Search NASA⌕ Search

SEARCH · Search NASA

Results for “box model”

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 487 records · Page 27
All sourcesNASA NTRSDOE OSTI
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↗
DOE OSTI2026

Building Operation Model (Morpheus) for Dallas Fort Worth Airport (CRADA CRD-19-16301 Final Report)

The primary objective of this project was to leverage digital twin technology to enhance the design and operation of DFW Airport terminals and their associated energy systems. To achieve this, Morpheus, a building digital twin, was developed to guide improvements in airport operations, specifically targeting reductions in peak power demand and overall energy consumption.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2021

Neural-network model for force prediction in multi-principal-element alloys

Atomistic simulations can provide useful insights into the physical properties of multi-principal-element alloys. However, classical potentials mostly fail to capture key quantum (electronic-structure) effects. Here we present a deep 3D convolutional neural network (3D CNN) based framework combined with a voxelization technique to design interatomic potentials for chemically complex alloys. We highlight the performance of the 3D CNN model and its efficacy in computing potentials using the medium-entropy alloy TaNbMo. In order to provide insights into the effect of voxel resolution, we implemented two approaches based on the inner and outer bounding boxes. An efficient 3D CNN model, which is as accurate as the density-functional theory (DFT) approach, for calculating potentials will provide a promising schema for accurate atomistic simulations of structure and dynamics of general multi-principle element alloys.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Black-Box Neural System Identification and Differentiable Programming to Improve Earth System Model Predictions February

Focal Area(s): Focal Area 2: AI for predictive modeling, including AI-derived model components, and AI-enabled applications including parameter optimization, data assimilation, and uncertainty quantification. Focal Area 1: AI-assisted data assimilation using numerical Earth system models. Science Challenge: Earth system models have structural biases that lead to uncertain predictions, and their complexity and expense makes it difficult constraining the models with data or improved physical understanding.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2022

Integrated Optimization and Control of a Hybrid Gas Turbine/sCO 2 Power System

During phase-I, the project team led by Echogen Power Systems (EPS) had two primary objectives based on investigating the application of gas turbines with supercritical carbon dioxide (sCO 2 ) power cycles. The first objective was to improve the overall efficiency and performance of a hybrid gas turbine/sCO 2 power system through a joint optimization of the two subsystems (gas turbine and sCO 2 power cycle) using non-linear optimization techniques that simultaneously evaluate thermal performance of the combined cycle. The hybrid power system included several points of interaction, including (but not limited to) gas turbine exhaust, fuel heating, inlet chilling and turbine cooling. The second objective was to establish a baseline transient response model of the hybrid power system and a notional microgrid and begin steps to integrate the control systems of the three major elements (gas turbine, sCO 2 cycle and grid controller). The project team established a baseline performance for a combined cycle power plant using a production gas turbine and scaled sCO 2 power cycle only utilizing exhaust heat recovery. Echogen’s non-linear techno-economic optimization code was extended by adding gas turbine component models derived from a in-house developed gas turbine design code. With the two cycles coupled by the gas turbine exhaust, design parameters of both cycles were allowed to vary simultaneously to determine performance opportunity versus isolated designs. Returning to the baseline gas turbine/sCO 2 power cycle transient models: Echogen had in-house developed sCO 2 cycle transient model in GT-Suite system simulation software, and had partnered with Siemens Finspång for gas turbine transient model, and Siemens PTI group to provide micro-grid load profile as well as hybrid power cycle generated load (power and frequency) analysis. The transient model for the SGT-750 Siemens gas turbine was a “black-box” functional mock-up interface (FMI) model developed by Siemens Industrial Turbomachinery in Finspång, Sweden. The SGT-750 is a twin-shaft gas turbine that produces 40 MW electricity with an efficiency of about 40% at ISO conditions. At 100% gas turbine throttle (load), the SGT-750 has average exhaust conditions of 114.6 kg/s and 469.8°C. The transient model for sCO 2 power cycle was developed by Echogen in GT-SUITE 1D system simulation software platform. The basic CO 2 flow circuit has single-shaft turbomachinery with net 11.5 MW electrical power output at design conditions. The power turbine has a double-ended shaft with one end connected to synchronous generator through a fixed-ratio gearbox. The other end of power turbine is connected to the compressor through a continuously variable transmission. The major components of the sCO 2 power cycle modeled include air cooled condenser/cooler, CO 2 compressor, recuperator, two waste heat exchanger coils, power turbine, continuous variable transmission, gearbox and generator. Integration of SGT-750 transient model and sCO 2 power cycle transient model was done in Matlab Simulink. In the integrated model, the gas turbine and sCO 2 power cycle interacted at two points, first one being the gas turbine exhaust gas flow rate and temperature, which were inputs to sCO 2 power cycle model. The second point was the distribution of micro-grid load demand signal between the SGT-750 generator and sCO 2 cycle generator. For a given combined-cycle load demand, the gas turbine load demand was equal to the total demand minus the sCO 2 cycle power generated. In the present study the integrated model was simulated for two cases of grid load demand: (i) for a step change, both positive-step and negative-step, in grid load demand (ii) for a micro-grid load demand curve provided by Siemens PTI group. Finally, the time series plots representing load demand versus integrated system response were presented including the sCO 2 power cycle control system performance plots. The actual generated power and frequency of both the generators, gas turbine and sCO 2 power cycle, was supplied to Siemens PTI group for dynamic grid assessment, results of which are provided in appendices.

03 NATURAL GAS↗
DOE OSTI2020

A Climatology and Extreme Value Analysis of Large Hail in China

In this paper, hail size records are analyzed at 2254 stations in China and a hail size climatology is developed based on gridded hail observations for the period 1960–2015. It is found that the annual percentiles of hail size records changed sharply and national-wide after 1980, therefore two periods, 1960–79 and 1980–2015, are studied. There are some similarities between the two periods in terms of the characteristics of hail size such as the spatial distribution patterns of mean annual maximum hail size and occurrence week of annual maximum hail size. The 1980–2015 period had higher observation density than the 1960–79 period, but showed smaller mean annual maximum hail size, especially in northern China. In the majority of grid boxes, the annual maximum hail size experienced a decreasing trend during the 1980–2015 period. A Gumbel extreme value model is fitted to each grid box to estimate the return periods of maximum hail size. The scale and location parameter of the fitted Gumbel distributions are higher in eastern China than in western China, thereby reflecting a greater likelihood of large hail in eastern China. In southern China, the maximum hail size exceeds 127 mm for a 10-yr return period, whereas in northern China maximum hail size exceeds this threshold for a 50-yr return period. Furthermore, the Gumbel model is found to potentially underestimate the maximum hail size for certain return periods, but provides a more informed picture of the spatial distribution of extreme hail size and the regional differences.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Sampling Low-Dimensional Markovian Dynamics for Preasymptotically Recovering Reduced Models from Data with Operator Inference

This work introduces a method for learning low-dimensional models from data of high-dimensional black-box dynamical systems. The novelty is that the learned models are exactly the reduced models that are traditionally constructed with classical projection-based model reduction techniques. Thus, the proposed approach learns models that are guaranteed to have the well-studied properties of reduced models known from model reduction, without requiring full knowledge of the governing equations and without requiring the operators of the high-dimensional systems. The key ingredient is a new data sampling scheme to obtain re-projected trajectories of high-dimensional systems that correspond to Markovian dynamics in low-dimensional subspaces. The exact recovery of reduced models from these re-projected trajectories is guaranteed pre-asymptotically under certain conditions for finite amounts of data and for a large class of systems with polynomial nonlinear terms. Numerical results demonstrate that the low-dimensional models learned with the proposed approach match reduced models from traditional model reduction up to numerical errors in practice. In conclusion, the numerical results further indicate that low-dimensional models fitted to re-projected trajectories are predictive even in situations where models fitted to trajectories without re-projection are inaccurate and unstable.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Moving beyond post hoc explainable artificial intelligence: a perspective paper on lessons learned from dynamical climate modeling

AI models are criticized as being black boxes, potentially subjecting climate science to greater uncertainty. Explainable artificial intelligence (XAI) has been proposed to probe AI models and increase trust. In this review and perspective paper, we suggest that, in addition to using XAI methods, AI researchers in climate science can learn from past successes in the development of physics-based dynamical climate models. Dynamical models are complex but have gained trust because their successes and failures can sometimes be attributed to specific components or sub-models, such as when model bias is explained by pointing to a particular parameterization. We propose three types of understanding as a basis to evaluate trust in dynamical and AI models alike: (1) instrumental understanding, which is obtained when a model has passed a functional test; (2) statistical understanding, obtained when researchers can make sense of the modeling results using statistical techniques to identify input–output relationships; and (3) component-level understanding, which refers to modelers' ability to point to specific model components or parts in the model architecture as the culprit for erratic model behaviors or as the crucial reason why the model functions well. We demonstrate how component-level understanding has been sought and achieved via climate model intercomparison projects over the past several decades. Such component-level understanding routinely leads to model improvements and may also serve as a template for thinking about AI-driven climate science. Currently, XAI methods can help explain the behaviors of AI models by focusing on the mapping between input and output, thereby increasing the statistical understanding of AI models. Yet, to further increase our understanding of AI models, we will have to build AI models that have interpretable components amenable to component-level understanding. We give recent examples from the AI climate science literature to highlight some recent, albeit limited, successes in achieving component-level understanding and thereby explaining model behavior. The merit of such interpretable AI models is that they serve as a stronger basis for trust in climate modeling and, by extension, downstream uses of climate model data.

54 ENVIRONMENTAL SCIENCES↗
NASA NTRS1994

Radiative force model performance for TOPEX/Poseidon precision orbit determination

The TOPEX/Poseidon spacecraft was launched to study the Earth's oceans. To maximize the benefit from the alimetric data collected, mission requirements dictate that TOPEX/Poseidon's orbit must be computed at extremely high accuracy. To meet these demands, a nonconservative force model which accounts for the satellite's complex geometry, attitude and surface properties has been developed. The 'box-wing' representation treats the spacecraft as the combination of flat plates arranged in the shape of a box and a connected solar array. Model performance and parameter sensitivities are discussed.

TRT-THEORETICAL↗
NASA NTRS1992

Nonconservative force model parameter estimation strategy for TOPEX/Poseidon precision orbit determination

The TOPEX/Poseidon spacecraft was launched on August 10, 1992 to study the Earth's oceans. To achieve maximum benefit from the altimetric data it is to collect, mission requirements dictate that TOPEX/Poseidon's orbit must be computed at an unprecedented level of accuracy. To reach our pre-launch radial orbit accuracy goals, the mismodeling of the radiative nonconservative forces of solar radiation, Earth albedo an infrared re-radiation, and spacecraft thermal imbalances cannot produce in combination more than a 6 cm rms error over a 10 day period. Similarly, the 10-day drag modeling error cannot exceed 3 cm rms. In order to satisfy these requirements, a 'box-wing' representation of the satellite has been developed in which, the satellite is modelled as the combination of flat plates arranged in the shape of a box and a connected solar array. The radiative/thermal nonconservative forces acting on each of the eight surfaces are computed independently, yielding vector accelerations which are summed to compute the total aggregate effect on the satellite center-of-mass. Select parameters associated with the flat plates are adjusted to obtain a better representation of the satellite acceleration history. This study analyzes the estimation of these parameters from simulated TOPEX/Poseidon laser data in the presence of both nonconservative and gravity model errors. A 'best choice' of estimated parameters is derived and the ability to meet mission requirements with the 'box-wing' model evaluated.

Luthcke, S. B.↗
NASA NTRS2011

Graphics Processing Unit (GPU) Acceleration of the Goddard Earth Observing System Atmospheric Model

The Goddard Earth Observing System 5 (GEOS-5) is the atmospheric model used by the Global Modeling and Assimilation Office (GMAO) for a variety of applications, from long-term climate prediction at relatively coarse resolution, to data assimilation and numerical weather prediction, to very high-resolution cloud-resolving simulations. GEOS-5 is being ported to a graphics processing unit (GPU) cluster at the NASA Center for Climate Simulation (NCCS). By utilizing GPU co-processor technology, we expect to increase the throughput of GEOS-5 by at least an order of magnitude, and accelerate the process of scientific exploration across all scales of global modeling, including: The large-scale, high-end application of non-hydrostatic, global, cloud-resolving modeling at 10- to I-kilometer (km) global resolutions Intermediate-resolution seasonal climate and weather prediction at 50- to 25-km on small clusters of GPUs Long-range, coarse-resolution climate modeling, enabled on a small box of GPUs for the individual researcher After being ported to the GPU cluster, the primary physics components and the dynamical core of GEOS-5 have demonstrated a potential speedup of 15-40 times over conventional processor cores. Performance improvements of this magnitude reduce the required scalability of 1-km, global, cloud-resolving models from an unfathomable 6 million cores to an attainable 200,000 GPU-enabled cores.

Putnam, Williama↗
NASA NTRS1973

Modelling a particular class of stochastic systems

In this paper a method is given for obtaining a mathematical model of a class of black boxes having multiple inputs and multiple outputs in terms of Ito stochastic integral equations. This method is applicable to the class of black boxes having ergodic correlation functions when there is zero applied input. The point of view adopted in this paper is phenomenological in that it is desired that calculations made using the mathematical model should be 'close' to what is actually observed at the output of the black box.

Eyman, E. D.↗
NASA NTRS2017

Optimization of Debris Shields on the NISAR Mission’s L-Band Radar Instrument

The NASA-ISRO Synthetic Aperture Radar (NISAR) space mission is a collaboration between NASA and the Indian Space Research Organization (ISRO), launching in the 2020s to a polar orbit of 747km altitude. The mission will provide spatial and temporal measurements of land surface changes (e.g. ice sheets, vegetation, earthquakes). Many of the SAR electronics boxes are mounted on the exterior of the structure. Their singlewall box lids efficiently radiate heat for thermal control, but are not very efficient debris shields. The initial design showed an unacceptably high impact risk as estimated with NASA’s ORDEM3 debris model and Bumper impact analysis tool. Each box has a different role in instrument functionality, and this was captured in a reliability model used to optimize the distribution of shield mass among the boxes: total added mass was minimized while maintaining a threshold of functionality and survival probability that was acceptable to the project.

Chinn, James Z.↗
DOE OSTI2023

Uncertainty quantification of a deep learning fuel property prediction model

Deep learning models are being widely used in the field of combustion. Given the black-box nature of typical neural network based models, uncertainty quantification (UQ) is critical to ensure the reliability of predictions as well as the training datasets, and for a principled quantification of noise and its various sources. Deep learning surrogate models for predicting properties of chemical compounds and mixtures have been recently shown to be promising for enabling data-driven fuel design and optimization, with the ultimate goal of improving efficiency and lowering emissions from combustion engines. In this study, UQ is performed for a multi-task deep learning model that simultaneously predicts the research octane number (RON), Motor Octane Number (MON), and Yield Sooting Index (YSI) of pure components and multicomponent blends. The deep learning model is comprised of three smaller networks: Extractor 1, Extractor 2, and Predictor, and a mixing operator. The molecular fingerprints of individual components are encoded via Extractor 1 and Extractor 2, the mixing operator generates fingerprints for mixtures/blends based on linear mixing operation, and the predictor maps the fingerprint to the target properties. Two different classes of UQ methods, Monte Carlo ensemble methods and Bayesian neural networks (BNNs), are employed for quantifying the epistemic uncertainty. Combinations of Bernoulli and Gaussian distributions with DropConnect and DropOut techniques are explored as ensemble methods. All the DropConnect, DropOut and Bayesian layers are applied to the predictor network. Aleatoric uncertainty is modeled by assuming that each data point has an independent uncertainty associated with it. The results of the UQ study are further analyzed to compare the performance of BNN and ensemble methods. Although this study is confined to UQ of fuel property prediction, the methodologies are applicable to other deep learning frameworks that are being widely used in the combustion community.

33 ADVANCED PROPULSION SYSTEMS↗
DOE OSTI2023

Learning effective stochastic differential equations from microscopic simulations: Linking stochastic numerics to deep learning

We identify effective stochastic differential equations (SDEs) for coarse observables of fine-grained particle- or agent-based simulations; these SDEs then provide useful coarse surrogate models of the fine scale dynamics. We approximate the drift and diffusivity functions in these effective SDEs through neural networks, which can be thought of as effective stochastic ResNets. The loss function is inspired by, and embodies, the structure of established stochastic numerical integrators (here, Euler–Maruyama and Milstein); our approximations can thus benefit from backward error analysis of these underlying numerical schemes. They also lend themselves naturally to “physics-informed” gray-box identification when approximate coarse models, such as mean field equations, are available. Existing numerical integration schemes for Langevin-type equations and for stochastic partial differential equations can also be used for training; we demonstrate this on a stochastically forced oscillator and the stochastic wave equation. Our approach does not require long trajectories, works on scattered snapshot data, and is designed to naturally handle different time steps per snapshot. We consider both the case where the coarse collective observables are known in advance, as well as the case where they must be found in a data-driven manner.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS1994

Static aeroelastic analysis of wings using Euler/Navier-Stokes equations coupled with improved wing-box finite element structures

Strong interactions between flow about an aircraft wing and the wing structure can result in aeroelastic phenomena which significantly impact aircraft performance. Time-accurate methods for solving the unsteady Navier-Stokes equations have matured to the point where reliable results can be obtained with reasonable computational costs for complex non-linear flows with shock waves, vortices and separations. The ability to combine such a flow solver with a general finite element structural model is key to an aeroelastic analysis in these flows. Earlier work involved time-accurate integration of modal structural models based on plate elements. A finite element model was developed to handle three-dimensional wing boxes, and incorporated into the flow solver without the need for modal analysis. Static condensation is performed on the structural model to reduce the structural degrees of freedom for the aeroelastic analysis. Direct incorporation of the finite element wing-box structural model with the flow solver requires finding adequate methods for transferring aerodynamic pressures to the structural grid and returning deflections to the aerodynamic grid. Several schemes were explored for handling the grid-to-grid transfer of information. The complex, built-up nature of the wing-box complicated this transfer. Aeroelastic calculations for a sample wing in transonic flow comparing various simple transfer schemes are presented and discussed.

Guruswamy, Guru P.↗
NASA NTRS1997

Application of Interface Technology in Nonlinear Analysis of a Stitched/RFI Composite Wing Stub Box

A recently developed interface technology was successfully employed in the geometrically nonlinear analysis of a full-scale stitched/RFI composite wing box loaded in bending. The technology allows mismatched finite element models to be joined in a variationally consistent manner and reduces the modeling complexity by eliminating transition meshing. In the analysis, local finite element models of nonlinearly deformed wide bays of the wing box are refined without the need for transition meshing to the surrounding coarse mesh. The COMET-AR finite element code, which has the interface technology capability, was used to perform the analyses. The COMET-AR analysis is compared to both a NASTRAN analysis and to experimental data. The interface technology solution is shown to be in good agreement with both. The viability of interface technology for coupled global/local analysis of large scale aircraft structures is demonstrated.

Wang, John T.↗
NASA NTRS2014

Aerostructural Level Set Topology Optimization for a Common Research Model Wing

The purpose of this work is to use level set topology optimization to improve the design of a representative wing box structure for the NASA common research model. The objective is to minimize the total compliance of the structure under aerodynamic and body force loading, where the aerodynamic loading is coupled to the structural deformation. A taxi bump case was also considered, where only body force loads were applied. The trim condition that aerodynamic lift must balance the total weight of the aircraft is enforced by allowing the root angle of attack to change. The level set optimization method is implemented on an unstructured three-dimensional grid, so that the method can optimize a wing box with arbitrary geometry. Fast matching and upwind schemes are developed for an unstructured grid, which make the level set method robust and efficient. The adjoint method is used to obtain the coupled shape sensitivities required to perform aerostructural optimization of the wing box structure.

Dunning, Peter D.↗