SEARCH · Search NASA

Results for “data representation”

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 1,621 records · Page 90

All sources NASA NTRS DOE OSTI

DOE OSTI2018

Soil moisture estimation using tomographic ground penetrating radar in a MCMC–Bayesian framework

In this paper, we focus on a hydrogeological inverse problem specifically targeting monitoring soil moisture variations using tomographic ground penetrating radar (GPR) travel time data. Technical challenges exist in the inversion of GPR tomographic data for handling non-uniqueness, nonlinearity and high-dimensionality of unknowns. We have developed a new method for estimating soil moisture fields from crosshole GPR data. It uses a pilot-point method to provide a low-dimensional representation of the relative dielectric permittivity field of the soil, which is the primary object of inference: the field can be converted to soil moisture using a petrophysical model. We integrate a multi-chain Markov chain Monte Carlo (MCMC)–Bayesian inversion framework with the pilot point concept, a curved-ray GPR travel time model, and a sequential Gaussian simulation algorithm, for estimating the dielectric permittivity at pilot point locations distributed within the tomogram, as well as the corresponding geostatistical parameters (i.e., spatial correlation range). We infer the dielectric permittivity as a probability density function, thus capturing the uncertainty in the inference. The multi-chain MCMC enables addressing high-dimensional inverse problems as required in the inversion setup. The method is scalable in terms of number of chains and processors, and is useful for computationally demanding Bayesian model calibration in scientific and engineering problems. The proposed inversion approach can successfully approximate the posterior density distributions of the pilot points, and capture the true values. The computational efficiency, accuracy, and convergence behaviors of the inversion approach were also systematically evaluated, by comparing the inversion results obtained with different levels of noises in the observations, increased observational data, as well as increased number of pilot points.

54 ENVIRONMENTAL SCIENCES↗

NASA NTRS1985

The impact of data density and data error on the evolution of mesoscale forecast error

The effects of data density and data errors on the accuracy of mesoscale weather forecasts were assessed by simulations of a period in February 1979 when a snowstorm occurred along the U.S. eastern seaboard. The simulations were initiated every 12 hr and the growth of the rms error was tracked as a function of time and varying data densities. A uniform grid of instrumentation separation and perfect boundary conditions were assumed for all the simulations. A bias error representation was defined to account for systematic measurement errors. A brief summary of features of the somewhat complex storm is provided, together with the parameters of the 11 different simulations with high, medium and low data densities. The accuracy of the forecasts were proportional to the data density, although the error decreased for all the simulations over a 24 hr period. The density of the vertical data had a significant impact on the accuracy of the forecast, whereas the horizontal data density did not. Finally, pathways by which errors were transferred among variables were identified.

Warner, T. T.↗

NASA NTRS2019

Macro-Scale Findings of the DebriSat Debris Field Obtained from X-Rays of the Catch Panels

The DebriSat project analyzes the debris from a hypervelocity impact test conducted on a 56 kg-simulated satellite target in 2014. Its purpose is to generate a comprehensive database of the size, mass, material, shape, and the radar and optical properties of fragments generated from orbital collisions of modern-built spacecraft. Three hundred ninety-one foam catch panels from the DebriSat test have been X-rayed to locate the particles within them for subsequent extraction and assessment. The data from the tens of thousands of extracted particles has heretofore been the focus of the debris assessment. The work presented here connects the X-ray sample location images to reconstruct the 3D distribution of the entire fragment cloud. Although some errors are inherent to the process, this 3D database is valuable for qualitative studies of size and energy distributions. Further, the 3D database can indicate the extent of bending or breaking of the fragments as they decelerated to full capture. By virtue of its full sphere perspective, the 3D database allows graphical insight to the physics of the collision and breakup. This 3D representation has been developed as a “point and click” graphical interface to retrieve the full data set on individual recovered and assessed fragments within the growing database. Overall, this macro-level screening technique has added enormous insight to the experiment in unexpected ways.

Allen, Anthony↗

NASA NTRS2019

Modeling Phenological Controls on Carbon Dynamics in Dryland Sagebrush Ecosystems

Dryland ecosystems play an important role in determining how precipitation anomalies affect terrestrial carbon fluxes at regional to global scales. Thus, to understand how climate change may affect the global carbon cycle, we must also be able to understand and model its effects on dryland vegetation. Dynamic Global Vegetation Models (DGVMs) are an important tool for modeling ecosystem dynamics, but they often struggle to reproduce seasonal patterns of plant productivity. Because the phenological niche of many plant species is linked to both total productivity and competitive interactions with other plants, errors in how process-based models represent phenology hinder our ability to predict climate change impacts. This may be particularly problematic in dryland ecosystems where many species have developed a complex phenology in response to seasonal variability in both moisture and temperature. Here, we examine how uncertainty in key parameters as well as the structure of existing phenology routines affect the ability of a DGVM to match seasonal patterns of leaf area index (LAI) and gross primary productivity (GPP) across a temperature and precipitation gradient. First, we optimized model parameters using a combination of site-level eddy covariance data and remotely-sensed LAI data. Second, we modified the model to include a semi-deciduous phenology type and added flexibility to the representation of grass phenology. While optimizing parameters reduced model bias, the largest gains in model performance were associated with the development of our new representation of phenology. This modified model was able to better capture seasonal patterns of both leaf area index (R(exp 2) = 0.75) and gross primary productivity (R(exp 2) = 0.84), though its ability to estimate total annual GPP depended on using eddy covariance data for optimization. The new model also resulted in a more realistic outcome of modeled competition between grass and shrubs. These findings demonstrate the importance of improving how DGVMs represent phenology in order to accurately forecast climate change impacts in dryland ecosystems.

Ecosystem model↗

DOE OSTI2023

Combining variational autoencoders and physical bias for improved microscopy data analysis *

Electron and scanning probe microscopy produce vast amounts of data in the form of images or hyperspectral data, such as electron energy loss spectroscopy or 4D scanning transmission electron microscope, that contain information on a wide range of structural, physical, and chemical properties of materials. To extract valuable insights from these data, it is crucial to identify physically separate regions in the data, such as phases, ferroic variants, and boundaries between them. In order to derive an easily interpretable feature analysis, combining with well-defined boundaries in a principled and unsupervised manner, here we present a physics augmented machine learning method which combines the capability of variational autoencoders to disentangle factors of variability within the data and the physics driven loss function that seeks to minimize the total length of the discontinuities in images corresponding to latent representations. Our method is applied to various materials, including NiO-LSMO, BiFeO 3 , and graphene. The results demonstrate the effectiveness of our approach in extracting meaningful information from large volumes of imaging data. The customized codes of the required functions and classes to develop phyVAE is available at https://github.com/arpanbiswas52/phy-VAE.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2025

A Digital Twin Framework Utilizing Machine Learning for Robust Predictive Maintenance: Enhancing Tire Health Monitoring

We introduce a novel digital twin (DT) framework for the predictive maintenance of long-term physical systems. Using monitoring tire health as an application, we show how the DT framework can be used to enhance automotive safety and efficiency, and how the technical challenges can be overcome using a three-step approach. First, to manage the data complexity over a long operation span, we employ data reduction techniques to concisely represent physical tires using historical performance and usage data. Relying on these data, for fast real-time prediction, we train a transformer-based model offline on our concise dataset to predict future tire health over time, represented as remaining casing potential (RCP). Based on our architecture, our model quantifies both epistemic and aleatoric uncertainties, providing reliable confidence intervals around predicted RCP. Second, to incorporate real-time data, we update the predictive model in the DT framework, ensuring its accuracy throughout its lifespan with the aid of hybrid modeling and the use of the discrepancy function. Third, to assist decision-making in predictive maintenance, we implement a tire state decision algorithm, which strategically determines the optimal timing for tire replacement based on RCP forecasted by our transformer model. This approach ensures that our DT accurately predicts system health, continually refines its digital representation, and supports predictive maintenance decisions. Furthermore, our framework effectively embodies a physical system, leveraging big data and machine learning (ML) for predictive maintenance, model updates, and decision-making.

advanced computing infrastructure↗

NASA NTRS1991

Identification of observer/Kalman filter Markov parameters: Theory and experiments

An algorithm to compute Markov parameters of an observer or Kalman filter from experimental input and output data is discussed. The Markov parameters can then be used for identification of a state space representation, with associated Kalman gain or observer gain, for the purpose of controller design. The algorithm is a non-recursive matrix version of two recursive algorithms developed in previous works for different purposes. The relationship between these other algorithms is developed. The new matrix formulation here gives insight into the existence and uniqueness of solutions of certain equations and gives bounds on the proper choice of observer order. It is shown that if one uses data containing noise, and seeks the fastest possible deterministic observer, the deadbeat observer, one instead obtains the Kalman filter, which is the fastest possible observer in the stochastic environment. Results are demonstrated in numerical studies and in experiments on an ten-bay truss structure.

Juang, Jer-Nan↗

NASA NTRS1991

Identification of observer/Kalman filter Markov parameters - Theory and experiments

Juang, Jer-Nan↗

DOE OSTI2022

Quantifying Errors in Effective Cluster Interactions of Lattice Gas Cluster Expansions

The promise of lattice gas (LG) cluster expansions (CEs) is that they can describe a given system property to any level of accuracy since the orthogonal “cluster basis functions” span the complete space of available configurations. Such an approach can be constructed to an arbitrarily large surface of a finite number of distinct adsorption sites. Unfortunately, this is only true for the case of an ideal, fixed lattice decorated with components at precise lattice points (the lattice “sites”) with no distortions or relaxations subsequently allowed. Since most systems, and surfaces specifically, do not conform to such an ideal set of constraints, errors in LG CEs must be expected or CE convergence severely hampered. Beyond this, numerical errors in the provided data can complicate the proper construction of a truly predictive and/or physically significant CE. In this work, we show here how reliance on typical statistical tools like confidence intervals cannot be expected to provide an accurate representation of the uncertainty of effective cluster interactions (ECIs) in the CE due to the nature of the target ab initio data and the nature of CEs themselves. We develop a method for estimating these errors that does not rely on statistical assumptions about the model or data. We then use these ECI errors to quantify fundamental consequences on the uncertainty of ECIs in CEs built from O/Fe(100) data whose surface and adsorbates have been allowed to relax in a typical manner and from O/Fe(100) data whose surface and adsorbates are fixed in ideal lattice positions. We also quantify the effect of using a different density functional theory exchange–correlation functional, using these ECI errors to assess the significance in any deviations. In both cases, our method is shown to have remarkable utility in the quantification of errors in the ECIs of CEs. While we stick to the lattice gas convention in this work, the method is equally applicable to the Ising convention or, in principle, any linear model of sufficient complexity.

08 HYDROGEN↗

DOE OSTI2018

Representation and evolution of urban weather boundary conditions in downtown Chicago

Our study presents a novel computing technique for data exchange and coupling between a high-resolution weather simulation model and a building energy model, with a goal of evaluating the impact of urban weather boundary conditions on energy performance of urban buildings. The Weather Research and Forecasting (WRF) model is initialized with the operational High-Resolution Rapid Refresh (HRRR) dataset to provide hourly weather conditions over the Chicago region. We utilize the building footprint, land use, and building stock datasets to generate building energy models using EnergyPlus. We mapped the building exterior surfaces to local air nodes to import simulated microclimate data and to export buildings' heat emissions to their local environment. Preliminary experiments for a test area in Chicago show that predicted building cooling energy use differs by about 4.7% for the selected date when compared with simulations using TMY weather data and without considering the urban microclimate boundary conditions.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

DOE OSTI2024

Reduced‐Order Modeling for Linearized Representations of Microphysical Process Rates

Abstract Representing cloud microphysical processes in large scale atmospheric models is challenging because many processes depend on the details of the droplet size distribution (DSD, the spectrum of droplets with different sizes in a cloud). While full or partial statistical moments of droplet size distributions are the typical variables used in bulk models, prognostic moments are limited in their ability to represent microphysical processes across the range of conditions experienced in the atmosphere. Microphysical parameterizations employing prognostic moments are known to suffer from structural uncertainty in their representations of inherently higher dimensional cloud processes, which limit model fidelity and lead to forecasting errors. Here we investigate how data‐driven reduced‐order modeling can be used to learn predictors for microphysical process rates in bulk microphysics schemes in an unsupervised manner from higher dimensional bin distributions. Using simulations characteristic of marine stratiform clouds, we simultaneously learn lower dimensional representations of droplet size distributions and predict the evolution of the microphysical state of the system. Droplet collision‐coalescence, the main process for generating warm rain, is estimated to have an intrinsic dimension of three. This intrinsic dimension provides a lower limit on the number of degrees of freedom needed to accurately represent collision‐coalescence in models. We demonstrate how deep learning based reduced‐order modeling can be used to discover intrinsic coordinates describing the microphysical state of the system, where process rates such as collision‐coalescence are globally linearized. These implicitly learned representations of the DSD retain more information about the DSD than typical moment‐based representations.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2019

The Case for Custom TMYs: Examples Using the NSRDB: Preprint

A typical meteorological year (TMY) essentially represents an hourly compilation of median months constructed using multiyear data sets. Although TMYs are generated from global horizontal irradiance, they are used in photovoltaic (PV) modeling for various plane-of-array (POA) representations. This paper demonstrates that a POA TMY generated with corresponding POA irradiance time-series information produces significantly different results than a POA TMY generated by transposing a TMY data set. In some months, the differences can be more than 3%. These results point to the need for generating TMYs using POA irradiance time series representing the orientation at which PV panels will be deployed.

14 SOLAR ENERGY↗

NASA NTRS2010

The Sensitivity of Simulated Ocean Biogeochemistry to Forcing Fields Derived from NCEP and MERRA Reanalysis Products

Ocean biogeochemistry models are typically forced by atmospheric and oceanic data derived from reanalysis products. For the NASA Ocean Biogeochemistry Model (NOBM) such reanalysis forcing fields include: surface wind stress, sea surface temperature, ice distributions, shortwave radiation, surface wind speeds and surface atmospheric pressure. Additionally, proper computation of ocean irradiance requires reanalysis products of relative humidity and precipitable water (in addition to aerosol and cloud information which is derived from satellite data). The question posed here is, does the choice of reanalysis products make a difference in the representation of ocean biology and biogeochemistry? NOBM was forced by NCEP and MERRA reanalysis products for the period 2002-2009. We find that in 2009 global distributions and abundances of biological variables (total chlorophyll and nutrients) and carbon (dissolved inorganic and organic carbon and surface pCO2) were similar between the two different forcing fields. Global statistical comparisons with satellite and in situ data also showed negligible differences.

Gregg, Watson↗

DOE OSTI2018

Learning Memento archive routing with Character-based Artificial Neural Networks

This white paper describes a series of tests that were performed to determine if a neural network could learn patterns from a service that maintains a cache of routing decisions for the discovery of version information for discrete web-at-large URLs. Labeled training data was derived from a log file that records the by-archive availability of Memento availability for a given URL. Training data sets were generated from this log file on a by-archive basis (thus making it a binary classification problem). Each training data set consisted of equal numbers of hit and miss URLs, selected at random. The URLs were converted to a normalized numeric representation where each integer in a URL training vector represents a character in that URL. The corresponding label indicates whether or not Mementos were available for that URL in the selected archive. The training data matrix and the label vector became input to a neural network. A number of neural network architectures and network hyperparameters were explored, however the log entries themselves were used as-is, without any feature engineering, beyond the aforementioned normalization.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2022

Mining for Strong Gravitational Lenses with Self-supervised Learning

We employ self-supervised representation learning to distill information from 76 million galaxy images from the Dark Energy Spectroscopic Instrument Legacy Imaging Surveys’ Data Release 9. Targeting the identification of new strong gravitational lens candidates, we first create a rapid similarity search tool to discover new strong lenses given only a single labeled example. We then show how training a simple linear classifier on the self-supervised representations, requiring only a few minutes on a CPU, can automatically classify strong lenses with great efficiency. We present 1192 new strong lens candidates that we identified through a brief visual identification campaign and release an interactive web-based similarity search tool and the top network predictions to facilitate crowd-sourcing rapid discovery of additional strong gravitational lenses and other rare objects: github.com/georgestein/ssl-legacysurvey.

79 ASTRONOMY AND ASTROPHYSICS↗

NASA NTRS1993

Visualizing the semantic content of large text databases using text maps

A methodology for generating text map representations of the semantic content of text databases is presented. Text maps provide a graphical metaphor for conceptualizing and visualizing the contents and data interrelationships of large text databases. Described are a set of experiments conducted against the TIPSTER corpora of Wall Street Journal articles. These experiments provide an introduction to current work in the representation and visualization of documents by way of their semantic content.

Combs, Nathan↗

NASA NTRS2006

Representation of the Geosynchronous Plasma Environment in Spacecraft Charging Calculations

Historically, our ability to predict and postdict spacecraft surface charging has been limited by the characterization of the plasma environment. One difficulty lies in the common practice of fitting the plasma data to a Maxwellian or Double Maxwellian distribution function, which may not represent the data well for charging purposes. We use electron and ion flux spectra measured by the Los Alamos National Laboratory (LANL) Magnetospheric Plasma Analyzer (MPA) to examine how the use of different spectral representations of the charged particle environment in computations of spacecraft potentials during magnetospheric substorms affects the accuracy of the results. We calculate the spacecraft potential using both the measured fluxes and several different fits to these fluxes. These measured fluxes have been corrected for the difference between the measured and calculated potential. The potential computed using the measured fluxes and the best available material properties of graphite carbon, with a secondary electron escape fraction of 81%, is within a factor of three of the measured potential for 87% of the data. Potentials calculated using a Kappa function fit to the incident electron flux distribution function and a Maxwellian function fit to the incident ion flux distribution function agree with measured potentials nearly as well as do potentials calculated using the measured fluxes. Alternative spectral representations gave less accurate estimates of potential. The use of all the components of the net flux, along with spacecraft specific average material properties, gives a better estimate of the spacecraft potential than the high energy flux alone.

Davis, V. A.↗

DOE OSTI2021

Importance of Engineered and Learned Molecular Representations in Predicting Organic Reactivity, Selectivity, and Chemical Properties

Machine-readable chemical structure representations are foundational in all attempts to harness machine learning for the prediction of reactivities, selectivities, and chemical properties directly from molecular structure. The featurization of discrete chemical structures into a continuous vector space is a critical phase undertaken before model selection, and the development of new ways to quantitatively encode molecules is an active area of research. Here, we highlight the application and suitability of different representations, from expert-guided “engineered” descriptors to automatically “learned” features, in different prediction tasks relevant to organic and organometallic chemistry, where differing amounts of training data are available. These tasks include statistical models of stereo- and enantioselectivity, thermochemistry, and kinetics developed using experimental and quantum chemical data. The use of expert-guided molecular descriptors provides an opportunity to incorporate chemical knowledge, domain expertise, and physical constraints into statistical modeling. In applications to stereoselective organic and organometallic catalysis, where data sets may be relatively small and 3D-geometries and conformations play an important role, mechanistically informed features can be used successfully to obtain predictive statistical models that are also chemically interpretable. We provide an overview of several recent applications of this approach to obtain quantitative models for reactivity and selectivity, where topological descriptors, quantum mechanical calculations of electronic and steric properties, along with conformational ensembles, all feature as essential ingredients of the molecular representations used. Alternatively, more flexible, general-purpose molecular representations such as attributed molecular graphs can be used with machine learning approaches to learn the complex relationship between a structure and prediction target. This approach has the potential to out-perform more traditional representation methods such as “hand-crafted” molecular descriptors, particularly as data set sizes grow. One area where this is particularly relevant is in the use of large sets of quantum mechanical data to train quantitative structure–property relationships. A general approach toward curating useful data sets and training highly accurate graph neural network models is discussed in the context of organic bond dissociation enthalpies, where this strategy outperforms regression using precomputed descriptors. Finally, we describe how graph neural network predictions can be incorporated into mechanistically informed statistical models of chemical reactivity and selectivity. Once trained, this approach avoids the expensive computational overhead associated with quantum mechanical calculations, while maintaining chemical interpretability. We illustrate examples for which fast predictions of bond dissociation enthalpy and of the identities of radicals formed through cleavage of a molecule’s weakest bond are used in simple physical models of site-selectivity and reactivity.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗