Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “GANs”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 577 records · Page 32
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Synthetic Data Generation Using Machine Learning

Robust machine learning techniques for image analysis require a substantial amount of data to yield confident results. In the nuclear domain, data scarcity is a substantial challenge because there are so few facilities worldwide. This research focuses on being able alleviate the data scarcity problem by generating synthetic data to bridge the gap between large and small datasets. This work achieves that goal using a Generative Adversarial Network (GAN) architectural approach, by training a model on real-world data and expands that small dataset through synthetic data amendments. Model performance is impacted by the size of the real-world dataset and the number of training epochs utilized. This means that 1) It is important to develop your GAN to be optimized with the specific data type, and 2) approaches taken when training the GAN should be specialized to encompass important aspects of the dataset that it is generating. By taking a step to improve dataset sizes in this way, the gap between models trained by parties with significant amount of data and those without access to large data, closes, allowing for robust analyses of satellite imagery for nuclear domain applications.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2024

Engagement: Hyperparameter Optimization of Generative Adversarial Network Models for High-Energy Physics Simulations

We present our SciDAC FASTMath-HEP partnership results for tuning generative adversarial models (GANs) for high energy physics applications. The GANs are used in hybrid simulations to accelerate otherwise time-consuming computations. We optimize for both, prediction accuracy and variability with the goal to find GAN architectures that are reliable and robust.

high energy physics↗
DOE OSTI2020

GaN/InGaN Blue Light-Emitting Diodes on Polycrystalline Molybdenum Metal Foils by Ion Beam-Assisted Deposition

Light-emitting diode (LED) arrays fabricated on a polycrystalline metal substrate are demonstrated using a novel technique that enables the growth of epitaxial metal-organic chemical vapor deposition (MOCVD) GaN layers on non-single-crystal substrates. Epitaxial GaN is deposited directly on metal foil using an intermediate ion beam-assisted deposition (IBAD) aligned layer. For a single 170 μm-diameter LED on the metal foil, electroluminescence (EL) spectrum shows a peak wavelength of ≈452 nm and a full width at half maximum (FWHM) of ≈24 nm. The current–voltage (I –V ) characteristics show a turn-on voltage of 3.7 V, a series resistance of 10 Ω. LEDs on metal show a relative external quantum efficiency (EQE) that is roughly 3× lower than that of similar LEDs fabricated on a sapphire substrate. Finally, InGaN LEDs on large-area non-single-crystal substrates such as metal foils enable large-area manufacturing, reducing production cost, and opening the door for new applications in lighting and displays.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Learning functional priors and posteriors from data and physics

In this work, we develop a new Bayesian framework based on deep neural networks to be able to extrapolate in space-time using historical data and to quantify uncertainties arising from both noisy and gappy data in physical problems. Specifically, the proposed approach has two stages: (1) prior learning and (2) posterior estimation. At the first stage, we employ the physics-informed Generative Adversarial Networks (PI-GAN) to learn a functional prior either from a prescribed function distribution, e.g., Gaussian process, or from historical data and physics. At the second stage, we employ the Hamiltonian Monte Carlo (HMC) method to estimate the posterior in the latent space of PI-GANs. In addition, we use two different approaches to encode the physics: (1) automatic differentiation, used in the physicsinformed neural networks (PINNs) for scenarios with explicitly known partial differential equations (PDEs), and (2) operator regression using the deep operator network (DeepONet) for PDE-agnostic scenarios. We then test the proposed method for (1) meta-learning for one-dimensional regression, and forward/inverse PDE problems (combined with PINNs); (2) PDE-agnostic physical problems (combined with DeepONet), e.g., fractional diffusion as well as saturated stochastic (100-dimensional) flows in heterogeneous porous media; and (3) spatial-temporal regression problems, i.e., inference of a marine riser displacement field using experimental data from the Norwegian Deepwater Programme (NDP). The results demonstrate that the proposed approach can provide accurate predictions as well as uncertainty quantification given very limited scattered and noisy data, since historical data could be available to provide informative priors. In summary, the proposed method is capable of learning flexible functional priors, e.g., both Gaussian and non-Gaussian process, and can be readily extended to big data problems by enabling mini-batch training using stochastic HMC or normalizing flows since the latent space is generally characterized as low dimensional.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2020

Hot-Hole versus Hot-Electron Transport at Cu/GaN Heterojunction Interfaces

Among all plasmonic metals, copper (Cu) has the greatest potential for realizing optoelectronic and photochemical hot-carrier devices, thanks to its CMOS compatibility and outstanding catalytic properties. Yet, relative to gold (Au) or silver (Ag), Cu has rarely been studied and the fundamental properties of its photoexcited hot carriers are not well understood. Here, we demonstrate that Cu nanoantennas on p-type gallium nitride (p-GaN) enable hot-hole-driven photodetection across the visible spectrum. Importantly, we combine experimental measurements of the internal quantum efficiency (IQE) with ab initio theoretical modeling to clarify the competing roles of hot-carrier energy and mean-free path on the performance of hot-hole devices above and below the interband threshold of the metal. We also examine Cu-based plasmonic photodetectors on corresponding n-type GaN substrates that operate via the collection of hot electrons. By comparing hot hole and hot electron photodetectors that employ the same metal/semiconductor interface (Cu/GaN), we further elucidate the relative advantages and limitations of these complementary plasmonic systems. In particular, we find that harnessing hot holes with p-type semiconductors is a promising strategy for plasmon-driven photodetection across the visible and ultraviolet regimes. Given the technological relevance of Cu and the fundamental insights provided by our combined experimental and theoretical approach, we anticipate that our studies will have a broad impact on the design of hot-carrier optoelectronic devices and plasmon-driven photocatalytic systems.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2023

A Smart Silicon Carbide Power Module With Pulse Width Modulation Over Wi-Fi and Wireless Power Transfer-Enabled Gate Driver, Featuring Onboard State of Health Estimator and High-Voltage Scaling Capabilities

A wide range of utility applications require controllable switches with features such as high-voltage blocking and high-current carrying capacity, especially at high pulse width modulation (PWM) frequency. Low- and medium-voltage utility applications such as motor drives and flexible AC transmission systems as well as solid state transformers could also benefit from a low-cost high-voltage switching module. Wide-bandgap semiconductors such as silicon carbide (SiC) and gallium nitride (GaN) metal oxide semiconductor field effect transistors (MOSFETs) are considered to be the present and next-generation device choices, although they have limitations. For relatively high-voltage applications with demanding thermal management, SiC is still the only choice, and GaN dominates the low-voltage regime. This manuscript proposes a new half-bridge power MOSFET module that is suitable for conventional H-bridge of multilevel configurations used in high-voltage applications. Constructed from bare SiC dies, this half-bridge module takes advantage of (1) optimized MOSFET placement inside the module, (2) customized heat exchanger, manifold, and cooling, (3) integrated gate driver module with pulse width modulation (PWM) over wi-fi to eliminate the need for low-voltage signals, (4) wireless power transfer (WPT)-enabled gate driver and other ancillary circuits, (5) and the option to incorporate an onboard state-of-health (SOH) estimator module. The entire architecture has been designed and built at the National Renewable Energy Laboratory (NREL) in Golden, CO.

Ga2O3 devices↗
DOE OSTI2021

EMI Mitigation of a Ćuk-Based Power-Electronic System Using Switching-Sequence-Based Control

Switching-sequence-based control (SBC) laws when designed based on topological switching behavior can have positive effects on slow- and fast-scale dynamics of a power-electronic system (PES). The slow-scale control can encompass fast PES state regulation and tracking, based on predefined objective, while fastscale control can address differential-mode (DM) and commonmode (CM) spectral-peak energy associated with PES switching operation. Such control laws may offer enhanced programmability to conventional PES design where bulky electromagnetic interference (EMI) filters have been traditionally used to reduce EMI of switching power converters to meet EMI regulatory standards. An EMI filter is always a less programmable solution since it is usually designed for the worst-case EMI mitigation and usually overkill for a PES operating under reduced load condition. The control scheme outlined in this article offers EMI mitigation across wide operating regions without compromising PES regulation. Moreover, it does so by use of switching sequences that guarantee the reachability of the PES dynamics using an advanced Lyapunov-function-based approach. SBC is a powerful tool to generate control actions for a PES based on multivariate PES state constraints. Hence, contemporary EMI regulatory standards are used as constraints in the SBC formulation to operate the PES under wide operating regime while autonomously mitigating the EMI levels. The work may be of paramount importance for operating the ultra-fast-transition recent wide-bandgap semiconductor devices like GaN–FET and SiC MOSFET under higher power with increasing switching frequencies, which is usually desirable for increased power density and reduced switching losses. Here, a hardware Cuk–PES operated ´ with GaN–FETs is fabricated and is used for case illustration. It is shown by experimental results how SBC mitigate DM and CM EMI noise of the PES while maintaining regulation even for the higher order nonminimum phase PES, while reducing sensor requirements using state observer derived from the switching model of the PES.

42 ENGINEERING↗
NASA NTRS1994

(abstract) Transmission Electron Microscopy of Al(sub x)Ga(sub 1-x)N/SiC Multilayer Structures

The potential of wide-band-gap III-V nitrides as solar-blind ultraviolet sensors and light emitters has prompted an increasing amount of work recently, including the fabrication of the first UV sensors from as-deposited single crystal GaN. We have used high resolution transmission electron microscopy (TEM) to study the microstructure of two novel developments of wide-band-gap III-V nitrides: the growth of ultra-short period GaN/AlN super lattices (digital alloys); and the incorporation of SiC layers into Al(sub x)Ga(sub 1-x)N structures. By varying the relative periods in a GaN/AlN superlattice, it should be possible to tailor the band gap of the composite to lie between the elemental values of 365 nm for GaN and 200 nm for AlN. The group IV semiconductor, SiC, has a close lattice match (< 3%) to Al(sub x)Ga(sub 1-x)N for growth on the basal plane. Demonstration of epitaxial growth for Al(sub x)Ga(sub 1-x)N/SiC multilayers could enable an extension of direct band-gap material towards the visible.

microscopy GaN AlN semiconductor SiC↗
NASA NTRS2018

Towards Informing an Intuitive Mission Planning Interface for Autonomous Multi-Asset Teams via Image Descriptions

Establishing a basis for certification of autonomous systems using trust and trustworthiness is the focus of Autonomy Teaming and TRAjectories for Complex Trusted Operational Reliability (ATTRACTOR). The Human-Machine Interface (HMI) team is working to capture and utilize the multitude of ways in which humans are already comfortable communicating mission goals and translate that into an intuitive mission planning interface. Several input/output modalities (speech/audio, typing/text, touch, and gesture) are being considered and investigated in the context human-machine teaming for the ATTRACTOR design reference mission (DRM) of Search and Rescue or (more generally) intelligence, surveillance, and reconnaissance (ISR). The first of these investigations, the Human Informed Natural-language GANs Evaluation (HINGE) data collection effort, is aimed at building an image description database to train a Generative Adversarial Network (GAN). In addition to building an image description database, the HMI team was interested if, and how, modality (spoken vs. written) affects different aspects of the image description given. The results will be analyzed to better inform the designing of an interface for mission planning.

Generative Adversarial Network (GAN)↗
NASA NTRS2020

Update on Wide Bandgap (WBG) Device Radiation Hardness Assurance

This presentation provides an overview of recent wide bandgap device radiation hardness assurance activities at NASA and via NASA-funded grant vehicles. Devices include enhancement mode power GaN HEMTs, RF GaN HEMTs, and SiC MOSFETs and JBS diodes.

wide bandgap↗
DOE OSTI2025

Domain Nucleation and Growth in an Epitaxially Grown Wurtzite Ferroelectric

Ferroelectric domain nucleation and growth in epitaxial (Al, B, Sc)N films grown on n-GaN substrates are explored using a combination of ferroelectric property measurements and scanning transmission electron microscopy, including novel in situ switching studies. The films are electrically switched to nitrogen-polar (N-polar) and metal-polar (M-polar) configurations, attaining a remanent polarization of 120 µC cm −2 with coercive fields of ≈6 MV cm −1 . In the initial switching cycle, the ferroelectric domains nucleate near the bottom n-GaN electrode and develop domain walls with zigzag morphologies, while residual “dead layers” that do not switch from the as-deposited orientation persist at the top and bottom electrodes. The in situ microscopy experiments reveal that domain walls propagate fastest in the lateral direction, parallel to the electrode/film interface. These findings provide insights into the domain dynamics and structural evolution of wurtzite ferroelectrics, offering implications for next-generation electronic devices.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Stable parallel training of Wasserstein conditional generative adversarial neural networks

In this work, we propose a stable, parallel approach to train Wasserstein conditional generative adversarial neural networks (W-CGANs) under the constraint of a fixed computational budget. Differently from previous distributed GANs training techniques, our approach avoids inter-process communications, reduces the risk of mode collapse and enhances scalability by using multiple generators, each one of them concurrently trained on a single data label. The use of the Wasserstein metric also reduces the risk of cycling by stabilizing the training of each generator. We illustrate the approach on the CIFAR10, CIFAR100, and ImageNet1k datasets, three standard benchmark image datasets, maintaining the original resolution of the images for each dataset. Performance is assessed in terms of scalability and final accuracy within a limited fixed computational time and computational resources. To measure accuracy, we use the inception score, the Fréchet inception distance, and image quality. An improvement in inception score and Fréchet inception distance is shown in comparison to previous results obtained by performing the parallel approach on deep convolutional conditional generative adversarial neural networks as well as an improvement of image quality of the new images created by the GANs approach. Weak scaling is attained on both datasets using up to 2000 NVIDIA V100 GPUs on the OLCF supercomputer Summit.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Microstructure Generation via Generative Adversarial Network for Heterogeneous, Topologically Complex 3D Materials

Using a large-scale, experimentally captured 3D microstructure dataset, we implement the generative adversarial network (GAN) framework to learn and generate 3D microstructures of solid oxide fuel cell electrodes. The generated microstructures are visually, statistically, and topologically realistic, with distributions of microstructural parameters, including volume fraction, particle size, surface area, tortuosity, and triple phase boundary density, being highly similar to those of the original microstructure. These results are compared and contrasted with those from an established, grain-based generation algorithm (DREAM.3D). Importantly, simulations of electrochemical performance, using a locally resolved finite element model, demonstrate that the GAN generated microstructures closely match the performance distribution of the original, while DREAM.3D leads to significant differences. Finally, the ability of the generative machine learning model to recreate microstructures with high fidelity suggests that the essence of complex microstructures may be captured and represented in a compact and manipulatable form.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Multi deep learning-based stochastic microstructure reconstruction and high-fidelity micromechanics simulation of time-dependent ceramic matrix composite response

A multi deep learning-based framework is developed for efficient, automated microstructure reconstruction and generation of stochastic representative volume elements (SRVEs) with periodic boundary conditions (PBCs) for accurate modeling of ceramic matrix composite (CMC) response. The methodology comprises a convolutional neural network coupled with regression layers to act as a vanilla regression network for semantic segmentation of the microstructure, allowing accurate characterization of the phases and their distributions at the microscale. Scanning electron microscope and confocal microscope are used to obtain C/SiNC and SiC/SiNC CMCs micrographs for vanilla regression testing. Microstructure variability in terms of fiber volume fraction and porosity are quantified through the output regression layer, ensuring accurate representation of material variability in SRVE construction. Generative adversarial network (GAN) and its variants are designed to produce high-fidelity SRVE, spanning CMCs microstructure variability space. A circular padding algorithm is developed to generate SRVEs with PBCs during training of GANs. The accuracy of the generated SRVEs is established through micromechanics simulations, where an efficient formulation of the high-fidelity generalized methods of cells (HFGMC) approach is used to compute the effective mechanical properties. Furthermore, an iterative algorithm is implemented in the HFGMC solver to simulate time-dependent deformation of SiC/SiNC subjected to creep loading conditions.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Adversarial sampling of unknown and high-dimensional conditional distributions

Many engineering problems require the prediction of realization-to-realization variability or a refined description of modeled quantities. In that case, it is necessary to sample elements from unknown high-dimensional spaces with possibly millions of degrees of freedom. While there exist methods able to sample elements from probability density functions (PDF) with known shapes, several approximations need to be made when the distribution is unknown. In this paper the sampling method, as well as the inference of the underlying distribution, are both handled with a data-driven method known as generative adversarial networks (GAN), which trains two competing neural networks to produce a network that can effectively generate samples from the training set distribution. In practice, it is often necessary to draw samples from conditional distributions. When the conditional variables are continuous, only one (if any) data point corresponding to a particular value of a conditioning variable may be available, which is not sufficient to estimate the conditional distribution. This work handles this problem using an a priori estimation of the conditional moments of a PDF. Herein, two approaches, stochastic estimation, and an external neural network are compared for computing these moments; however, any preferred method can be used. The algorithm is demonstrated in the case of the deconvolution of a filtered turbulent flow field. It is shown that all the versions of the proposed algorithm effectively sample the target conditional distribution with minimal impact on the quality of the samples compared to state-of-the-art methods. Additionally, the procedure can be used as a metric for the diversity of samples generated by a conditional GAN (cGAN) conditioned with continuous variables.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2025

Epitaxial columnar growth of strain-free antiferromagnetic Weyl semimetal Mn 3 Sn on wurtzite c -plane GaN/Al 2 O 3 (0001)

Weyl semimetal thin films with excellent crystalline quality are of great interest for antiferromagnetic spintronics. Mn 3 Sn is one Weyl semimetal with great properties and promise for exciting science and applications. It has proven very challenging, however, to grow Mn 3 Sn thin films with smooth surfaces, negligible strain, and excellent crystallinity. In this work, we discuss the successful preparation of epitaxial Mn 3 Sn (0001)-oriented thin films via molecular beam epitaxial growth on c -plane wurtzite GaN which was grown by MBE on Al 2 O 3 (0001). We present the reflection high energy electron diffraction analysis along with x-ray diffraction in order to demonstrate the crystalline quality of the film, and we give atomic models to explain the epitaxial orientation relationships between the crystal lattices of the substrate, GaN layer, and Mn 3 Sn layer. Importantly, we discuss the film lattice parameters as compared to expected values, demonstrating negligible strain both in-plane and out-of-plane . Atomic force microscopy reveals an epitaxial columnar growth mode characterized by flat-top-mesa islands, while scanning tunneling microscopy shows the atomically smooth surfaces of the mesa-top structures. Finally, Rutherford backscattering informs the stoichiometry of the film as well as the layer thicknesses.

Chiral antiferromagnetic material↗