SEARCH · Search NASA

Results for “Robustness analysis”

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.

SourceResearch connectionFrom yearThrough year

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

At least 289 records · Page 16

All sources NASA NTRS DOE OSTI

DOE OSTI2021

Building Trustworthy Machine Learning Models for Astronomy

Astronomy is entering an era of data-driven discovery, due in part to modern machine learning (ML) techniques enabling powerful new ways to interpret observations. This shift in our scientific approach requires us to consider whether we can trust the black box. Here, we overview methods for an often-overlooked step in the development of ML models: building community trust in the algorithms. Trust is an essential ingredient not just for creating more robust data analysis techniques, but also for building confidence within the astronomy community to embrace machine learning methods and results.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

DOE OSTI2023

n-Propylbenzene Droplet Combustion: Experiments and Numerical Modeling

n-Propylbenzene (PB, C9H12, 432K boiling point) is important as a constituent in surrogates for some transportation fuels but little is understood about its droplet burning characteristics. We present an experimental and numerical study of a fuel droplet burning under conditions where the sole means of gas transport is fuel evaporation at the droplet surface thereby creating a one-dimensional gas transport that is amenable to detailed numerical modeling. The one-dimensional gas transport that results is amenable to detailed numerical modeling. Initial droplet diameters (Do) ranged from 0.3 mm to about 5 mm to better understand the mechanisms that control burning over such a wide range of Do. For Do < 1 mm droplets were burned while anchored to 14 m SiC fibers within a sealed container in free-fall during which time the droplets were ignited by spark discharge and the burning process recorded by video cameras. Larger and free-floating droplets were studied in the spaced-based platform of the orbiting International Space Station for Do > 1.0 mm which provided unlimited times for experimental observation. Measurements of droplet, flame and soot shell diameters were made and are discussed in the presentation. The experimental results showed formation of optically thick soot clouds developing during burning that required a robust image analysis algorithm to enable measurement of droplet diameter. The measured droplet diameters were well predicted by a numerical model that included soot formation, a PB kinetic mechanism comprising 576 species and 27369 reactions, unsteady gas and liquid transport, variable properties and radiative transport. Flame diameters showed more variability compared to the simulations, depending on how the flame was defined (peak gas temperature and OH concentration). Droplets burned slower as Do was increased and the PB sooting propensity increased with Do. Flame extinction was observed for Do > 3.3 mm indicating a strong effect of radiation as Do was increased compared to the smaller droplet diameters examined where extinction was not experimentally observed. For the largest droplets examined the possibility that the observed extinction is indicative of a transition to cool flame burning is still under investigation.

Guo, Songtao↗

DOE OSTI2016

NORM Concentration Determination in Common Materials in an Urban Environment

The United States government performs searches for illicit radiation sources before and during large special events, such as State of the Union addresses, presidential inaugurations, or the recent papal visit. Clear detection of sources is difficult due to natural background radiation, which can vary greatly throughout a given search area. Background radiation is mostly a function of the materials used in roadways, sidewalks, and building exteriors. A team of scientists from Oak Ridge National Laboratory (ORNL), Lawrence Livermore National Laboratory (LLNL), the Remote Sensing Laboratory (RSL), and Lawrence Berkeley National Laboratory (LBNL) is exploring the nature and variability of background count rates in order to improve the performance of searches in urban areas. A series of background measurements have been and will continue to be taken at a Military Operations on Urban Terrain (MOUT) environment at the Fort Indiantown Gap (FTIG) National Guard training facility in Pennsylvania. This site replicates a small town with asphalt roads, concrete intersections, buildings, sidewalks, curbs, and gravel parking areas. Unlike an actual urban area, traffic and pedestrian access can be easily controlled to avoid interference with measurements. This paper documents the efforts focused on characterizing the various concentrations of radioactive material found in these materials at the site. The NORM concentrations determined from this work will be incorporated in to a large-scale simulation (a 'virtual test-bed') that can be used to investigate various search techniques and algorithms, detector suite configurations, and to produce representative synthetic data for future work. A combination of measurement and simulation was used to determine the concentration of {sup 40}K and the radioactive nuclides in the uranium series and the thorium series in several materials at the FTIG MOUT site. The predicted concentrations are consistent with other published results, and they matched favorably with synthetic data generated with MCNP6 models using the determined NORM concentrations and estimates of the FTIG environment as described by available construction plans and using standard material assumptions. Future measurements will incorporate lessons learned from this study. More measurements will be made of the concrete intersections at the site. The methodology explained in this report will also be carried out on building materials in an effort to characterize NORM concentrations in these background sources. Other modeling and simulation work that should be conducted includes a robust sensitivity analysis in which the effects of small changes to the model (and measurement) system can be characterized to help bound the uncertainty on the final NORM concentration predictions. (authors)

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2026

Dark Energy Survey Year 6 Results: Weak Lensing and Galaxy Clustering Cosmological Analysis Framework

We present the methodology for the weak lensing and galaxy clustering analyses of the Dark Energy Survey (DES) Year 6 data set. In this work, we design and validate the analysis pipeline for the cosmic shear, galaxy clustering plus galaxy$-$galaxy lensing ($2 \times 2$pt), and the joint analysis in the $3 \times 2$pt. Our framework accounts for key theoretical uncertainties, such as baryonic feedback and galaxy bias, incorporating both linear and non-linear models. We apply scale cuts in regimes where theoretical modeling becomes unreliable. The robustness of the pipeline is validated using mock data and simulations, confirming unbiased cosmological constraints and highlighting the importance of posterior projection effects in the validation process. As a result, we deliver robust and validated analysis pipelines for cosmic shear, $2 \times 2$pt, and $3 \times 2$pt in $Λ$CDM and $w$CDM scenarios, including a well-defined set of scales suitable for real data analysis, a robust prescription for theoretical systematics, and the theoretical covariance of the signal. This comprehensive methodology also lays the groundwork for future galaxy surveys such as the Vera C. Rubin Observatory Legacy Survey of Space and Time.

Sanchez-Cid, D. [Zurich U.; Madrid, CIEMAT; Madrid↗

NASA NTRSDate not supplied

Robust Trajectory Optimization and GN&C Performance Analysis for NRHO Rendezvous

This paper evaluates several candidate Near-Rectilinear Halo Orbits (NRHO) rendezvous trajectory designs using linear covariance (LinCov) analysis and determines the optimal locations for NRHO rendezvous translational maneuver locations. The performance of several candidate relative trajectory designs are determined as a function of relative navigation accuracy (angles only), inertial optical navigation (OpNav), range observability maneuvers, maneuver execution errors, relative maneuver targeting, and environment uncertainties. Further, the optimal locations of rendezvous maneuvers are determined for each of the candidate reference trajectories. The long-term goal of this research is to utilize LinCov and a genetic optimization algorithm (GA) to determine a complete end-to-end optimal NRHO trajectory design that is robust to navigation errors, maneuver execution errors, and environment uncertainties. This paper represents a first step toward this goal. Three candidate rendezvous trajectories with varying numbers of range-observability maneuvers are evaluated for their robustness to uncertainties, errors, and total trajectory correction delta-v performance. Some key elements of this analysis include relative navigation performance in an NRHO, relative trajectory dispersion performance, and total 3-sigma delta-v performance. This development provides the foundation to then determine an optimal and robust end-to-end NRHO rendezvous trajectory, including the determination of the optimal locations of range observability maneuvers, if needed.

Linear Covariance Analysis↗

DOE OSTI2024

“Understanding Robustness Lottery”: A Geometric Visual Comparative Analysis of Neural Network Pruning Approaches

Deep learning approaches have provided state-of-the-art performance in many applications by relying on large and overparameterized neural networks. However, such networks are very brittle and are difficult to deploy on resource-limited platforms. Model pruning, i.e., reducing the size of the network, is a widely adopted strategy that can lead to a more robust and compact model. Many heuristics exist for model pruning, but our understanding of the pruning process remains limited due to the black-box nature of a neural network model. Empirical studies show that some heuristics improve performance whereas others can make models more brittle. Here, this work aims to shed light on how different pruning methods alter the network’s internal feature representation and the corresponding impact on model performance. To facilitate a comprehensive comparison and characterization of the high-dimensional model feature space, we introduce a visual geometric analysis of feature representations. We evaluated a set of critical geometric concepts decomposed from the commonly adopted classification loss and used them to design a visualization system to compare and highlight the impact of pruning on model performance and feature representation. The proposed tool provides an environment for an in-depth comparison of pruning methods and a comprehensive understanding of how the model responds to common data corruption. By leveraging the proposed visualization, machine learning researchers can reveal the similarities between pruning methods and redundancy in robustness evaluation benchmarks, obtain geometric insights about the differences between pruned models that achieve superior robustness performance, and identify samples that are robust or fragile to model pruning and common data corruption.

Li, Zhimin [Univ. of Utah, Salt Lake City, UT (Uni↗

NASA NTRSDate not supplied

Robust Design Under Uncertainty of Hypersonic Inlets

The objective of this work is to optimize scramjet inlet designs under uncertainty at multiple fidelity levels. Optimizations are performed for both deterministic and stochastic conditions. For stochastic conditions, robust design is used to minimize the variance in system performance in addition to optimizing mean system performance. Uncertainty quantification is performed using non-intrusive polynomial chaos. Comparisons are made between low-fidelity analysis and high-fidelity analysis, as well as deterministic optimization and robust optimization. Results indicate that the general design trend of previous inlet optimization techniques is recovered for the external compression portion of an inlet. For the internal compression portion of an inlet, when viscous effects become more significant and boundary layer separation becomes more likely, the optimal trend departs from previous results. Multifidelity uncertainty quantification methods are found to substantially reduce the computation time of robust design while obtaining results that nearly match those of high-fidelity analysis. Robust optimization is found to decrease the standard deviation of throat Mach number by up to 40% compared to deterministic optimization. Changes in the design variables and total turning that lead to robust scramjet inlets are identified and linked back to fundamental physical principles through the theta-beta-Mach function.

Hypersonic↗

NASA NTRS1998

Wavelet Filtering to Reduce Conservatism in Aeroservoelastic Robust Stability Margins

Wavelet analysis for filtering and system identification was used to improve the estimation of aeroservoelastic stability margins. The conservatism of the robust stability margins was reduced with parametric and nonparametric time-frequency analysis of flight data in the model validation process. Nonparametric wavelet processing of data was used to reduce the effects of external desirableness and unmodeled dynamics. Parametric estimates of modal stability were also extracted using the wavelet transform. Computation of robust stability margins for stability boundary prediction depends on uncertainty descriptions derived from the data for model validation. F-18 high Alpha Research Vehicle aeroservoelastic flight test data demonstrated improved robust stability prediction by extension of the stability boundary beyond the flight regime.

Brenner, Marty↗

NASA NTRS1991

Grid adaptation to multiple functions for applied aerodynamic analysis

A fast and robust method of grid adaptation to multiple functions has been developed for flow analysis in 3D space. It is based on the equidistribution principle and the alternate direction adaptation method. The grid adaptation algorithm has been fully integrated with a space marching flow solver as contained within a 3D Navier-Stokes code and PAB3D-v2. The grid adaptation strategy, details of numerical implementation, examples of application, and potential extensions of the current method are presented in this paper.

Pao, S. P.↗

NASA NTRS2014

The Evolution of the Galaxy Rest-Frame Ultraviolet Luminosity Function Over the First Two Billion Years

We present a robust measurement and analysis of the rest-frame ultraviolet (UV) luminosity function at z = 4 to 8. We use deep Hubble Space Telescope imaging over the CANDELS/GOODS fields, the Hubble Ultra Deep Field and the Hubble Frontier Field deep parallel observations near the Abell 2744 and MACS J0416.1- 2403 clusters. The combination of these surveys provides an effective volume of 0.6-1.2 ×10(exp 6) Mpc(exp 3) over this epoch, allowing us to perform a robust search for bright (M(sub UV) less than −21) and faint (M(sub UV) = −18) galaxies. We select galaxies using a well-tested photometric redshift technique with careful screening of contaminants, finding a sample of 7446 galaxies at 3.5 less than z less than 8.5, with more than 1000 galaxies at z of approximately 6 - 8. We measure both a stepwise luminosity function for galaxies in our redshift samples, as well as a Schechter function, using a Markov Chain Monte Carlo analysis to measure robust uncertainties. At the faint end our UV luminosity functions agree with previous studies, yet we find a higher abundance of UV-bright galaxies at z of greater than or equal to 6. Our bestfit value of the characteristic magnitude M* is consistent with −21 at z of greater than or equal to 5, different than that inferred based on previous trends at lower redshift. At z = 8, a single power-law provides an equally good fit to the UV luminosity function, while at z = 6 and 7, an exponential cutoff at the bright-end is moderately preferred. We compare our luminosity functions to semi-analytical models, and find that the lack of evolution in M* is consistent with models where the impact of dust attenuation on the bright-end of the luminosity function decreases at higher redshift, though a decreasing impact of feedback may also be possible. We measure the evolution of the cosmic star-formation rate (SFR) density by integrating our observed luminosity functions to M(sub UV) = −17, correcting for dust attenuation, and find that the SFR density declines proportionally to (1 + z)((exp −4.3)(+/-)(0.5)) at z greater than 4, consistent with observations at z greater than or equal to 9. Our observed luminosity functions are consistent with a reionization history that starts at redshift of approximately greater than 10, completes at z greater than 6, and reaches a midpoint (x(sub HII) = 0.5) at 6.7 less than z less than 9.4. Finally, using a constant cumulative number density selection and an empirically derived rising star-formation history, our observations predict that the abundance of bright z = 9 galaxies is likely higher than previous constraints, though consistent with recent estimates of bright z similar to 10 galaxies.

Galaxy↗

DOE OSTI2023

Joint analysis of Dark Energy Survey Year 3 data and CMB lensing from SPT and Planck . I. Construction of CMB lensing maps and modeling choices

Joint analyses of cross-correlations between measurements of galaxy positions, galaxy lensing, and lensing of the cosmic microwave background (CMB) offer powerful constraints on the large-scale structure of the Universe. In a forthcoming analysis, we will present cosmological constraints from the analysis of such cross-correlations measured using Year 3 data from the Dark Energy Survey (DES), and CMB data from the South Pole Telescope (SPT) and Planck. Here we present two key ingredients of this analysis: (1) an improved CMB lensing map in the SPT-SZ survey footprint and (2) the analysis methodology that will be used to extract cosmological information from the cross-correlation measurements. Relative to previous lensing maps made from the same CMB observations, we have implemented techniques to remove contamination from the thermal Sunyaev Zel’dovich effect, enabling the extraction of cosmological information from smaller angular scales of the cross-correlation measurements than in previous analyses with DES Year 1 data. We describe our model for the cross-correlations between these maps and DES data, and validate our modeling choices to demonstrate the robustness of our analysis. We then forecast the expected cosmological constraints from the galaxy survey-CMB lensing auto and cross-correlations. We find that the galaxy-CMB lensing and galaxy shear-CMB lensing correlations will on their own provide a constraint on S 8 =σ 8 $\sqrt{Ωm/0.3}$ at the few percent level, providing a powerful consistency check for the DES-only constraints. Finally, we explore scenarios where external priors on shear calibration are removed, finding that the joint analysis of CMB lensing cross-correlations can provide constraints on the shear calibration amplitude at the 5% to 10% level.

79 ASTRONOMY AND ASTROPHYSICS↗

NASA NTRS2002

Advanced Vibration Analysis Tools and New Strategies for Robust Design of Turbine Engine Rotors

The adverse effects of small, random structural irregularities among the blades, called mistuning, can result in blade forced-response amplitudes and stresses that are much larger than those predicted for a perfectly tuned rotor. Manufacturing tolerances, deviations in material properties, or nonuniform operational wear causes mistuning; therefore, mistuning is unavoidable. Furthermore, even a small mistuning can have a dramatic effect on the vibratory behavior of a rotor because it can lead to spatial localization of the vibration energy (see the following photographs). As a result, certain blades may experience forced response amplitudes and stresses that are substantially larger than those predicted by an analysis of the nominal (tuned) design. Unfortunately, these random uncertainties in blade properties, and the immense computational effort involved in obtaining statistically reliable design data, combine to make this aspect of rotor design cumbersome.

Min, James B.↗

NASA NTRS2002

Advanced Vibration Analysis Tools and New Strategies for Robust Design of Turbine Engine Rotors

Min, James B.↗

DOE OSTI2021

A study of real-world micrograph data quality and machine learning model robustness

Abstract Machine-learning (ML) techniques hold the potential of enabling efficient quantitative micrograph analysis, but the robustness of ML models with respect to real-world micrograph quality variations has not been carefully evaluated. We collected thousands of scanning electron microscopy (SEM) micrographs for molecular solid materials, in which image pixel intensities vary due to both the microstructure content and microscope instrument conditions. We then built ML models to predict the ultimate compressive strength (UCS) of consolidated molecular solids, by encoding micrographs with different image feature descriptors and training a random forest regressor, and by training an end-to-end deep-learning (DL) model. Results show that instrument-induced pixel intensity signals can affect ML model predictions in a consistently negative way. As a remedy, we explored intensity normalization techniques. It is seen that intensity normalization helps to improve micrograph data quality and ML model robustness, but microscope-induced intensity variations can be difficult to eliminate.

36 MATERIALS SCIENCE↗

DOE OSTI2022

Robust two-colour pyrometry uncertainty analysis to acquire spatially-resolved measurements

Two-colour pyrometry (2CP) has been used over several decades to study engine-relevant combustion processes, but results are generally regarded as qualitative or semi-quantitative. In many current 2CP systems, large measurement errors are introduced by parallax because the two measured wavelengths are not from the same line of sight. Here this work presents a spatially-resolved 2CP system with the objective of quantifying and reducing measurement uncertainty. An optical setup that eliminates parallax in 2CP is used together with pixel-by-pixel calibration of the camera sensor to increase measurement accuracy. Primary uncertainty terms are identified, and an error propagation analysis is performed to compute uncertainties in the final results of soot temperature, soot concentration parameter, KL, and soot mass. These methodologies are applied to investigate an auto-igniting fuel spray in a constant pressure flow rig at diesel-like conditions of high ambient pressure and temperature. Results show bias uncertainty of around 200 K (≈10%) for temperature and about 40%–60% for KL. High uncertainty was found to occur on the diffusion flame front where both optical thickness and soot concentrations are small. However, these uncertain measurement zones with relatively low soot concentrations contribute minimally to the total soot mass present in the reacting jet during the temporal evolution of the flame.

42 ENGINEERING↗

DOE OSTI2023

Primordial non-Gaussianity with angular correlation function: integral constraint and validation for DES

ABSTRACT Local primordial non-Gaussianity (PNG) is a promising observable of the underlying physics of inflation, characterized by $f_{\rm NL}^{\rm loc}$. We present the methodology to measure $f_{\rm NL}^{\rm loc}$ from the Dark Energy Survey (DES) data using the two-point angular correlation function (ACF) with scale-dependent bias. One of the focuses of the work is the integral constraint (IC). This condition appears when estimating the mean number density of galaxies from the data and is key in obtaining unbiased $f_{\rm NL}^{\rm loc}$ constraints. The methods are analysed for two types of simulations: ∼246 goliat-png N-body small area simulations with fNL equal to −100 and 100, and 1952 Gaussian ice-cola mocks with fNL = 0 that follow the DES angular and redshift distribution. We use the ensemble of goliat-png mocks to show the importance of the IC when measuring PNG, where we recover the fiducial values of fNL within the 1σ when including the IC. In contrast, we found a bias of ΔfNL ∼ 100 when not including it. For a DES-like scenario, we forecast a bias of ΔfNL ∼ 23, equivalent to 1.8σ, when not using the IC for a fiducial value of fNL = 100. We use the ice-cola mocks to validate our analysis in a realistic DES-like set-up finding it robust to different analysis choices: best-fitting estimator, the effect of IC, BAO damping, covariance, and scale choices. We forecast a measurement of fNL within σ(fNL) = 31 when using the DES-Y3 BAO sample, with the ACF in the 1 deg < θ < 20 deg range.

79 ASTRONOMY AND ASTROPHYSICS↗

NASA NTRS1991

INS integrated motion analysis for autonomous vehicle navigation

The use of inertial navigation system (INS) measurements to enhance the quality and robustness of motion analysis techniques used for obstacle detection is discussed with particular reference to autonomous vehicle navigation. The approach to obstacle detection used here employs motion analysis of imagery generated by a passive sensor. Motion analysis of imagery obtained during vehicle travel is used to generate range measurements to points within the field of view of the sensor, which can then be used to provide obstacle detection. Results obtained with an INS integrated motion analysis approach are reviewed.

Roberts, Barry↗

NASA NTRS1994

Robustness of reduced-order multivariable state-space self-tuning controller

In this paper, we present a quantitative analysis of the robustness of a reduced-order pole-assignment state-space self-tuning controller for a multivariable adaptive control system whose order of the real process is higher than that of the model used in the controller design. The result of stability analysis shows that, under a specific bounded modelling error, the adaptively controlled closed-loop real system via the reduced-order state-space self-tuner is BIBO stable in the presence of unmodelled dynamics.

Yuan, Zhuzhi↗