Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “separated representations”

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 181 records · Page 10
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

CAD-VAE: Leveraging Correlation-Aware Latents for Comprehensive Fair Disentanglement

While deep generative models have significantly advanced representation learning, they may inherit or amplify biases and fairness issues by encoding sensitive attributes alongside predictive features. Enforcing strict independence in disentanglement is often unrealistic when target and sensitive factors are naturally correlated. To address this challenge, we propose CAD-VAE(Correlation-Aware Disentangled VAE), which introduces a correlated latent code to capture the information shared between the target and sensitive attributes. Given this correlated latent, our method effectively separates over-lapping factors without extra domain knowledge by directly minimizing the conditional mutual information between target and sensitive codes. A relevance-driven optimization strategy refines the correlated code by efficiently capturing essential correlated features and eliminating redundancy. Extensive experiments on benchmark datasets demonstrate that CAD-VAE produces fairer representations, realistic counterfactuals, and improved fairness-aware image editing.

Ma, Chenrui [University of California Irvine]↗
NASA NTRS2005

Probabilistic Analysis of Impact of Wake Vortices on Closely-Spaced Parallel Approaches

One of the primary constraints on the capacity of the nation's air transportation system is the landing capacity of its largest airports. Many airports with closely spaced parallel runways suffer a severe runway acceptance rate when the weather conditions do not allow full utilization of these parallel runways. The present requirement for simultaneous independent landings in Instrument Meteorological Conditions, IMC, is at least 4300 feet of lateral runway spacing (as close as 3000 feet for runways with a Precision Runway Monitor). Operations in Visual Meteorological Conditions, VMC, to Closely Spaced Parallel Approaches only require a lateral runway spacing greater than 750 feet. A study by Hardy and Lewis integrated and extended earlier studies and concepts in lateral traffic separation, longitudinal station keeping, wake prediction, wake display, and the concepts of R N P into a preliminary system concept for Closely Spaced Parallel Approaches in IMC. This system allows IMC airport acceptance rates to approach those for VMC. The system concept that was developed, presented traffic and wake information on the NAVigation Display, NAV, and developed operational procedures for a mix of conventional and Runway Independent Aircraft with different approach speeds to Closely Spaced Parallel Runways. This paper first describes some improvements made on the technology needed to better predict and formulate a probabilistic representation for the time-dependent motion and spreading of the hazardous region associated with the lift-generated vortex wakes of preceding aircraft. In this way, the time at which the vortex wakes of leading aircraft intrude into the airspace of adjacent flight-corridor/runway combinations can be more reliably predicted. Such a prediction is needed because it determines restraints to be placed on in-trail separation distances; or, the allowable time intervals between aircraft executing nearly simultaneous landings or takeoffs on very closely-spaced runways. Improved estimates of wake spreading are achieved by inclusion of representations in the equations for wake spreading due to ambient turbulence and due to the long-wave instability of a vortex pair. Wake motion and spreading due to the time-averaged wind and its variations with time, are retained. The more detailed representation of wake spreading presented here permits the development of probabilistically-based uncertainty estimates for wake spreading. Measurements needed within actual aircraft wake vortices to validate and support this analysis are also described. The second part of the paper uses the improvements in the accuracy of the location of wake vortices to extend the preliminary system concept for Closely Spaced Parallel Approaches described earlier with more robust operational procedures. Additionally, improvements in longitudinal station keeping, wake display, and risk assessment methodologies are incorporated and described.

Hardy, Gordon H.↗
NASA NTRS1978

Synthetic atmospheric transmittance spectra near 15 and 4.3 microns

Synthetic monochromatic atmospheric transmittance spectra are presented for infrared intervals in the vicinity of the 15 and 4.3 microns CO2 bands. The intervals are nominally 20 kaysers in width, and the spectra partition the intervals 560, 780 kayser and 2180, 2400 kayser, respectively. The spectra are for a vertical atmospheric path. A given spectrum is presented at that pressure for which the mean weighting function is a maximum. In order to indicate the molecular origin of the features in a given spectrum, separate spectra are also presented for those species which make center line contributions to the absorption over the interval. The transmittance model for the calculations is described. In addition to the representation of spectral features for intervals which comprise these two CO2 bands, the spectra afford a utility in the design of high precision satellite temperature radiometers.

Susskind, J.↗
NASA NTRS2014

Time and Frequency-Domain Cross-Verification of SLS 6DOF Trajectory Simulations

The SLS GNC team and its partners have developed several time- and frequency-based simulations for development and analysis of the proposed SLS launch vehicle. The simulations differ in fidelity and some have unique functionality that allows them to perform specific analyses. Some examples of the purposes of the various models are: trajectory simulation, multi-body separation, Monte Carlo, hardware in the loop, loads, and frequency domain stability analyses. While no two simulations are identical, many of the models are essentially six degree-of-freedom (6DOF) representations of the SLS plant dynamics, hardware implementation, and flight software. Thus at a high level all of those models should be in agreement. Comparison of outputs from several SLS trajectory and stability analysis tools are ongoing as part of the program's current verification effort. The purpose of these comparisons is to highlight modeling and analysis differences, verify simulation data sources, identify inconsistencies and minor errors, and ultimately to verify output data as being a good representation of the vehicle and subsystem dynamics. This paper will show selected verification work in both the time and frequency domain from the current design analysis cycle of the SLS for several of the design and analysis simulations. In the time domain, the tools that will be compared are MAVERIC, CLVTOPS, SAVANT, STARS, ARTEMIS, and POST 2. For the frequency domain analysis, the tools to be compared are FRACTAL, SAVANT, and STARS. The paper will include discussion of these tools including their capabilities, configurations, and the uses to which they are put in the SLS program. Determination of the criteria by which the simulations are compared (matching criteria) requires thoughtful consideration, and there are several pitfalls that may occur that can severely punish a simulation if not considered carefully. The paper will discuss these considerations and will present a framework for responding to these issues when they arise. For example, small event timing differences can lead to large differences in mass properties if the criteria are to measure those properties at the same time, or large differences in altitude if the criteria are to measure those properties when the simulation experiences a staging event. Similarly, a tiny difference in phase can lead to large gain margin differences for frequency-domain comparisons of gain margins.

VanZwieten, Tannen↗
NASA NTRS2017

Time and Frequency-Domain Cross-Verification of SLS 6DOF Trajectory Simulations

The Space Launch System (SLS) Guidance, Navigation, and Control (GNC) team and its partners have developed several time- and frequency-based simulations for development and analysis of the proposed SLS launch vehicle. The simulations differ in fidelity and some have unique functionality that allows them to perform specific analyses. Some examples of the purposes of the various models are: trajectory simulation, multi-body separation, Monte Carlo, hardware in the loop, loads, and frequency domain stability analyses. While no two simulations are identical, many of the models are essentially six degree-of-freedom (6DOF) representations of the SLS plant dynamics, hardware implementation, and flight software. Thus at a high level all of those models should be in agreement. Comparison of outputs from several SLS trajectory and stability analysis tools are ongoing as part of the program's current verification effort. The purpose of these comparisons is to highlight modeling and analysis differences, verify simulation data sources, identify inconsistencies and minor errors, and ultimately to verify output data as being a good representation of the vehicle and subsystem dynamics. This paper will show selected verification work in both the time and frequency domain from the current design analysis cycle of the SLS for several of the design and analysis simulations. In the time domain, the tools that will be compared are MAVERIC, CLVTOPS, SAVANT, STARS, ARTEMIS, and POST 2. For the frequency domain analysis, the tools to be compared are FRACTAL, SAVANT, and STARS. The paper will include discussion of these tools including their capabilities, configurations, and the uses to which they are put in the SLS program. Determination of the criteria by which the simulations are compared (matching criteria) requires thoughtful consideration, and there are several pitfalls that may occur that can severely punish a simulation if not considered carefully. The paper will discuss these considerations and will present a framework for responding to these issues when they arise. For example, small event timing differences can lead to large differences in mass properties if the criteria are to measure those properties at the same time, or large differences in altitude if the criteria are to measure those properties when the simulation experiences a staging event. Similarly, a tiny difference in phase can lead to large gain margin differences for frequency-domain comparisons of gain margins.

Johnson, Matthew↗
NASA NTRS2018

Relevence of 3D Simulations and Sandwich Core Topology for the Modeling of Honeycomb Core Sandwich Panels Undergoing Mode I Delamination

A recently developed cohesive zone traction-separation law, which includes the effects of fiber bridging in a novel way, is extended from 2D to 3D. The proposed cohesive model is applied to low fidelity (i.e. homogenized core representation) and high fidelity (i.e. directly accounting for the core topology) finite element models of a composite panel comprised of carbon fiber reinforced plastic facesheets and a honeycomb sandwich core. This enables the investigation of 2D to 3D parameter transferability, width-dependent effects such as thumbnail-shaped crack growth, and the verification of plane strain / plane stress assumptions. A pronounced curvature of the initial interface-related crack front is observed, while the bridging-related crack front is straight. Furthermore, it is found that the cohesive parameters can easily be transferred from 2D to 3D under plane stress assumptions, but not under plane strain assumptions. The numerical predictions are compared to experimental load-displacement and R-curves.

Composites↗
NASA NTRS1997

Neural Network Research on Tilt-Rotor Hover Performance

Results from a neural network study on XV-15 tilt-rotor hover performance are presented. Two XV-15 test data bases, acquired during separate tests conducted in the National Full-Scale Aerodynamics Complex at NASA Ames, were used in this study. The two XV-15 test data bases used were: the 1997 80- by 120-Foot Wind Tunnel test data base and the 1984 Outdoor Aerodynamic Research Facility (OARF) test data base. The objectives associated with the 80- by 120-Foot Wind Tunnel test data were to conduct data quality checks, obtain neural network representations, and demonstrate sensitivity to test conditions. The objectives associated with the OARF outdoor test data were to obtain a 'zero wind' neural network representation, and formulate and implement a neural network based wind correction procedure. An additional objective, common to both test data bases, was to conduct error comparisons. The conclusions from the present study were as follows: Compared to measured parameters (i.e., collective pitch angle), derived parameters (i.e., thrust coefficient) were found preferable as neural network inputs. For the 80- by 120-Foot Wind Tunnel test data base, the neural network based figure of merit representations did not extrapolate rotor stall. Using the OARF outdoor test database, a neural network based wind correction procedure was formulated and successfully implemented. Based on an error comparison, the present neural network based wind correction procedure was found to be more accurate compared to the existing, momentum-theory-based wind correction procedure. Also, a separate error comparison showed that the 80- by 120-Foot Wind Tunnel hover performance data were comparable to non-wind-corrected outdoor test data.

Kozapalli, S.↗
DOE OSTI2023

Review of rotating wing dynamic stall: Experiments and flow control

Dynamic stall has been a technical challenge and a fluid dynamical subject of interest for more than fifty years; but in the last decade significant advances have been made in the understanding, prediction, modeling, and control of dynamic stall on rotors. This paper provides a summary of the state of the art of dynamic stall experiments and future directions in the understanding of dynamic stall on rotors. Experimental data sets are discussed, as well the direction of future research for control of dynamic stall. Coordinated testing between airfoils and rotating blades, as well as close integration between computational and experimental studies were found to be productive approaches. Advanced analysis methods, including statistical methods, modal representations, and artificial intelligence methods have led to significant advances in the understanding of dynamic stall. Investigations of dynamic stall control devices have allowed many useful targeted investigations of the transition to separated flow, but have not yet resulted in a commercially implemented device.

42 ENGINEERING↗
NASA NTRS1978

Upgraded viscous flow analysis of multielement airfoils

A description of an improved version of the NASA/Lockheed multielement airfoil analysis computer program is presented. The improvements include several major modifications of the aerodynamic model as well as substantial changes of the computer code. The modifications of the aerodynamic model comprise the representation of the boundary layer and wake displacement effects with an equivalent source distribution, the prediction of wake parameters with Green's lag-entrainment method, the calculation of turbulent boundary layer separation with the method of Nash and Hicks, the estimation of the onset of confluent boundary layer separation with a modified form of Goradia's method, and the prediction of profile drag with the formula of Squire and Young. The paper further describes the modifications of the computer program for which the structured approach to computer software development was employed. Important aspects of the structured program development such as the functional decomposition of the aerodynamic theory and its numerical implementation, the analysis of the data flow within the code, and the application of a pseudo code are discussed.

Brune, G. W.↗
NASA NTRS1979

Upgraded viscous flow analysis of multi-element airfoils

A description of an improved version of the NASA/Lockheed multi-element airfoil analysis computer program is presented. The improvements include several major modifications of the aerodynamic model as well as substantial changes of the computer code. The modifications of the aerodynamic model comprise the representation of the boundary layer and wake displacement effects with an equivalent source distribution, the prediction of wake parameters with Green's lag-entrainment method, the calculation of turbulent boundary layer separation with the method of Nash and Hicks, the estimation of the onset of confluent boundary layer separation with a modified form of Goradia's method, and the prediction of profile drag with the formula of Squire and Young. The modifications of the computer program for which the structured approach to computer software development was employed are also described. Important aspects of the structured program development such as the functional decomposition of the aerodynamic theory and its numerical implementation, the analysis of the data flow within the code, and the application of a pseudo code are discussed.

Brune, G. W.↗
NASA NTRSDate not supplied

Analyzing Natural Language Context in Human-Machine Teaming using Supervised Machine Learning

Building a foundation for trustworthiness and trust verification in multi-asset teaming is the research challenge of Autonomy Teaming and TRAjectories for Complex Trusted Operational Reliability (ATTRACTOR). The Design Reference Mission (DRM) for ATTRACTOR is a search and rescue mission objective governed by a multi-member team consisting of human and machine operators. A crucial component to the effort is the communication between humans and autonomous agents throughout both planning and execution stages of the mission. Intuitive communication methods and modalities are posited as critical enablers for certifying trust and trustworthiness. This paper reports on the data collection and analysis conducted in support of the Human Informed Natural-language GANs Evaluation (HINGE)project to attain explainable and trusted communication between human-machine assets. Two identically curated image description datasets were acquired for HINGE, both consisting of two unique input modalities (typed vs. verbal) and retrieved in two distinct contexts (general vs. specific). The gathered datasets were assessed and compared using Parts-of-Speech (POS)features, sentence similarity metrics, and linguistic analysis. Then, the datasets were modeled and tested separately and in combination with one another using machine learning algorithms. The comparison and testing results reveal a superior dataset, by which a preferred context and input is understood, for generating image representations of missing persons using a Generative Adversarial Network (GAN).

Bryan A Barrows↗
DOE OSTI2025

Entanglement Cost for Infinite-Dimensional Physical Systems

We prove that the entanglement cost equals the regularized entanglement of formation for any infinite-dimensional quantum state ρ ΑΒ with finite quantum entropy on at least one of the subsystems A or B. This generalizes a foundational result in quantum information theory that was previously formulated only for operations and states on finite-dimensional systems. The extension to infinite-dimensional systems is nontrivial because the conventional tools for establishing both the direct and converse bounds, i.e., strong typicality, monotonicity, and asymptotic continuity, are no longer directly applicable. To address this problem, we construct a new entanglement dilution protocol for infinite-dimensional states implementable by local operations and a finite amount of one-way classical communication (one-way LOCC), using weak and strong typicality multiple times. We also prove the optimality of this protocol among all protocols, even under infinite-dimensional separable operations, by developing an argument based on alternative forms of monotonicity and asymptotic continuity of the entanglement of formation for infinite-dimensional states. Along the way, we derive a new integral representation for the quantum entropy of infinite-dimensional states, which we believe to be of independent interest. Our results allow us to fully characterize an important operational entanglement measure—the entanglement cost—for all infinite-dimensional physical systems.

Complexity↗
DOE OSTI2021

Navigating Transition-Metal Chemical Space: Artificial Intelligence for First-Principles Design

Conspectus The variability of chemical bonding in open-shell transition-metal complexes not only motivates their study as functional materials and catalysts but also challenges conventional computational modeling tools. Here, tailoring ligand chemistry can alter preferred spin or oxidation states as well as electronic structure properties and reactivity, creating vast regions of chemical space to explore when designing new materials atom by atom. Although first-principles density functional theory (DFT) remains the workhorse of computational chemistry in mechanism deduction and property prediction, it is of limited use here. DFT is both far too computationally costly for widespread exploration of transition-metal chemical space and also prone to inaccuracies that limit its predictive performance for localized d electrons in transition-metal complexes. These challenges starkly contrast with the well-trodden regions of small-organic-molecule chemical space, where the analytical forms of molecular mechanics force fields and semiempirical theories have for decades accelerated the discovery of new molecules, accurate DFT functional performance has been demonstrated, and gold-standard methods from correlated wavefunction theory can predict experimental results to chemical accuracy. The combined promise of transition-metal chemical space exploration and lack of established tools has mandated a distinct approach. In this Account, we outline the path we charted in exploration of transition-metal chemical space starting from the first machine learning (ML) models (i.e., artificial neural network and kernel ridge regression) and representations for the prediction of open-shell transition-metal complex properties. The distinct importance of the immediate coordination environment of the metal center as well as the lack of low-level methods to accurately predict structural properties in this coordination environment first motivated and then benefited from these ML models and representations. Once developed, the recipe for prediction of geometric, spin state, and redox potential properties was straightforwardly extended to a diverse range of other properties, including in catalysis, computational “feasibility”, and the gas separation properties of periodic metal–organic frameworks. Interpretation of selected features most important for model prediction revealed new ways to encapsulate design rules and confirmed that models were robustly mapping essential structure–property relationships. Encountering the special challenge of ensuring that good model performance could generalize to new discovery targets motivated investigation of how to best carry out model uncertainty quantification. Distance-based approaches, whether in model latent space or in carefully engineered feature space, provided intuitive measures of the domain of applicability. With all of these pieces together, ML can be harnessed as an engine to tackle the large-scale exploration of transition-metal chemical space needed to satisfy multiple objectives using efficient global optimization methods. In practical terms, bringing these artificial intelligence tools to bear on the problems of transition-metal chemical space exploration has resulted in ML-model assessments of large, multimillion compound spaces in minutes and validated new design leads in weeks instead of decades.

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

Clustering High-dimensional Toxicogenomics Data with Rare Signals

Toxicogenomics studies the gene and protein activities to drug treatments or toxic exposures. As the drugs and genes are numerous, toxicogenomics data are naturally high dimensional, with dimension sizes up to millions. In addition, the distribution of toxicogenomics data is oftentimes skewed, and they contain rare but important signals representing a cell or organism’s response to toxicity. The combination of high dimension and extremely skewed distribution of toxicogenomics data makes clustering analysis extremely challenging.We present our study of clustering toxicogenomics data using classical approaches such as principal component analysis as well as deep learning approaches such as auto-encoders. Our experiments show that these approaches fail to preserve rare signals and produce high-quality clusters. We then explore augmenting matrix factorization with deep learning techniques such as attention mechanism to produce latent representations for clustering. Our technique is able to better preserve rare signals after dimensionality reduction than prior approaches. Furthermore, we combine our augmented matrix factorization with a mechanism similar to autoencoder to balance separable clusters and low regeneration errors. Our experiments demonstrate better clustering with our proposed approach.

Cong, Guojing↗
DOE OSTI2026

AIF for Vis (Active Inference for simulating human interpretation of data visualization) [SWR-26-084]

AIF for Vis contains the Active Inference models and analysis scripts used to study a simple visualization-interpretation task: estimating the average value of two bars in a bar chart. The work is a proof of concept for translating hypothesized cognitive strategies into executable, inspectable process models. We implement two idealized strategies inspired by dual-process accounts of visualization-aided decision making: *Fast model: a compressed, heuristic strategy that estimates the visual midpoint of the two bars and maintains a single belief over their average. *Slow model: a sequential, analytic strategy that estimates the two bar heights separately and maintains them in working memory before computing an average. Both models use a common Active-Inference-inspired framework for sequential perception, belief updating, action selection, and reporting. Their different internal representations produce distinct predicted vulnerabilities: *the Fast model is more susceptible to tick-salience bias; *the Slow model is more susceptible to working-memory decay. The repository includes the model implementations, scripts used for the experiments reported in the paper, precomputed trial-level results, and plotting scripts.

Goldwyn, Harrison [National Laboratory of the Rock↗
DOE OSTI2020

CMIP5 model selection for ISMIP6 ice sheet model forcing: Greenland and Antarctica

Abstract. The ice sheet model intercomparison project for CMIP6 (ISMIP6) effort brings together the ice sheet and climate modeling communities to gain understanding of the ice sheet contribution to sea level rise. ISMIP6 conducts stand-alone ice sheet experiments that use space- and time-varying forcing derived from atmosphere–ocean coupled global climate models (AOGCMs) to reflect plausible trajectories for climate projections. The goal of this study is to recommend a subset of CMIP5 AOGCMs (three core and three targeted) to produce forcing for ISMIP6 stand-alone ice sheet simulations, based on (i) their representation of current climate near Antarctica and Greenland relative to observations and (ii) their ability to sample a diversity of projected atmosphere and ocean changes over the 21st century. The selection is performed separately for Greenland and Antarctica. Model evaluation over the historical period focuses on variables used to generate ice sheet forcing. For stage (i), we combine metrics of atmosphere and surface ocean state (annual- and seasonal-mean variables over large spatial domains) with metrics of time-mean subsurface ocean temperature biases averaged over sectors of the continental shelf. For stage (ii), we maximize the diversity of climate projections among the best-performing models. Model selection is also constrained by technical limitations, such as availability of required data from RCP2.6 and RCP8.5 projections. The selected top three CMIP5 climate models are CCSM4, MIROC-ESM-CHEM, and NorESM1-M for Antarctica and HadGEM2-ES, MIROC5, and NorESM1-M for Greenland. This model selection was designed specifically for ISMIP6 but can be adapted for other applications.

54 ENVIRONMENTAL SCIENCES↗
NASA NTRS1984

Two-dimensional wind-tunnel interference from measurements on two contours

This paper describes how wall-induced velocities near a model in a two-dimensional wind tunnel can be estimated from upwash distributions measured along two contours surrounding a model. The method is applicable to flows that can be represented by linear theory. It was derived by applying the Schwarz Integral Formula separately to the two contours and by exploiting the free-air relationship between upwashes along the contours. Advantages of the method are that only one flow quantity need by measured and no representation of the model is required. A weakness of the method is that it assumes streamwise interference velocity vanishes far upstream of the model. This method was applied to a simple theoretical model of flow in a solid-wall wind tunnel. The theoretical interference velocities and the velocities computed using the method were in excellent agreement. The method was then used to analyze experimental data acquired during adaptive-wall experiments at Ames Research Center. This analysis confirmed that the wall adjustments reduced wall-induced velocities near the model.

Schairer, E. T.↗
NASA NTRS1989

Advances in the computation of transonic separated flows over finite wings

Problems encountered in numerical simulations of transonic wind-tunnel experiments with low-aspect-ratio wings are surveyed and illustrated. The focus is on the zonal Euler/Navier-Stokes program developed by Holst et al. (1985) and its application to shock-induced separation. The physical basis and numerical implementation of the method are reviewed, and results are presented from studies of the effects of artificial dissipation, boundary conditions, grid refinement, the turbulence model, and geometry representation on the simulation accuracy. Extensive graphs and diagrams and typical flow visualizations are provided.

Kaynak, Unver↗