Search NASA⌕ Search

SEARCH · Search NASA

Results for “domain transfer”

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 19 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

Scaling deep learning for material imaging with a pseudo 3D model for domain transfer

The recent introduction of deep learning methods for image processing has greatly advanced the characterization of materials using three-dimensional (3D) X-ray imaging techniques. However, deep learning models often have difficulty performing consistently across images owing to unavoidable variations in imaging conditions, which create inconsistencies even for the same material. As a result, networks must frequently be retrained for new datasets, limiting their applicability and generalization. Thus, it is critical to reduce the variations between images to enable a single model to process multiple datasets. Herein, we introduce P3T-Net, a pseudo-3D domain transfer network that transfers diverse 3D images into a uniform domain before processing using deep learning models. Remarkably, P3T-Net enables the reuse of previously trained networks for processing new images and considerably reduces the computational cost of transferring 3D images across domains. These unique capabilities were demonstrated in the following scenarios: (i) image enhancement of fast scans for geological rock and hydrogen fuel cells, (ii) enhancement of images to match the quality of multi-source imaging for lithium-ion batteries, (iii) accurate segmentation of images captured under different conditions, and (iv) tera-scale 3D transfer (10 11 voxels) on a single GPU. Overall, the proposed approach addresses cross-domain inconsistencies across various materials and conditions, thereby enabling more robust and generalizable deep learning solutions for a wide range of material imaging tasks.

25 ENERGY STORAGE↗
DOE OSTI2025

Machine learning for domain transfer between simulated and experimental 2D X-ray diffraction patterns using generative adversarial networks

X-ray diffraction (XRD) is a well-established technique for analyzing materials at an atomic level. Dynamic compression experiments (DCE), in which materials are subject to extreme pressures, can provide fundamental understanding to pressure-induced phase transitions and compression of the crystal lattice. The analysis of XRD patterns from highly compressed samples is non-trivial given the sparsity of data, high experimental costs, and the fact that the data is often marred with X-ray background and other artifacts. While accurate computational frameworks exist, they solve the forward problem—from structures and orientations to XRD patterns. Solving the inverse problem for 2D experimental diffraction patterns is currently a complex manual process of matching and comparing experimentally observed patterns to computationally generated ones. Machine learning is a promising tool for automating the matching process but often requires data-intensive architectures. Here, in this study, we use a CycleGAN to translate the domain of limited experimental data to a domain in which there is readily available simulated data. This domain shift allows data-intensive machine learning models that have only been trained on simulated XRD patterns to be used in the analysis of experiments.

Brozak, Samantha Jean [Sandia National Laboratorie↗
DOE OSTI2021

Learning Universal Authorship Representations

This code contains all the utilities required to reproduce the results of our EMNLP 2021 paper "Learning Universal Authorship Representations". It contains the utilities required for download the datasets, training our model, and performing all evaluations necessary for reproducing the results in the paper. Here's the abstract of our work: Determining whether two documents were composed by the same author, also known as authorship verification, has traditionally been tackled using statistical methods. Recently, authorship representations learned using neural networks have been found to outperform alternatives, particularly in large-scale settings involving hundreds of thousands of authors. But do such representations learned in a particular domain transfer to other domains? Or are these representations inherently entangled with domain-specific features? To study these questions, we conduct the first large-scale study of cross-domain transfer for authorship verification considering zero-shot transfers involving three disparate domains: Amazon reviews, fanfiction short stories, and Reddit comments. We find that although a surprising degree of transfer is possible between certain do- mains, it is not so successful between others. We examine properties of these domains that influence generalization and propose simple but effective methods to improve transfer.

Rivera Soto, RafaelA↗
NASA NTRS2015

Variable-Domain Displacement Transfer Functions for Converting Surface Strains into Deflections for Structural Deformed Shape Predictions

Variable-Domain Displacement Transfer Functions were formulated for shape predictions of complex wing structures, for which surface strain-sensing stations must be properly distributed to avoid jointed junctures, and must be increased in the high strain gradient region. Each embedded beam (depth-wise cross section of structure along a surface strain-sensing line) was discretized into small variable domains. Thus, the surface strain distribution can be described with a piecewise linear or a piecewise nonlinear function. Through discretization, the embedded beam curvature equation can be piece-wisely integrated to obtain the Variable-Domain Displacement Transfer Functions (for each embedded beam), which are expressed in terms of geometrical parameters of the embedded beam and the surface strains along the strain-sensing line. By inputting the surface strain data into the Displacement Transfer Functions, slopes and deflections along each embedded beam can be calculated for mapping out overall structural deformed shapes. A long tapered cantilever tubular beam was chosen for shape prediction analysis. The input surface strains were analytically generated from finite-element analysis. The shape prediction accuracies of the Variable- Domain Displacement Transfer Functions were then determined in light of the finite-element generated slopes and deflections, and were fofound to be comparable to the accuracies of the constant-domain Displacement Transfer Functions

displacement transfer functions↗
DOE OSTI2022

Realistic galaxy image simulation via score-based generative models

ABSTRACT We show that a denoising diffusion probabilistic model (DDPM), a class of score-based generative model, can be used to produce realistic mock images that mimic observations of galaxies. Our method is tested with Dark Energy Spectroscopic Instrument (DESI) grz imaging of galaxies from the Photometry and Rotation curve OBservations from Extragalactic Surveys (PROBES) sample and galaxies selected from the Sloan Digital Sky Survey. Subjectively, the generated galaxies are highly realistic when compared with samples from the real data set. We quantify the similarity by borrowing from the deep generative learning literature, using the ‘Fréchet inception distance’ to test for subjective and morphological similarity. We also introduce the ‘synthetic galaxy distance’ metric to compare the emergent physical properties (such as total magnitude, colour, and half-light radius) of a ground truth parent and synthesized child data set. We argue that the DDPM approach produces sharper and more realistic images than other generative methods such as adversarial networks (with the downside of more costly inference), and could be used to produce large samples of synthetic observations tailored to a specific imaging survey. We demonstrate two potential uses of the DDPM: (1) accurate inpainting of occluded data, such as satellite trails, and (2) domain transfer, where new input images can be processed to mimic the properties of the DDPM training set. Here we ‘DESI-fy’ cartoon images as a proof of concept for domain transfer. Finally, we suggest potential applications for score-based approaches that could motivate further research on this topic within the astronomical community.

79 ASTRONOMY AND ASTROPHYSICS↗
NASA NTRSDate not supplied

Improving Sim-to-Real Transfer in Vision-Based Robot Navigation Via Instance-Level GAN-Based Data Augmentation

Achieving robust vision-based robotic tasks requires large amounts of data, which are often difficult to obtain in real-world scenarios. Simulators and synthetic data offer a cost-effective alternative, but the visual gap between simulation and reality hinders the performance of models when deployed in real-world environments. In this paper, we present a data augmentation pipeline that integrates a foundation model (Segment Anything Model) with an unsupervised image-to-image translation model (CycleGAN) for instance-level domain transfer from simulation to reality. This pipeline enables the generation of realistic labeled data from synthetic images for training supervised machine learning models in vision-based navigation tasks. We evaluate our approach on real-world data for ego-vehicle pose estimation, a critical autonomous navigation task involving the prediction of cross-track position and heading angle relative to road center line markings. The results of our tests show that our GAN-based data augmentation pipeline significantly outperforms models trained solely on simulation data or on data processed with standard image augmentation methods for sim-to-real transfer, enhancing model robustness and generalizability in real-world scenarios. Our method provides a scalable and flexible data augmentation tool for leveraging large synthetic datasets to enhance vision-based robotic navigation tasks.

artificial intelligence↗
NASA NTRS1990

Correlation of AH-1G airframe flight vibration data with a coupled rotor-fuselage analysis

The formulation and features of the Rotor-Airframe Comprehensive Analysis Program (RACAP) is described. The analysis employs a frequency domain, transfer matrix approach for the blade structural model, a time domain wake or momentum theory aerodynamic model, and impedance matching for rotor-fuselage coupling. The analysis is applied to the AH-1G helicopter, and a correlation study is conducted on fuselage vibration predictions. The purpose of the study is to evaluate the state-of-the-art in helicopter fuselage vibration prediction technology. The fuselage vibration predicted using RACAP are fairly good in the vertical direction and somewhat deficient in the lateral/longitudinal directions. Some of these deficiencies are traced to the fuselage finite element model.

Sangha, K.↗
NASA NTRS1978

Human operator identification model and related computer programs

Four computer programs which provide computational assistance in the analysis of man/machine systems are reported. The programs are: (1) Modified Transfer Function Program (TF); (2) Time Varying Response Program (TVSR); (3) Optimal Simulation Program (TVOPT); and (4) Linear Identification Program (SCIDNT). The TV program converts the time domain state variable system representative to frequency domain transfer function system representation. The TVSR program computes time histories of the input/output responses of the human operator model. The TVOPT program is an optimal simulation program and is similar to TVSR in that it produces time histories of system states associated with an operator in the loop system. The differences between the two programs are presented. The SCIDNT program is an open loop identification code which operates on the simulated data from TVOPT (or TVSR) or real operator data from motion simulators.

Kessler, K. M.↗
DOE OSTI2022

Smoothed particle hydrodynamics modeling and analysis of oxide reduction process for uranium oxides

A common kinetic feature for oxide reduction chemical/electrochemical processes is oxygen transport via a porous metallic layer, which has been considered as a rate-determining step for reducing uranium oxides to metallic uranium. Accounting this kinetic behavior must involve the resolution of the moving reactive interface between shrinking oxide and expanding metal phases. This study presents a numerical model using smoothed particles hydrodynamics (SPH) to effectively deal with the evolution of the shrinking core reaction interface and oxygen transport via mass transfer of lithium oxide (Li 2 O) species in multiple mass transfer domains. We successfully validated the proposed model against a theoretical derivation for the oxide reduction process handling a shrinking oxide core with molten salt and metal ash medium on a simple planar geometry. Armed with successful validation results, the model examined a realistic reactant geometry to extend the arguments beyond the one-dimensional analyses, allowing the proposed model to apply to general application scenarios with multi-dimensional geometry. Here, this study demonstrated that the proposed model could simulate and evaluate an arbitrary reaction basket design without iterative experimental trials, which is prohibitive for a scaled high-temperature molten salt study in an inert environment. The construct potentially provides not only deep insights on multiphysics behaviors governing the process dynamics but also a robust framework for evaluating and screening candidate basket designs in the most cost-effective manner.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2026

Dynamic Multipoles in the RCS

The Rapid Cycling Synchrotron (RCS) of the Electron--Ion Collider (EIC) will accelerate polarized electrons up to 18 GeV at a repetition rate of 1 Hz. Dynamic multipoles driven by eddy currents in the vacuum chamber can affect beam dynamics and polarization preservation during the energy ramp. Building on the analytical treatment of S.Y. Lee for a ramped normal-conducting dipole magnet, we use conformal transformations to derive compact expressions for dynamic multipoles in higher-order ramped magnets with circular beam pipes. We further derive analytical frequency-domain transfer functions for both self-response and cross-response multipoles, providing compact dynamical models for their evolution during arbitrary ramp waveforms. The resulting analytical predictions are benchmarked against finite-element simulations, including the effects of practical pole geometry and beam-pipe misalignment. Finally, the expected dynamic multipoles in the RCS magnets are estimated and their implications for beam dynamics and polarization preservation are discussed.

43 PARTICLE ACCELERATORS↗
NASA NTRS1974

Hardware realization of sampled-data controllers

This paper examines the hardware implementation process for digital controllers (filters), beginning with the desired z-domain transfer function of the digital compensator and ending in the selection of a hardware realization scheme. The final hardware realization is shown to be affected by quantization noise, dynamic range considerations, limit cycles, coefficient roundoff, and filter structure. A design example is included to illustrate the implementation process for a digital rate filter for vehicle control of the space shuttle. This paper develops open-loop techniques which may be extended to the closed-loop case.

Nagle, H. T., Jr.↗
NASA NTRS1981

Theoretical linear approach to the combined man-manipulator system in manual control of an aircraft

An approach to the calculation of the dynamic characteristics of the combined man manipulator system in manual aircraft control was derived from a model of the neuromuscular system. This model combines the neuromuscular properties of man with the physical properties of the manipulator system which is introduced as pilot manipulator model into the manual aircraft control. The assumption of man as a quasilinear and time invariant control operator adapted to operating states, depending on the flight phases, of the control system gives rise to interesting solutions of the frequency domain transfer functions of both the man manipulator system and the closed loop pilot aircraft control system. It is shown that it is necessary to introduce the complete precision pilot manipulator model into the closed loop pilot aircraft transfer function in order to understand the well known handling quality criteria, and to derive these criteria directly from human operator properties.

Brauser, K.↗
NASA NTRS1988

Experimental implementation of adaptive control for flexible space structures

On-going research at The Aerospace Corporation studying the feasibility of applying adaptive control methodologies to the control of flexible space structures is described. A laboratory testbed was established to test system identification and control approaches. The laboratory set-up and controller design approach are discussed. The ARX least squares parameter estimation technique is analyzed in terms of frequency domain transfer function bias error. This analysis approach enables the determination of the effects of sampling rate, sensor type, and data prefiltering on the estimation performance. The ability to identify space structure dynamics over a range of frequencies is shown to be heavily dependent on these factors.

Mcgraw, Gary A.↗
NASA NTRS1998

Comparison of Five System Identification Algorithms for Rotorcraft Higher Harmonic Control

This report presents an analysis and performance comparison of five system identification algorithms. The methods are presented in the context of identifying a frequency-domain transfer matrix for the higher harmonic control (HHC) of helicopter vibration. The five system identification algorithms include three previously proposed methods: (1) the weighted-least- squares-error approach (in moving-block format), (2) the Kalman filter method, and (3) the least-mean-squares (LMS) filter method. In addition there are two new ones: (4) a generalized Kalman filter method and (5) a generalized LMS filter method. The generalized Kalman filter method and the generalized LMS filter method were derived as extensions of the classic methods to permit identification by using more than one measurement per identification cycle. Simulation results are presented for conditions ranging from the ideal case of a stationary transfer matrix and no measurement noise to the more complex cases involving both measurement noise and transfer-matrix variation. Both open-loop identification and closed- loop identification were simulated. Closed-loop mode identification was more challenging than open-loop identification because of the decreasing signal-to-noise ratio as the vibration became reduced. The closed-loop simulation considered both local-model identification, with measured vibration feedback and global-model identification with feedback of the identified uncontrolled vibration. The algorithms were evaluated in terms of their accuracy, stability, convergence properties, computation speeds, and relative ease of implementation.

Jacklin, Stephen A.↗