The collection of data for AUVs is difficult, but necessary in order to have more flexibility in the model’s parameters due to the dynamic environment of the ocean. Yet, traditional system identification methods are still dominant today, despite the recent breakthroughs in Deep Learning. Therefore, in this thesis, we aim for a data-driven approach that embeds laws from physics in order to learn the state-space model of an AUV. More precisely, exploring the LNN framework for higher-dimensional systems. Furthermore, we also extend the LNN to account for non-conservative forces acting upon the system, such as damping and control inputs. The networks are trained to learn from simulated data of a second-order ordinary differential equation of an AUV. The trained model is evaluated by integrating paths from different initial states and comparing them to the true dynamics. The results yielded a model capable of predicting the output acceleration of the state space model but struggled in learning the direction of the forward movement with time.<br>I den här uppsatsen utforskas Lagrangianska Neurala Nätverk (LNN) för systemidentifiering av Autonoma Undervattensfordon (AUV) med 6 frihetsgrader. En av de största utmaningarna med AUV är deras begränsningar när det kommer till trådlös kommunikation och navigering under vatten. Ett krav för att ha fungerande AUV är deras förmåga att navigera och utföra uppdrag under okända undervattensförhållanden med begränsad och brusig sensordata. Dessutom är ett kritiskt krav för lokalisering och adaptiv reglerteknik att ha noggranna modeller av systemets olinjära dynamik, samtidigt som den dynamiska miljön i havet tas i beaktande. De flesta sådana modeller tar inte i beaktande sensordata för att reglera dess parameterar. Insamling av sådan data för AUVer är besvärligt, men nödvändigt för att skapa större flexibilitet hos modellens parametrar. Trots de senaste genombrotten inom djupinlärning är traditionella metoder av systemidentifiering dominanta än idag för AUV. Det är av dessa anledningar som vi i denna uppsats strävar efter en datadriven metod, där vi förankrar lagar från fysik under inlärningen av systemets state-space modell. Mer specifikt utforskar vi LNN för ett system med högre dimension. Vidare expanderar vi även LNN till att även ta ickekonservativa krafter som verkar på systemet i beaktande, såsom dämpning och styrsignaler. Nätverket tränas att lära sig från simulerad data från en andra ordningens differentialekvation som beskriver en AUV. Den tränade modellen utvärderas genom att iterativt integrera fram dess rörelse från olika initialstillstånd, vilket jämförs med den korrekta modellen. Resultaten visade en modell som till viss del var kapabel till att förutspå korrekt acceleration, med begränsad framgång i att lära sig korrekt rörelseriktning framåt i tiden.
2
Cedergren, Linnéa. "Physics-informed Neural Networks for Biopharma Applications." Thesis, Umeå universitet, Institutionen för fysik, 2021. http://urn.kb.se/resolve?urn=urn:nbn:se:umu:diva-185423.
Full textAbstract:
Physics-Informed Neural Networks (PINNs) are hybrid models that incorporate differential equations into the training of neural networks, with the aim of bringing the best of both worlds. This project used a mathematical model describing a Continuous Stirred-Tank Reactor (CSTR), to test two possible applications of PINNs. The first type of PINN was trained to predict an unknown reaction rate law, based only on the differential equation and a time series of the reactor state. The resulting model was used inside a multi-step solver to simulate the system state over time. The results showed that the PINN could accurately model the behaviour of the missing physics also for new initial conditions. However, the model suffered from extrapolation error when tested on a larger reactor, with a much lower reaction rate. Comparisons between using a numerical derivative or automatic differentiation in the loss equation, indicated that the latter had a higher robustness to noise. Thus, it is likely the best choice for real applications. A second type of PINN was trained to forecast the system state one-step-ahead based on previous states and other known model parameters. An ordinary feed-forward neural network with an equal architecture was used as baseline. The second type of PINN did not outperform the baseline network. Further studies are needed to conclude if or when physics-informed loss should be used in autoregressive applications.
3
Ding, Simon. "Advancing cosmological field-level inference with physics-informed Bayesian neural networks." Electronic Thesis or Diss., Sorbonne université, 2025. http://www.theses.fr/2025SORUS050.
Full textAbstract:
La cosmologie repose essentiellement sur des observations passives, obtenues avec des télescopes sophistiqués, pour comprendre l'origine, la dynamique et le destin ultime de l'Univers. Les nouveaux relevés de galaxies propulsent la cosmologie dans une ère dominée par les données, nécessitant des adaptations significatives de nos méthodes d'analyse. L'un des principaux défis de la cosmologie moderne est d'extraire des informations physiques significatives des prochains relevés cosmologiques. Parallèlement à cette révolution, l'apprentissage automatique s'impose comme une technique puissante pour les applications nécessitant de grandes quantités de données, au prix d'un grand nombre de paramètres. Cette thèse vise à développer des modèles précis pour surmonter les limites actuelles de l'inférence cosmologique basée sur la structuration à grande échelle des galaxies. Nous explorons deux nouvelles approches intégrant des concepts physiques pour incorporer des modèles d'apprentissage automatique, tels que les réseaux de neurones bayésiens, dans une chaîne complète d'inférence bayésienne. La première méthode concerne la modélisation du biais des galaxies, une relation statistique entre la densité de matière et les galaxies. La formation des galaxies est un processus complexe dont la résolution par des simulations à N-corps nécessite un temps de calcul conséquent, tandis que les techniques d'inférence requièrent des évaluations rapides. Par conséquent, des modèles de biais approximatifs sont utilisés, au détriment de la précision. Nous avons conçu un réseau de neurones capable de générer des halos avec une structuration spatiale comparable à celle obtenue par les simulations à N-corps. Ce modèle repose sur un ensemble réduit de paramètres (19 à 35), obtenus grâce aux symétries et à une expertise phénoménologique. Il permet de synthétiser des halos quasi directement à partir de la théorie des perturbations, de manière numériquement efficace. Nous avons appliqué ce modèle à la génération de catalogues de halos, avec des propriétés similaires aux simulations à N-corps, notamment pour les fonctions de corrélation à N-points. Nous démontrons que ce modèle peut être intégré de manière transparente dans une méthode d'inférence au niveau du champ, tout en maintenant un comportement d'échantillonnage stable. La deuxième méthode présentée ici aborde le problème du choix des priors pertinents pour l'inférence bayésienne. Elle s'appuie sur des informations issues de la physique pour tenir compte des effets systématiques ou des mis-spécifications. Les modèles introduisent des paramètres supplémentaires pour l'inférence cosmologique. Sans priors contraignants, l'augmentation du nombre de paramètres entraîne une perte d'information sur les paramètres pertinents, notamment en cosmologie. Cet effet est exacerbé par l'utilisation de modèles précis mais sur-paramétrés issus des techniques d'apprentissage automatique. À cette fin, nous présentons une nouvelle approche pour dériver des priors à partir de simulations à N-corps afin de réduire l'espace des modèles aux seules solutions physiquement plausibles. Ces priors supplémentaires sont formulés selon les principes du raisonnement bayésien. Nous présentons l'intégration de nos nouveaux priors dans une inférence bayésienne pour contraindre le modèle de biais des halos de matière noire. Enfin, nous montrons que ces priors permettent d'améliorer les contraintes sur les paramètres du biais ainsi que l'efficacité de l'échantillonnage. Notre approche constitue une extension importante des cadres d'inférence bayésienne existants en physique. De plus, les méthodes basées sur la physique développées dans cette thèse initient l'incorporation rigoureuse de modèles complexes d'apprentissage automatique pour contraindre la cosmologie<br>Cosmology may only derive insights into our Universe's origin, dynamics, and ultimate fate through observations. To accomplish this, we design and build sophisticated telescopes, and with next-generation surveys, cosmology will transform into a data-driven era. However, accommodating the subsequent change in the data landscape will require significant adaptations in our current data analysis frameworks. One of the main challenges in modern cosmology will be to design accurate data models in order to extract meaningful physical information from the upcoming surveys. Concurrently, machine learning has emerged as a powerful technique for data-driven applications. The present thesis aims to develop accurate data models to address the current limitations in cosmological inference from galaxy clustering surveys. For this purpose, we explore two novel physics-informed approaches to incorporate machine-learning models, such as Bayesian neural networks, into end-to-end Bayesian inference frameworks. The first method pertains to modelling the galaxy bias, the link between matter density and observable galaxies. Galaxy formation is a complex process, and resolving this with N-body simulations, especially if magneto-hydrodynamics are included, requires significant computational time. Inversely, inference frameworks require fast model evaluation; therefore, approximate bias models are utilised at the cost of accuracy, which becomes one of the limiting factors in inference frameworks. As an initial step, we designed a neural network to produce halos comparable to N-body codes. Our presented model relies on a reduced set of 19 to 35 parameters, realized through symmetries and domain-knowledge choices motivated by first principles. Furthermore, the halos can be generated from approximate dark matter density fields, enhancing the model's computational efficiency. In its first application, we show that our network can be adopted as a mock halo generator, producing halo catalogues consistent with N-body halos across different N-point correlation functions. Subsequently, due to the physics-informed architecture with a minimal parameter set, we demonstrate that our model can be seamlessly integrated into an existing field-level inference framework, showing stable sampling behaviour. The second physics-informed framework addresses the challenge of specifying meaningful parameter priors for Bayesian inference. In order to account for unknown systematics or to incorporate additional physical effects, data models introduce additional parameters for cosmological inference. Without setting constraining priors, the growing number of parameters will cause the data to be less informative on the physical parameters of interest, notably the ones related to cosmology. This effect will be exacerbated if we want to include accurate but over-parametrised models from machine learning as Bayesian neural networks. To this end, we present a novel approach to derive priors from N-body simulations in order to constrain the effective parameter space of inference models to physically plausible solutions. These additional priors are formulated following the principles of Bayesian reasoning. We present the integration of our physics-informed prior approach into an existing Bayesian inference engine to constrain the halo bias in the hierarchical data model. Finally, we show that the additional priors were able to improve constraints on the bias parameters while also improving their sampling efficiency. Consequently, we demonstrate that our approaches are valuable extensions to existing Bayesian inference frameworks. Moreover, the physics-informed frameworks developed in this thesis serve as a starting point for incorporating intricate machine-learning models in a rigorous manner for more accurate and constraining cosmological inferences from upcoming galaxy surveys
4
Elhawary, Mohamed. "Apprentissage profond informé par la physique pour les écoulements complexes." Electronic Thesis or Diss., Paris, ENSAM, 2024. http://www.theses.fr/2024ENAME068.
Full textAbstract:
Ce travail de doctorat étudie deux problèmes spécifiques concernant les turbomachines en utilisant des algorithmes d'apprentissage automatique. Le premier se concentre sur un compresseur axial, en abordant les problèmes de décrochage tournant, qui sont des phénomènes instables limitant la plage de fonctionnement des compresseurs. Les avancées récentes comprennent le développement de techniques de contrôle d'écoulement, telles que des jets au niveau du carter et du bord d’attaque du rotor, qui ont montré un potentiel pour étendre les plages de fonctionnement des compresseurs. Cependant, l’optimisation de ces stratégies de contrôle représente un défi en raison du grand nombre de paramètres et de configurations, y compris le nombre de jets, la vitesse d’injection et l’angle d’injection dans le cadre fixe. Cela soulève la question suivante : les algorithmes d'apprentissage automatique peuvent-ils aider à explorer cet vaste espace de paramètres et à optimiser la stratégie de contrôle ? À cette fin, une base de données complète des résultats expérimentaux issus de divers paramètres de contrôle et évaluations des performances du compresseur sur un compresseur axial a été utilisée, avec des tests effectués sur le banc d'essai CME2 au laboratoire LMFL. Le deuxième problème examine un diffuseur lisse radial, un composant statorique annulaire positionné en aval du rotor dans les pompes et compresseurs radiaux. Son rôle principal est de décélérer le fluide tout en augmentant la pression statique et l'enthalpie. Malgré sa fonction apparemment simple, prédire le comportement de l'écoulement à l'intérieur du diffuseur est assez difficile en raison du manque de guidage du fluide, de la structure complexe du flux de jet à l'entrée, des instabilités d'écoulement et de la nature tridimensionnelle du flux. Cela mène à la question suivante : les algorithmes d'apprentissage automatique peuvent-ils prédire efficacement cet écoulement ? Pour cette analyse, nous utilisons une base de données composée de simulations numériques (URANS) réalisées sur une géométrie de pompe centrifuge effectuées au laboratoire LMFL. Nous avons employé deux approches d'apprentissage automatique pour étudier ces sujets distincts liés aux dispositifs de turbomachinerie. La première approche utilise des réseaux de neurones (NN) et des algorithmes génétiques (GA) pour explorer des stratégies de contrôle actif du flux dans un compresseur axial. La deuxième approche applique des réseaux de neurones informés par la physique (PINN) pour modéliser un écoulement turbulent en 2D dans le diffuseur d'une pompe radiale<br>This PhD work investigates two specific problems concerning turbomachinery using machine learning algorithms. The first focuses on the axial flow compressor, addressing the issues of rotating stall and surge which is unstable phenomena that limit the operational range of compressors. Recent advancements include the development of flow control techniques, such as jets at the casing and leading edge of the rotor, which have shown promise in extending compressor operating ranges. However, optimizing these control strategies poses a challenge due to the large number of parameters and configurations, including the number of jets, the injection velocity, and the injection angle in the fixed frame. This raises the question: can ML algorithms assist in exploring this extensive parameter space and optimizing the control strategy? To this end, a comprehensive database of experimental results from various control parameters and compressor performance evaluations on an axial flow compressor has been utilized, with tests conducted on the CME2 test bench at LMFL laboratory. The second problem examines the radial vaneless diffuser, an annular stator component positioned downstream of the rotor in radial pumps and compressors. Its primary role is to decelerate the fluid while increasing static pressure and enthalpy. Despite its seemingly straightforward function, predicting the flow behaviour within the diffuser is quite challenging due to the lack of fluid guidance, the complex jet wake flow structure at the inlet, flow instabilities, three-dimensional nature of the flow. This leads to the inquiry: can ML algorithms effectively predict this flow? For this analysis, we utilize a database consisting of numerical simulations (URANS) obtained on a radial flow pump geometry performed at LMFL laboratory. We employed two machine learning approaches to investigate these distinct topics related to turbomachinery devices. The first approach utilizes Neural Networks (NNs) and Genetic Algorithms (GAs) to explore active flow control strategies in an axial compressor. The second approach applies Physics-Informed Neural Networks (PINNs) to model 2D turbulent flow in the vaneless diffuser of a radial pump
5
Wu, Dawen. "Solving Some Nonlinear Optimization Problems with Deep Learning." Electronic Thesis or Diss., université Paris-Saclay, 2023. http://www.theses.fr/2023UPASG083.
Full textAbstract:
Cette thèse considère quatre types de problèmes d'optimisation non linéaire, à savoir les jeux de bimatrice, les équations de projection non linéaire (NPEs), les problèmes d'optimisation convexe non lisse (NCOPs) et les jeux à contraintes stochastiques (CCGs). Ces quatre classes de problèmes d'optimisation non linéaire trouvent de nombreuses applications dans divers domaines tels que l'ingénierie, l'informatique, l'économie et la finance. Notre objectif est d'introduire des algorithmes basés sur l'apprentissage profond pour calculer efficacement les solutions optimales de ces problèmes d'optimisation non linéaire.Pour les jeux de bimatrice, nous utilisons des réseaux neuronaux convolutionnels (CNNs) pour calculer les équilibres de Nash. Plus précisément, nous concevons une architecture de CNN où l'entrée est un jeu de bimatrice et la sortie est l'équilibre de Nash prédit pour le jeu. Nous générons un ensemble de jeux de bimatrice suivant une distribution de probabilité donnée et utilisons l'algorithme de Lemke-Howson pour trouver leurs véritables équilibres de Nash, constituant ainsi un ensemble d'entraînement. Le CNN proposé est formé sur cet ensemble de données pour améliorer sa précision. Une fois l'apprentissage terminée, le CNN est capable de prédire les équilibres de Nash pour des jeux de bimatrice inédits. Les résultats expérimentaux démontrent l'efficacité computationnelle exceptionnelle de notre approche basée sur CNN, au détriment de la précision.Pour les NPEs, NCOPs et CCGs, qui sont des problèmes d'optimisation plus complexes, ils ne peuvent pas être directement introduits dans les réseaux neuronaux. Par conséquent, nous avons recours à des outils avancés, à savoir l'optimisation neurodynamique et les réseaux neuronaux informés par la physique (PINNs), pour résoudre ces problèmes. Plus précisément, nous utilisons d'abord une approche neurodynamique pour modéliser un problème d'optimisation non linéaire sous forme de système d'équations différentielles ordinaires (ODEs). Ensuite, nous utilisons un modèle basé sur PINN pour résoudre le système d'ODE résultant, où l'état final du modèle représente la solution prédite au problème d'optimisation initial. Le réseau neuronal est formé pour résoudre le système d'ODE, résolvant ainsi le problème d'optimisation initial. Une contribution clé de notre méthode proposée réside dans la transformation d'un problème d'optimisation non linéaire en un problème d'entraînement de réseau neuronal. En conséquence, nous pouvons maintenant résoudre des problèmes d'optimisation non linéaire en utilisant uniquement PyTorch, sans compter sur des solveurs d'optimisation convexe classiques tels que CVXPY, CPLEX ou Gurobi<br>This thesis considers four types of nonlinear optimization problems, namely bimatrix games, nonlinear projection equations (NPEs), nonsmooth convex optimization problems (NCOPs), and chance-constrained games (CCGs).These four classes of nonlinear optimization problems find extensive applications in various domains such as engineering, computer science, economics, and finance.We aim to introduce deep learning-based algorithms to efficiently compute the optimal solutions for these nonlinear optimization problems.For bimatrix games, we use Convolutional Neural Networks (CNNs) to compute Nash equilibria.Specifically, we design a CNN architecture where the input is a bimatrix game and the output is the predicted Nash equilibrium for the game.We generate a set of bimatrix games by a given probability distribution and use the Lemke-Howson algorithm to find their true Nash equilibria, thereby constructing a training dataset.The proposed CNN is trained on this dataset to improve its accuracy. Upon completion of training, the CNN is capable of predicting Nash equilibria for unseen bimatrix games.Experimental results demonstrate the exceptional computational efficiency of our CNN-based approach, at the cost of sacrificing some accuracy.For NPEs, NCOPs, and CCGs, which are more complex optimization problems, they cannot be directly fed into neural networks.Therefore, we resort to advanced tools, namely neurodynamic optimization and Physics-Informed Neural Networks (PINNs), for solving these problems.Specifically, we first use a neurodynamic approach to model a nonlinear optimization problem as a system of Ordinary Differential Equations (ODEs).Then, we utilize a PINN-based model to solve the resulting ODE system, where the end state of the model represents the predicted solution to the original optimization problem.The neural network is trained toward solving the ODE system, thereby solving the original optimization problem.A key contribution of our proposed method lies in transforming a nonlinear optimization problem into a neural network training problem.As a result, we can now solve nonlinear optimization problems using only PyTorch, without relying on classical convex optimization solvers such as CVXPY, CPLEX, or Gurobi
6
Quattromini, Michele. "Graph Neural Networks for fluid mechanics : data-assimilation and optimization." Electronic Thesis or Diss., université Paris-Saclay, 2024. http://www.theses.fr/2024UPAST161.
Full textAbstract:
Cette thèse de doctorat explore l'application des réseaux de neurones en graphes (GNN) dans le domaine de la dynamique des fluides numérique (CFD), avec un accent particulier sur l'assimilation de données et l'optimisation. Le travail est structuré en trois parties principales: assimilation de données pour les équations de Navier-Stokes moyennées à la Reynolds (RANS) basée sur des modèles GNN; assimilation de données augmentée par les GNN avec des contraintes physiques imposées par la méthode adjointe; optimisation des systèmes fluides par des techniques d'apprentissage automatique (ML).Dans la première partie, la thèse examine le potentiel des GNN pour contourner les modèles de fermeture traditionnels, qui nécessitent souvent une calibration manuelle et sont sujets à des inexactitudes. En exploitant des données de simulation à haute fidélité, les GNN sont entraînés à apprendre directement les quantités non résolues de l'écoulement, offrant ainsi un cadre plus flexible pour le problème de fermeture des équations RANS. Cette approche élimine le besoin de modèles de fermeture calibrés manuellement, fournissant une alternative généralisée et basée sur les données. De plus, dans cette première partie, une étude approfondie de l'impact de la quantité de données sur les performances des GNN est réalisée, avec la conception d'une stratégie d'Active Learning pour sélectionner les données les plus informatives parmi celles disponibles. Sur la base de ces résultats, la deuxième partie de la thèse aborde un défi critique souvent rencontré par les modèles d'apprentissage automatique : l'absence de garantie de cohérence physique dans leurs prédictions. Afin de garantir que les GNN non seulement minimisent les erreurs, mais produisent également des résultats physiquement valides, cette partie intègre des contraintes physiques directement dans le processus d'entraînement des GNN. En incorporant les principes clés de la mécanique des fluides dans le cadre de l'apprentissage automatique, le modèle produit des prédictions à la fois fiables et cohérentes avec les lois physiques sous-jacentes, améliorant ainsi son applicabilité aux problèmes réels. Dans la troisième partie, la thèse démontre l'application des GNN pour optimiser les systèmes de dynamique des fluides, avec un accent particulier sur la conception des éoliennes. Ici, les GNN sont utilisés comme modèles de substitution, permettant des prédictions rapides de diverses configurations de conception sans avoir besoin de réaliser une simulation CFD complète à chaque itération. Cette approche accélère considérablement le processus de conception et montre le potentiel de l'optimisation basée sur l'apprentissage automatique dans le cadre de la CFD, permettant une exploration plus efficace des espaces de conception et une convergence plus rapide vers des solutions optimales. Sur le plan méthodologique, la thèse introduit une architecture GNN sur mesure spécifiquement adaptée aux applications CFD. Contrairement aux réseaux de neurones traditionnels, les GNN sont intrinsèquement capables de gérer des données de maillage non structurées, ce qui est courant dans les problèmes de mécanique des fluides impliquant des géométries irrégulières et des domaines d'écoulement complexes. À cette fin, la thèse présente une interface en deux parties entre les solveurs de la méthode des éléments finis (FEM) et l'architecture GNN. Cette interface transforme les champs vectoriels FEM en tenseurs numériques pouvant être traités efficacement par le réseau neuronal, permettant ainsi l'échange de données entre l'environnement de simulation et le modèle d'apprentissage<br>This PhD thesis investigates the application of Graph Neural Networks (GNNs) in the field of Computational Fluid Dynamics (CFD), with a focus on data-assimilation and optimization. The work is structured into three main parts: data-assimilation for Reynolds-Averaged Navier-Stokes (RANS) equations based on GNN models; data-assimilation augmented by GNN and adjoint-based enforced physical constraint; fluid systems optimization by ML techniques. In the first part, the thesis explores the potential of GNNs to bypass traditional closure models, which often require manual calibration and are prone to inaccuracies. By leveraging high-fidelity simulation data, GNNs are trained to directly learn the unresolved flow quantities, offering a more flexible framework for the RANS closure problem. This approach eliminates the need for manually tuned closure models, providing a generalized and data-driven alternative. Moreover, in this first part, a comprehensive study of the impact of data quantity on GNN performance is conducted, designing an Active Learning strategy to select the most informative data among those available. Building on these results, the second part of the thesis addresses a critical challenge often faced by ML models: the lack of guaranteed physical consistency in their predictions. To ensure that the GNNs not only minimize errors but also produce physically valid results, this part integrates physical constraints directly into the GNN training process. By embedding key fluid mechanics principles into the machine learning framework, the model produces predictions that are both reliable and consistent with the underlying physical laws, enhancing its applicability to real-world problems. In the third part, the thesis demonstrates the application of GNNs to optimize fluid dynamics systems, with a particular focus on wind turbine design. Here, GNNs are employed as surrogate models, enabling rapid predictions of various design configurations without the need for performing a full CFD simulation at each iteration. This approach significantly accelerates the design process and demonstrates the potential of ML-driven optimization in CFD workflows, allowing for more efficient exploration of design spaces and faster convergence toward optimal solutions. On the methodology side, the thesis introduces a custom GNN architecture specifically tailored for CFD applications. Unlike traditional neural networks, GNNs are inherently capable of handling unstructured mesh data, which is common in fluid mechanics problems involving irregular geometries and complex flow domains. To this end, the thesis presents a two-fold interface between Finite Element Method (FEM) solvers and the GNN architecture. This interface transforms FEM vector fields into numerical tensors that can be efficiently processed by the neural network, allowing data exchange between the simulation environment and the learning model
7
Doriat, Aurélien. "Caractérisation des couplages aéro-thermo-mécaniques lors d’un vieillissement par thermo-oxydation de composites à matrice polymère soumis à un écoulement rapide et chauffé." Electronic Thesis or Diss., Chasseneuil-du-Poitou, Ecole nationale supérieure de mécanique et d'aérotechnique, 2024. http://www.theses.fr/2024ESMA0018.
Full textAbstract:
Les matériaux composites à matrice organique renforcés de fibres de carbone (CFRP) sont largement utilisés dans les structures aéronautiques froides. Dans les applications de moteurs aéronautiques, comme les aubes de FAN, ces matériaux peuvent être soumis à des conditions environnementales particulièrement sévères, avec des températures pouvant atteindre 120°C et une vitesse d’écoulement proche de Mach 1.Il est bien établi que les polymères époxy sont sujets à des phénomènes de thermo-oxydation lorsqu’ils sont exposés à des températures élevées. Ce phénomène implique la diffusion et la réaction de l’oxygène au sein du polymère, entraînant des changements de couleur, une antiplasticisation du matériau, une fragilisation. Jusqu’à présent, les essais de vieillissement ont été principalement effectués dans des fours à air statique, fournissant une compréhension détaillée du phénomène dans ces conditions. Cependant, l’impact de l’écoulement d’air sur la thermo-oxydation reste à explorer. Cette étude vise ainsi à approfondir la compréhension du couplage entre l’écoulement d’air et la dégradation du matériau par thermo-oxydation. Des échantillons ont été vieillis dans un four sous air à pression atmosphérique et dans la soufflerie BATH, adaptée pour ces essais et capable de générer un écoulement d’air à plus de 150 ◦C et Mach 1, reproduisant ainsi les conditions d’usage les plus sévère rencontrées dans des moteurs d’avion. Cette comparaison entre essai en four et soufflerie a montré une accélération du vieillissement en soufflerie. Pour obtenir ce résultat, une technique expérimentale basée sur le changement de couleur induit par l’oxydation a été développée et utilisée. Cette technique a été validée avec des essais d’indentation. Avec cette meilleure compréhension de l’accélération du vieillissement, un modèle couplé entre l’écoulement, la chimie de l’oxydation et les changements de propriétés mécaniques a été mis en place afin de mieux comprendre les mécanismes à l’interface. Cette modélisation comprends trois étapes. Les champs de pression et de température à la surface de l’échantillon ont été calculés par simulation fluide moyennée (RANS). Puis, un modèle mécanistique a été utilisé décrivant les réactions chimiques lors de l’oxydation. Enfin, sur la base des mesures de couleur, un réseau de neurones informé par la physique (PINN) a été mis en place pour coupler les quantités chimiques aux propriétés mécaniques<br>Carbon fiber-reinforced polymer matrix composites (CFRP) are widely used in cold aeronautical structures. In aeronautical engine applications, such as fan blades, these materials can be subjected to particularly severe environmental conditions, with temperatures reaching up to 120 ◦C and airflow speeds close to Mach 1. It is well established that epoxy polymers are prone to thermo-oxidation phenomena when exposed to high temperatures.This phenomenon involves the diffusion and reaction of oxygen within the polymer, leading to color changes, antiplasticization of the material, and embrittlement. Until now, aging tests have been mainly conducted in static air ovens, providing a detailed understanding of the phenomenon under these conditions. However, the impact of airflow on thermo-oxidation remains to be explored.This study thus aims to deepen the understanding of the coupling between airflow and material degradation due to thermo-oxidation.Samples were aged in an oven under air at atmospheric pressure and in the BATH wind tunnel, adapted for these tests and capable of generating an airflow at over 150 ◦C and Mach 1, thereby reproducing the most severe usage conditions encountered in aircraft engines. This comparison between oven and wind tunnel tests showed an acceleration of aging in the wind tunnel. To achieve this result, an experimental technique based on the color change induced by oxidation was developed and used. This technique was validated with indentation tests. With this improved understanding of the accelerated aging, a coupled model between the airflow, oxidation chemistry, and changes in mechanical properties was established to better understand the interfacial mechanisms. This modeling comprises three steps. The pressure and temperature fields at the sample surface were calculated using Reynolds-Averaged Navier-Stokes (RANS) fluid simulations. Then, a mechanistic model was used to describe the chemical reactions during oxidation. Finally, based on thecolor measurements, a physics-informed neural network (PINN) was implemented to couple the chemical quantities to the mechanical properties
8
(8828960), Sukirt. "Physics Informed Neural Networks for Engineering Systems." Thesis, 2020.
Find full textAbstract:
<div>This thesis explores the application of deep learning techniques to problems in fluid mechanics, with particular focus on physics informed neural networks. Physics</div><div>informed neural networks leverage the information gathered over centuries in the</div><div>form of physical laws mathematically represented in the form of partial differential</div><div>equations to make up for the dearth of data associated with engineering and physi-</div><div>cal systems. To demonstrate the capability of physics informed neural networks, an</div><div>inverse and a forward problem are considered. The inverse problem involves discov-</div><div>ering a spatially varying concentration ?field from the observations of concentration</div><div>of a passive scalar. A forward problem involving conjugate heat transfer is solved as</div><div>well, where the boundary conditions on velocity and temperature are used to discover</div><div>the velocity, pressure and temperature ?fields in the entire domain. The predictions of</div><div>the physics informed neural networks are compared against simulated data generated</div><div>using OpenFOAM.</div>
9
Yadav, Sangeeta. "Data Driven Stabilization Schemes for Singularly Perturbed Differential Equations." Thesis, 2023. https://etd.iisc.ac.in/handle/2005/6095.
Full textAbstract:
This thesis presents a novel way of leveraging Artificial Neural Network (ANN) to aid
conventional numerical techniques for solving Singularly Perturbed Differential Equation (SPDE). SPDEs are challenging to solve with conventional numerical techniques
such as Finite Element Methods (FEM) due to the presence of boundary and interior
layers. Often the standard numerical solution shows spurious oscillations in the vicinity
of these layers. Stabilization techniques are often employed to eliminate these spurious
oscillations in the numerical solution. The accuracy of the stabilization technique depends on a user-chosen stabilization parameter whose optimal value is challenging to
find. A few formulas for the stabilization parameter exist in the literature, but none extends well for high-dimensional and complex problems. In order to solve this challenge,
we have developed the following ANN-based techniques for predicting this stabilization
parameter:
1) SPDE-Net:
As a proof of concept, we have developed an ANN called SPDE-Net for one-dimensional
SPDEs. In the proposed method, we predict the stabilization parameter for the Streamline
Upwind Petrov Galerkin (SUPG) stabilization technique. The prediction task is modelled as a regression problem using equation coefficients and domain parameters as
inputs to the neural network. Three training strategies have been proposed, i.e. supervised learning, L
2-Error minimization (global) and L2-Error minimization (local).
The proposed method outperforms existing state-of-the-art ANN-based partial differential equations (PDE) solvers, such as Physics Informed Neural Networks (PINNs).
2) AI-stab FEM
With an aim for extending SPDE-Net for two-dimensional problems, we have also developed an optimization scheme using another Neural Network called AI-stab FEM and
showed its utility in solving higher-dimensional problems. Unlike SPDE-Net, it minimizes the equation residual along with the crosswind derivative term and can be classified as an unsupervised method. We have shown that the proposed approach yields
stable solutions for several two-dimensional benchmark problems while being more accurate than other contemporary ANN-based PDE solvers such as PINNs and Variational
Neural Networks for the Solution of Partial Differential Equations (VarNet)
3) SPDE-ConvNet
In the last phase of the thesis, we attempt to predict a cell-wise stabilization parameter to
treat the interior/boundary layer regions adequately by developing an oscillations-aware
neural network. We present SPDE-ConvNet, Convolutional Neural Network (CNN),
for predicting the local (cell-wise) stabilization parameter. For the network training,
we feed the gradient of the Galerkin solution, which is an indirect metric for representing oscillations in the numerical solution, along with the equation coefficients, to
the network. It obtains a cell-wise stabilization parameter while sharing the network
parameters among all the cells for an equation. Similar to AI-stab FEM, this technique
outperforms PINNs and VarNet.
We conclude the thesis with suggestions for future work that can leverage our current
understanding of data-driven stabilization schemes for SPDEs to develop and improve
the next-generation neural network-based numerical solvers for SPDEs.
10
(10141679), Haoyang Zheng. "Quantifying implicit and explicit constraints on physics-informed neural processes." Thesis, 2021.
Find full textAbstract:
<p>Due to strong interactions among various phases and among the phases
and fluid motions, multiphase flows (MPFs) are so complex that lots of efforts
have to be paid to predict its sequential
patterns of phases and motions. The present paper applies the physical
constraints inherent in MPFs and enforces them to a physics-informed neural
network (PINN) model either explicitly or implicitly, depending on the type of
constraints. To predict the unobserved order
parameters (OPs) (which locate the phases) in the future steps, the conditional
neural processes (CNPs) with long short-term memory (LSTM, combined as CNPLSTM)
are applied to quickly infer the dynamics of the phases after encoding only a
few observations. After that, the multiphase consistent and conservative
boundedness mapping (MCBOM) algorithm is implemented the correction the predicted OPs from
CNP-LSTM so that the mass conservation, the summation of the volume fractions of
the phases being unity, the consistency of reduction, and the boundedness of the OPs are
strictly satisfied. Next, the density of the
fluid mixture is computed from the corrected OPs. The observed velocity and
density of the fluid mixture then encode in a physics-informed conditional
neural processes and long short-term memory (PICNP-LSTM) where the constraint
of momentum conservation is included in the loss function. Finally, the unobserved
velocity in future steps is predicted from PICNP-LSTM. The proposed physics-informed neural
processes (PINPs) model (CNP-LSTM-MCBOM-PICNP-LSTM) for MPFs avoids unphysical behaviors
of the OPs, accelerates the convergence, and requires fewer data. The proposed
model successfully predicts several canonical MPF problems, i.e., the horizontal shear
layer (HSL) and dam break (DB) problems, and its performances are validated.</p>
Book chapters on the topic "Physics-Informed neural network"
1
Beniwal, Kirti, and Vivek Kumar. "Gradient-Based Physics-Informed Neural Network." In Third Congress on Intelligent Systems. Springer Nature Singapore, 2023. http://dx.doi.org/10.1007/978-981-19-9379-4_54.
Full text
2
Madenci, Erdogan, Pranesh Roy, and Deepak Behera. "Peridynamics for Physics Informed Neural Network." In Advances in Peridynamics. Springer International Publishing, 2022. http://dx.doi.org/10.1007/978-3-030-97858-7_16.
Full text
3
Li, Yao, Yuanxun Xu, Shengzhu Shi, and Boying Wu. "Adversarial Adaptive Sampling for Physics-Informed Neural Network." In Lecture Notes in Networks and Systems. Springer Nature Switzerland, 2024. https://doi.org/10.1007/978-3-031-77688-5_41.
Full text
4
Ho, Ya-Chi, Chia-Lin Chang, Tai-Te Lee, and Yu-Hui Huang. "Physics-Informed Neural Network for Shock Absorber Design." In Communications in Computer and Information Science. Springer Nature Singapore, 2025. https://doi.org/10.1007/978-981-96-4596-1_19.
Full text
5
Naveen Raj, R. "Physics Informed Neural Network for Solution of Duffing Oscillators." In Springer Proceedings in Physics. Springer Nature Switzerland, 2024. https://doi.org/10.1007/978-3-031-69146-1_14.
Full text
6
Dhamirah Mohamad, Najwa Zawani, Akram Yousif, Nasiha Athira Binti Shaari, et al. "Heat Transfer Modelling with Physics-Informed Neural Network (PINN)." In Studies in Systems, Decision and Control. Springer International Publishing, 2022. http://dx.doi.org/10.1007/978-3-031-04028-3_3.
Full text
7
Mahesh, Ragini Bal, Jorge Leandro, and Qing Lin. "Physics Informed Neural Network for Spatial-Temporal Flood Forecasting." In Lecture Notes in Civil Engineering. Springer Singapore, 2021. http://dx.doi.org/10.1007/978-981-16-5501-2_7.
Full text
8
Oh, Dong Keun. "Pure Physics-Informed Echo State Network of ODE Solution Replicator." In Artificial Neural Networks and Machine Learning – ICANN 2023. Springer Nature Switzerland, 2023. http://dx.doi.org/10.1007/978-3-031-44201-8_19.
Full text
9
Wu, Mingyu, Yafei Wang, Yichen Zhang, and Zexing Li. "Physics-Informed Neural Network for Mining Truck Suspension Parameters Identification." In Lecture Notes in Mechanical Engineering. Springer Nature Switzerland, 2024. http://dx.doi.org/10.1007/978-3-031-70392-8_94.
Full textAbstract:
AbstractMining truck suspensions are prone to performance degradation under complex external excitation of the mining area, leading to high safety risks and maintenance costs. However, the lack of unsprung kinematic information and harsh operating environments lead to inadequate accuracy of current physical models. On the other hand, data-driven methods partially address the issue of incomplete information, but suffer from the absence of interpretability and generalization. To address these challenges, this paper introduces a Physics-Informed Neural Network (PINN) for precise suspension characteristic identification of mining trucks. Specifically, the physical model of the longitudinal-vertical dynamics of the mining truck is established. Then, based on the model, baseline values of suspension parameters are regressed through the instrumental variable method. Therefore, the hybrid modeling architecture is established to precisely identify suspension parameters by utilizing a recurrent neural network. Under this architecture, the states of the mining truck can be effectively updated. Real truck experiments demonstrate the proposed hybrid model outperforms traditional physical and data-driven models in estimating suspension nonlinear parameters and truck dynamic characteristics under typical longitudinal motions.
10
Ibrahim, Abdul Qadir, Sebastian Götschel, and Daniel Ruprecht. "Parareal with a Physics-Informed Neural Network as Coarse Propagator." In Euro-Par 2023: Parallel Processing. Springer Nature Switzerland, 2023. http://dx.doi.org/10.1007/978-3-031-39698-4_44.
Full textAbstract:
AbstractParallel-in-time algorithms provide an additional layer of concurrency for the numerical integration of models based on time-dependent differential equations. Methods like Parareal, which parallelize across multiple time steps, rely on a computationally cheap and coarse integrator to propagate information forward in time, while a parallelizable expensive fine propagator provides accuracy. Typically, the coarse method is a numerical integrator using lower resolution, reduced order or a simplified model. Our paper proposes to use a physics-informed neural network (PINN) instead. We demonstrate for the Black-Scholes equation, a partial differential equation from computational finance, that Parareal with a PINN coarse propagator provides better speedup than a numerical coarse propagator. Training and evaluating a neural network are both tasks whose computing patterns are well suited for GPUs. By contrast, mesh-based algorithms with their low computational intensity struggle to perform well. We show that moving the coarse propagator PINN to a GPU while running the numerical fine propagator on the CPU further improves Parareal’s single-node performance. This suggests that integrating machine learning techniques into parallel-in-time integration methods and exploiting their differences in computing patterns might offer a way to better utilize heterogeneous architectures.
Conference papers on the topic "Physics-Informed neural network"
1
Miao, Yuyang, Haolin Li, and Danilo Mandic. "GPINN: Physics-Informed Neural Network with Graph Embedding." In 2024 International Joint Conference on Neural Networks (IJCNN). IEEE, 2024. http://dx.doi.org/10.1109/ijcnn60899.2024.10651053.
Full text
2
Jahanbakhsh, Amirmohammad, Rojan Firouznia, Sina Nazifi, and Hadi Ghasemi. "Physics-Informed Neural Network on Thin Film Evaporation." In 2024 23rd IEEE Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems (ITherm). IEEE, 2024. http://dx.doi.org/10.1109/itherm55375.2024.10709585.
Full text
3
Bakirtzis, Stefanos, Marco Fiore, and Ian Wassell. "Towards Physics-Informed Graph Neural Network-based Computational Electromagnetics." In 2024 IEEE International Symposium on Antennas and Propagation and INC/USNC‐URSI Radio Science Meeting (AP-S/INC-USNC-URSI). IEEE, 2024. http://dx.doi.org/10.1109/ap-s/inc-usnc-ursi52054.2024.10686000.
Full text
4
Glosemeyer, Tom, Julian Lich, Robert Kuschmierz, and Juergen Czarske. "3D Diffuser Encoded Imaging and Physics-Informed Neural Network Reconstruction." In Frontiers in Optics. Optica Publishing Group, 2024. https://doi.org/10.1364/fio.2024.fw6d.1.
Full textAbstract:
Minimally invasive fiber endoscopy offers a high potential for biomedical imaging applications. By utilizing a diffuser for encoding and a coherent fiber bundle in conjunction with neural networks for reconstruction, single-shot 3D imaging enabled. Full-text article not available; see video presentation
5
Serrano, Gil, Marcelo Jacinto, José Ribeiro-Gomes, et al. "Physics-Informed Neural Network for Multirotor Slung Load Systems Modeling." In 2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2024. http://dx.doi.org/10.1109/icra57147.2024.10610582.
Full text
6
Zhao, Michael Yong, Olzhas Mukhmetov, Aigerim Mashekova, et al. "Application of Physics Informed Neural Network for Breast Cancer Detection." In 2024 9th International Conference on Automation, Control and Robotics Engineering (CACRE). IEEE, 2024. http://dx.doi.org/10.1109/cacre62362.2024.10635033.
Full text
7
Eeckhout, Victor, Hossein Fani, Md Umar Hashmi, and Geert Deconinck. "Improved Physics-Informed Neural Network based AC Power Flow for Distribution Networks." In 2024 IEEE PES Innovative Smart Grid Technologies Europe (ISGT EUROPE). IEEE, 2024. https://doi.org/10.1109/isgteurope62998.2024.10863674.
Full text
8
Chen, Likun, Xuzhu Dong, Yifan Wang, Wei Sun, Bo Wang, and Gareth Harrison. "Physics-Informed Neural Network for Microgrid Forward/Inverse Ordinary Differential Equations." In 2024 IEEE Power & Energy Society General Meeting (PESGM). IEEE, 2024. http://dx.doi.org/10.1109/pesgm51994.2024.10688678.
Full text
9
Zhang, Qian, Yuan Sui, Stefan Rothe, and Jürgen W. Czarske. "Learning to decompose multimode fibers using a physics-informed neural network." In Emerging Topics in Artificial Intelligence (ETAI) 2024, edited by Giovanni Volpe, Joana B. Pereira, Daniel Brunner, and Aydogan Ozcan. SPIE, 2024. http://dx.doi.org/10.1117/12.3027588.
Full text
10
Luan, Xinmeng, Marco Olivieri, Mirco Pezzoli, Fabio Antonacci, and Augusto Sarti. "Complex - Valued Physics-Informed Neural Network for Near-Field Acoustic Holography." In 2024 32nd European Signal Processing Conference (EUSIPCO). IEEE, 2024. http://dx.doi.org/10.23919/eusipco63174.2024.10715295.
Full textReports on the topic "Physics-Informed neural network"
1
Ellis, Kai, Nilanjan Banerjee, and Christopher Pierce. Modeling a Thermionic Electron Source Using a Physics-Informed Neural Network. Office of Scientific and Technical Information (OSTI), 2023. http://dx.doi.org/10.2172/2008057.
Full text
2
Pettit, Chris, and D. Wilson. A physics-informed neural network for sound propagation in the atmospheric boundary layer. Engineer Research and Development Center (U.S.), 2021. http://dx.doi.org/10.21079/11681/41034.
Full textAbstract:
We describe what we believe is the first effort to develop a physics-informed neural network (PINN) to predict sound propagation through the atmospheric boundary layer. PINN is a recent innovation in the application of deep learning to simulate physics. The motivation is to combine the strengths of data-driven models and physics models, thereby producing a regularized surrogate model using less data than a purely data-driven model. In a PINN, the data-driven loss function is augmented with penalty terms for deviations from the underlying physics, e.g., a governing equation or a boundary condition. Training data are obtained from Crank-Nicholson solutions of the parabolic equation with homogeneous ground impedance and Monin-Obukhov similarity theory for the effective sound speed in the moving atmosphere. Training data are random samples from an ensemble of solutions for combinations of parameters governing the impedance and the effective sound speed. PINN output is processed to produce realizations of transmission loss that look much like the Crank-Nicholson solutions. We describe the framework for implementing PINN for outdoor sound, and we outline practical matters related to network architecture, the size of the training set, the physics-informed loss function, and challenge of managing the spatial complexity of the complex pressure.
3
Keidar, Michael, and Li Lin. Generative Physics-Informed Neural Network Solving Multi-Scale and Multi-Phase Plasma Chemical Flow Field. Office of Scientific and Technical Information (OSTI), 2024. https://doi.org/10.2172/2478929.
Full text
4
Wells, Daniel, Benjamin Baker, and Kristine Pankow. The Feasibility of Incorporating a 3D Velocity Model Into Earthquake Location Around Salt Lake City, UT Using a Physics Informed Neural Network. Office of Scientific and Technical Information (OSTI), 2023. http://dx.doi.org/10.2172/2430497.
Full text
5
Nadiga, Balasubramanya, and Robert Lowrie. Physics Informed Neural Networks as Computational Physics Emulators. Office of Scientific and Technical Information (OSTI), 2023. http://dx.doi.org/10.2172/1985825.
Full text
6
Guan, Jiajing, Sophia Bragdon, and Jay Clausen. Predicting soil moisture content using Physics-Informed Neural Networks (PINNs). Engineer Research and Development Center (U.S.), 2024. http://dx.doi.org/10.21079/11681/48794.
Full textAbstract:
Environmental conditions such as the near-surface soil moisture content are valuable information in object detection problems. However, such information is generally unobtainable at the necessary scale without active sensing. Richards’ equation is a partial differential equation (PDE) that describes the infiltration process of unsaturated soil. Solving the Richards’ equation yields information about the volumetric soil moisture content, hydraulic conductivity, and capillary pressure head. However, Richards’ equation is difficult to approximate due to its nonlinearity. Numerical solvers such as finite difference method (FDM) and finite element method (FEM) are conventional in approximating solutions to Richards’ equation. But such numerical solvers are time-consuming when used in real-time. Physics-informed neural networks (PINNs) are neural networks relying on physical equations in approximating solutions. Once trained, these networks can output approximations in a speedy manner. Thus, PINNs have attracted massive attention in the numerical PDE community. This project aims to apply PINNs to the Richards’ equation to predict underground soil moisture content under known precipitation data.
7
D'Elia, Marta, Michael L. Parks, Guofei Pang, and George Karniadakis. nPINNs: nonlocal Physics-Informed Neural Networks for a parametrized nonlocal universal Laplacian operator. Algorithms and Applications. Office of Scientific and Technical Information (OSTI), 2020. http://dx.doi.org/10.2172/1614899.
Full text
8
Bailey Bond, Robert, Pu Ren, James Fong, Hao Sun, and Jerome F. Hajjar. Physics-informed Machine Learning Framework for Seismic Fragility Analysis of Steel Structures. Northeastern University, 2024. http://dx.doi.org/10.17760/d20680141.
Full textAbstract:
The seismic assessment of structures is a critical step to increase community resilience under earthquake hazards. This research aims to develop a Physics-reinforced Machine Learning (PrML) paradigm for metamodeling of nonlinear structures under seismic hazards using artificial intelligence. Structural metamodeling, a reduced-fidelity surrogate model to a more complex structural model, enables more efficient performance-based design and analysis, optimizing structural designs and ease the computational effort for reliability fragility analysis, leading to globally efficient designs while maintaining required levels of accuracy. The growing availability of high-performance computing has improved this analysis by providing the ability to evaluate higher order numerical models. However, more complex models of the seismic response of various civil structures demand increasing amounts of computing power. In addition, computational cost greatly increases with numerous iterations to account for optimization and stochastic loading (e.g., Monte Carlo simulations or Incremental Dynamic Analysis). To address the large computational burden, simpler models are desired for seismic assessment with fragility analysis. Physics reinforced Machine Learning integrates physics knowledge (e.g., scientific principles, laws of physics) into the traditional machine learning architectures, offering physically bounded, interpretable models that require less data than traditional methods. This research introduces a PrML framework to develop fragility curves using the combination of neural networks of domain knowledge. The first aim involves clustering and selecting ground motions for nonlinear response analysis of archetype buildings, ensuring that selected ground motions will include as few ground motions as possible while still expressing all the key representative events the structure will probabilistically experience in its lifetime. The second aim constructs structural PrML metamodels to capture the nonlinear behavior of these buildings utilizing the nonlinear Equation of Motion (EOM). Embedding physical principles, like the general form of the EOM, into the learning process will inform the system to stay within known physical bounds, resulting in interpretable results, robust inferencing, and the capability of dealing with incomplete and scarce data. The third and final aim applies the metamodels to probabilistic seismic response prediction, fragility analysis, and seismic performance factor development. The efficiency and accuracy of this approach are evaluated against existing physics-based fragility analysis methods.
9
SECOND-ORDER ANALYSIS OF BEAM-COLUMNS BY MACHINE LEARNING-BASED STRUCTURAL ANALYSIS THROUGH PHYSICS-INFORMED NEURAL NETWORKS. The Hong Kong Institute of Steel Construction, 2023. http://dx.doi.org/10.18057/ijasc.2023.19.4.10.
Full textAbstract:
The second-order analysis of slender steel members could be challenging, especially when large deflection is involved. This paper proposes a novel machine learning-based structural analysis (MLSA) method for second-order analysis of beam-columns, which could be a promising alternative to the prevailing solutions using over-simplified analytical equations or traditional finite-element-based methods. The effectiveness of the conventional machine learning method heavily depends on both the qualitative and the quantitative of the provided data. However, such data are typically scarce and expensive to obtain in structural engineering practices. To address this problem, a new and explainable machine learning-based method, named Physics-informed Neural Networks (PINN), is employed, where the physical information will be utilized to orientate the learning process to create a self-supervised learning procedure, making it possible to train the neural network with few or even no predefined datasets to achieve an accurate approximation. This research extends the PINN method to the problems of second-order analysis of steel beam-columns. Detailed derivations of the governing equations, as well as the essential physical information for the training process, are given. The PINN framework and the training procedure are provided, where an adaptive loss weight control algorithm and the transfer learning technic are adopted to improve numerical efficiency. The practicability and accuracy of which are validated by four sets of verification examples.