SEARCH · Search NASA

Results for “conditional generative models”

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 55 records · Page 3

All sources NASA NTRS DOE OSTI

DOE OSTI2022

Operation-adversarial scenario generation

This paper proposes a modified conditional generative adversarial network (cGAN) model to generate net load scenarios for power systems that are statistically credible, conditioned by given labels (e.g., seasons), and, at the same time, “stressful” to the system operations and dispatch decisions. The measure of stress used in this paper is based on the operating cost increases due to net load changes. The proposed operation-adversarial cGAN (OA-cGAN) internalizes a DC optimal power flow model and seeks to maximize the operating cost and achieve a worst-case data generation. The training and testing stages employed in the proposed OA-cGAN use historical day-ahead net load forecast errors and has been implemented for the realistic NYISO 11-zone system. In conclusion, our numerical experiments demonstrate that the generated operation-adversarial forecast errors lead to more cost-effective and reliable dispatch decisions.

42 ENGINEERING↗

DOE OSTI2021

Machine Learning Augmented Predictive and Generative Model for Rupture Life in Ferritic and Austenitic Steels

The Larson-Miller parameter (LMP) offers an efficient and fast scheme to estimate the creep rupture life of alloy materials for high temperature applications. However, owing to poor generalizability and dependence on the constant C, which is typically not known a-priori, estimations using the Larson-Miller parameter often result in suboptimal performance for a wide range of materials. At best it is useful in comparing alloys of similar composition. In this work, three machine learning (ML) schemes were developed for rupture life prediction for 9-12% Cr ferritic-martensitic steels and austenitic stainless steels, i.e., a hierarchical model to parameterize LMP using the LMP constant C to compute rupture life, a hierarchical model to parameterize both C and LMP to compute rupture life, and a direct prediction of rupture life. Specifically, we show that the third scheme, using a gradient boosting algorithm, can be used to train ML models for very accurate prediction of rupture life in a variety of alloys (Pear-son Correlation Coefficient > 0.9 for 9-12% Cr and > 0.8 for austenitic stainless steels). In addition, the Shapley value was used to quantify feature importance, making the model interpretable by identifying the effect of various features on the model performance. Furthermore, a variational autoencoder-based generative model was built by conditioning on the experimental dataset to sample hypothetical synthetic candidate alloys from the learnt joint distribution not existing in both 9-12 % Cr ferritic-martensitic steel and austenitic stainless steel datasets. Finally, based on the predictive and generative model, a reinforcement learning strategy has been proposed to guide experimentalists into designing better heat resistant alloys.

Mamun, Md Osman G.↗

DOE OSTI2022

‘Flux+Mutability’: a conditional generative approach to one-class classification and anomaly detection

Abstract Anomaly Detection is becoming increasingly popular within the experimental physics community. At experiments such as the Large Hadron Collider, anomaly detection is growing in interest for finding new physics beyond the Standard Model. This paper details the implementation of a novel Machine Learning architecture, called Flux+Mutability, which combines cutting-edge conditional generative models with clustering algorithms. In the ‘flux’ stage we learn the distribution of a reference class. The ‘mutability’ stage at inference addresses if data significantly deviates from the reference class. We demonstrate the validity of our approach and its connection to multiple problems spanning from one-class classification to anomaly detection. In particular, we apply our method to the isolation of neutral showers in an electromagnetic calorimeter and show its performance in detecting anomalous dijets events from standard QCD background. This approach limits assumptions on the reference sample and remains agnostic to the complementary class of objects of a given problem. We describe the possibility of dynamically generating a reference population and defining selection criteria via quantile cuts. Remarkably this flexible architecture can be deployed for a wide range of problems, and applications like multi-class classification or data quality control are left for further exploration.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

NASA NTRS2020

Aircraft Engine Run-To-Failure Data Set Under Real Flight Conditions

The generation of data-driven prognostics models requires the availability of datasets with run-to-failure trajectories. In order to contribute to the development of these methods, the dataset provides a new realistic dataset of run-to-failure trajectories for a small fleet of aircraft engines under realistic flight conditions. The damage propagation modelling used for the generation of this synthetic dataset builds on the modelling strategy from previous work [1] and incorporates two new levels of fidelity. First, it considers real flight conditions as recorded on board of a commercial jet [2]. Secondly, it extends the degradation modelling by relating the degradation process to the operation history. The dataset was generated with the Commercial Modular Aero-Propulsion System Simulation (C-MAPSS) dynamical model [3]. More details about the generation process can be found in [4].

CMAPSS↗

NASA NTRSDate not supplied

Selection of Next Priority IMPACT Medical Conditions Based on Available Terrestrial and Spaceflight Data

BACKGROUND: As the era of exploration class missions begins, identification of medical conditions that may occur and require management becomes essential for the modeling of medical risk. To this end, NASA has developed IMPACT (Informing Mission Planning via Analysis of Complex Tradespaces), a suite of tools to assist in assessment of medical risk analysis. It has incorporated an initial list of the 120 conditions of highest concern, labeled the IMPACT condition list 1.0 (ICL 1.0). This abstract describes a method for prioritizing the 92 conditions included on the Proposed Future Conditions List (PFCL) for inclusion in future iterations of the ICL. OVERVIEW: To construct the Prioritized Proposed Future Conditions List (P-PFCL), each condition on the PFCL was scored as “low,” “medium,” or “high” on each of four variables: incidence, likelihood of significant task impairment, diagnostic and treatment complexity, and treatment futility. Qualitative assessment using clinical judgement was utilized to score complexity, futility, and likelihood of impairment. Incidence was assessed quantitatively using spaceflight data and/or analog populations where available then assigned a score using established cutoffs. Logarithmic numerical values were assigned to each category label. A Prioritization Score was generated for each condition by taking the product of incidence and likelihood of task impairment (risk) divided by the product of complexity and futility (difficulty of care), with higher values corresponding to higher priority for future inclusion in the ICL. DISCUSSION: The described methods allow for the generation of a ranked P-PFCL to act as a decision support tool for selection of the next generation of modeled medical conditions. Some of the conditions ranked highly on the P-PFCL include EVA-related upper and lower extremity sprain/strain, iron deficiency, delirium, and hypertension, among others. While this effort does not attempt to quantify the absolute risk associated with each condition, it does attempt to semi-quantitatively estimate the risk of each condition relative to the other possible conditions. This tool in concert with subject matter expert opinion could optimize the future use of limited resources thereby producing a more accurate medical risk model, which will be essential to the upcoming exploration class missions.

Michael Pohlen↗

DOE OSTI2015

An improved hindcast approach for evaluation and diagnosis of physical processes in global climate models

Here, we present an improved procedure of generating initial conditions (ICs) for climate model hindcast experiments with specified sea surface temperature and sea ice. The motivation is to minimize errors in the ICs and lead to a better evaluation of atmospheric parameterizations' performance in the hindcast mode. We apply state variables (horizontal velocities, temperature and specific humidity) from the operational analysis/reanalysis for the atmospheric initial states. Without a data assimilation system, we apply a two-step process to obtain other necessary variables to initialize both the atmospheric (e.g., aerosols and clouds) and land models (e.g., soil moisture). First, we nudge only the model horizontal velocities towards operational analysis/reanalysis values, given a 6-hour relaxation time scale, to obtain all necessary variables. Compared to the original strategy in which horizontal velocities, temperature and specific humidity are nudged, the revised approach produces a better representation of initial aerosols and cloud fields which are more consistent and closer to observations and model's preferred climatology. Second, we obtain land ICs from an offline land model simulation forced with observed precipitation, winds, and surface fluxes. This approach produces more realistic soil moisture in the land ICs. With this refined procedure, the simulated precipitation, clouds, radiation, and surface air temperature over land are improved in the Day 2 mean hindcasts. Following this procedure, we propose a “Core” integration suite which provides an easily repeatable test allowing model developers to rapidly assess the impacts of various parameterization changes on the fidelity of modelled cloud-associated processes relative to observations.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2025

Latent diffusion can map beam loss to two-dimensional phase-space projections

Beam loss monitors (BLMs) and beam current monitors (BCMs) are ubiquitous at particle accelerators around the world. These simple devices provide noninvasive high-level beam measurements but give no insight into the detailed 6D (𝑥,𝑦,𝑧,𝑝 𝑥 ,𝑝 𝑦 ,𝑝 𝑧 ) beam phase-space distributions or dynamics. We show that generative conditional latent diffusion models can learn intricate patterns to solve the extreme inverse problem of mapping waveforms of tens of BLMs or BCMs along an accelerator to detailed 2D projections of a charged particle beam’s 6D phase-space density. This transformational method can be used at any particle accelerator to transform simple noninvasive devices into detailed beam phase-space diagnostics. We demonstrate this concept via multiparticle simulations of the high-intensity beam in the kilometer-long Los Alamos Neutron Science Center linear proton accelerator.

43 PARTICLE ACCELERATORS↗

NASA NTRS1987

A linearized Euler analysis of unsteady flows in turbomachinery

A method for calculating unsteady flows in cascades is presented. The model, which is based on the linearized unsteady Euler equations, accounts for blade loading shock motion, wake motion, and blade geometry. The mean flow through the cascade is determined by solving the full nonlinear Euler equations. Assuming the unsteadiness in the flow is small, then the Euler equations are linearized about the mean flow to obtain a set of linear variable coefficient equations which describe the small amplitude, harmonic motion of the flow. These equations are discretized on a computational grid via a finite volume operator and solved directly subject to an appropriate set of linearized boundary conditions. The steady flow, which is calculated prior to the unsteady flow, is found via a Newton iteration procedure. An important feature of the analysis is the use of shock fitting to model steady and unsteady shocks. Use of the Euler equations with the unsteady Rankine-Hugoniot shock jump conditions correctly models the generation of steady and unsteady entropy and vorticity at shocks. In particular, the low frequency shock displacement is correctly predicted. Results of this method are presented for a variety of test cases. Predicted unsteady transonic flows in channels are compared to full nonlinear Euler solutions obtained using time-accurate, time-marching methods. The agreement between the two methods is excellent for small to moderate levels of flow unsteadiness. The method is also used to predict unsteady flows in cascades due to blade motion (flutter problem) and incoming disturbances (gust response problem).

Hall, Kenneth C.↗

NASA NTRS1989

Calculation of unsteady flows in turbomachinery using the linearized Euler equations

Hall, Kenneth C.↗

NASA NTRS1991

The role of the sea-surface temperature distribution on numerically simulated cyclogenesis during ERICA

The goal was to quantify the extent to which a sea surface temperature (SST) front can influence cyclogenesis. The approach was to use the Drexel Limited-Area Mesoscale Prediction System (LAMPS) dynamical model to simulate cyclogenesis over various SST fields. Research during the past year focused on the development and testing of a four dimensional data assimilation (FDDA) technique within LAMPS. The technique is a continuous dynamical assimilation where forcing terms are added to the governing model equations to gradually nudge the model solution toward a gridded analysis. Here, the nudging is used as a dynamic initialization tool during a 12 hour preforecast to generate model balanced initial conditions for a subsequent 24 hour numerical prediction. Tests were performed to determine which variables to nudge and how to specify the four dimensional weighting function used to scale the nudging terms. To date, optimal results were obtained by nudging the u and v components of the wind along with the potential temperature. The weighting function ranged from 0 to 1 and varies in time as a quadratic polynomial. It was initialized at 0, reached its maximum at 9 hours into the preforecast, and fell back at 0 to 12 hours. The nudging terms are included in the model equations for all grid points except those within the model predicted oceanic boundary layer. This design attempts to confine changes imposed by the specified SST field to the oceanic boundary layer during the preforecast period.

Lapenta, William M.↗

NASA NTRS1992

Lunar PMAD technology assessment

This report documents an initial set of power conditioning models created to generate 'ballpark' power management and distribution (PMAD) component mass and size estimates. It contains converter, rectifier, inverter, transformer, remote bus isolator (RBI), and remote power controller (RPC) models. These models allow certain studies to be performed; however, additional models are required to assess a full range of PMAD alternatives. The intent is to eventually form a library of PMAD models that will allow system designers to evaluate various power system architectures and distribution techniques quickly and consistently. The models in this report are designed primarily for space exploration initiative (SEI) missions requiring continuous power and supporting manned operations. The mass estimates were developed by identifying the stages in a component and obtaining mass breakdowns for these stages from near term electronic hardware elements. Technology advances were then incorporated to generate hardware masses consistent with the 2000 to 2010 time period. The mass of a complete component is computed by algorithms that calculate the masses of the component stages, control and monitoring, enclosure, and thermal management subsystem.

Metcalf, Kenneth J.↗

NASA NTRS2014

Simulation of VSPT Experimental Cascade Under High and Low Free-Stream Turbulence Conditions

Variable-Speed Power Turbines (VSPT) for rotorcraft applications operate at low Reynolds number and over a wide range in incidence associated with shaft speed change. A comprehensive linear cascade data set obtained includes the effects of Reynolds number, free-stream turbulence and incidence is available and this paper concerns itself with the presentation and numerical simulation of conditions resulting in a selected set of those data. As such, post-dictions of blade pressure loading, total-pressure loss and exit flow angles under conditions of high and low turbulence intensity for a single Reynolds number are presented. Analyses are performed with the three-equation turbulence models of Walters-Leylek and Walters and Cokljat. Transition, loading, total-pressure loss and exit angle variations are presented and comparisons are made with experimental data as available. It is concluded that at the low freestream turbulence conditions the Walters-Cokljat model is better suited to predictions while for high freestream conditions the two models generate similar predications that are generally satisfactory.

low pressure trubine↗

NASA NTRS2015

Simulation of VSPT Experimental Cascade Under High and Low Free-Stream Turbulence Conditions

Variable-Speed Power Turbines (VSPT) for rotorcraft applications operate at low Reynolds number and over a wide range in incidence associated with shaft speed change. A comprehensive linear cascade data set obtained includes the effects of Reynolds number, free-stream turbulence and incidence is available and this paper concerns itself with the presentation and numerical simulation of conditions resulting in a selected set of those data. As such, post-dictions of blade pressure loading, total-pressure loss and exit flow angles under conditions of high and low turbulence intensity for a single Reynolds number are presented. Analyses are performed with the three-equation turbulence models of Walters- Leylek and Walters and Cokljat. Transition, loading, total-pressure loss and exit angle variations are presented and comparisons are made with experimental data as available. It is concluded that at the low freestream turbulence conditions the Walters-Cokljat model is better suited to predictions while for high freestream conditions the two models generate similar predications that are generally satisfactory.

turbulence modeling↗

NASA NTRS2004

Ejecta Generation and Redistribution on 433 Eros: Modeling Ejecta Launch Conditions

The NEAR-Shoemaker mission to asteroid 433 Eros presents an unprecedented opportunity to gain fundamental new knowledge about the processes governing regolith formation and redistribution on small bodies. NEAR-Shoemaker’s high-resolution imaging of the surface of Eros makes the asteroid a valuable and heretofore unparalleled laboratory for the detailed study of impact ejecta reaccretion and regolith redistribution on low-gravity (of order 10 -3 g) objects. Regolith is produced on asteroids by impact cratering, and the existence of regolith on the smallest solar system bodies supports the view that some of the ejecta from impact events on such objects may be retained. Impact craters and retained ejecta on low-gravity objects like Eros represent valuable natural laboratories for evaluating various models of impact cratering processes, since they may present crater structures or ejecta features that either do not form or are hidden on higher-gravity bodies like the Moon. Further, quantifying the extent to which impact processes generate and redistribute regoliths on small body surfaces (excavation depths, retained fraction, turnover timescales, etc.) is pivotal to the issue of how to relate meteoritical samples to their asteroidal parent bodies when surficial processes ( i.e., “space weathering”) may disguise or cover up underlying material and confound the ability of remote sensing techniques to provide reliable mineralogical assays of the parent objects. The rich variety of data on Eros’ regolith properties and distribution returned by NEAR-Shoemaker now require detailed analysis in order to take full advantage of the clues these observations offer for elucidating details of the impact cratering process on small bodies. Complicating simple interpretations of crater and ejecta morphology are dynamical effects on ejecta emplacement resulting from Eros’ irregular shape, rapid (5.27 hr) rotation, and low gravity. Figure 1 shows the very different ejecta deposit morphology that can result if the effects of rotation alone are neglected. Considering the additional complicating factors of Eros’ irregular shape and complex gravitational field, simple calculations of the extent and thickness of ejecta blankets and the spatial distribution of ejecta blocks from basic crater scaling laws or numerical hydrocodes alone do not suffice. In order to fully interpret the suite of NEAR-Shoemaker observations of regolith features across the surface of Eros and to evaluate various impact models for specific craters on the asteroid, detailed dynamical modeling of the deposition of crater ejecta from those craters is required . Here, I describe some modifications and improvements to the dynamical model being used for these studies.

D D Durda↗

DOE OSTI2019

Advanced Laboratory and Field Arrays: Accelerate the development of next-generation marine energy conversion systems (Task 3)

This task successfully improved the representation of extreme conditions in wave resource characterization. The previous state of the art to characterize extreme conditions was a purely statistical representation of buoy data; Generalized Extreme Value (GEV) techniques are used to extrapolate joint histograms of significant wave height Hs and energy period T e for a given return period. The task confirmed that H s and T e alone are a poor representation of the actual hazards in extreme events, during which WECs many experience significant impacts from wave groups and breaking waves (i.e., it is not simply the largest wave that may produce the worst condition for a WEC). This task improved both prediction of extreme conditions with “third generation” spectral models, such as WAVEWATCHIII and WAM, and observational quantification of the wave breaking impacts during extreme conditions.

13 HYDRO ENERGY↗

DOE OSTI2024

Accelerating Multiphase Simulations With Denoising Diffusion Model Driven Initializations

This study introduces a hybrid fluid simulation approach that integrates generative diffusion models with physics‐based simulations, aiming at reducing the computational costs of flow simulations while still honoring all the physical properties of interest. Pore‐scale simulations enhance our understanding of applications such as assessing hydrogen and storage efficiency in underground reservoirs. Nevertheless, they are computationally expensive and the presence of non‐unique solutions can require multiple simulations within a single geometry. To overcome the computational cost hurdle, we propose a method that couples generative diffusion models and physics‐based simulations. While training the data‐driven model, we simultaneously generate initial conditions and perform physics‐based simulations using these. This integrated approach enables us to receive real‐time feedback on a single compute node equipped with both CPUs and GPUs. By efficiently managing these processes within a single compute node, we can continuously monitor performance and halt training once the model meets the specified criteria. To test our model, we generate realizations in a real Berea sandstone fracture which shows that our technique is up to 4.4 times faster than commonly used flow simulation initializations.

36 MATERIALS SCIENCE↗

DOE OSTI2017

A probabilistic drought forecasting framework: A combined dynamical and statistical approach

In order to improve drought forecasting skill, this study develops a probabilistic drought forecasting framework comprised of dynamical and statistical modeling components. The novelty of this study is to seek the use of data assimilation to quantify initial condition uncertainty with the Monte Carlo ensemble members, rather than relying entirely on the hydrologic model or land surface model to generate a single deterministic initial condition, as currently implemented in the operational drought forecasting systems. Next, the initial condition uncertainty is quantified through data assimilation and coupled with a newly developed probabilistic drought forecasting model using a copula function. The initial condition at each forecast start date are sampled from the data assimilation ensembles for forecast initialization. Finally, seasonal drought forecasting products are generated with the updated initial conditions. This study introduces the theory behind the proposed drought forecasting system, with an application in Columbia River Basin, Pacific Northwest, United States. Results from both synthetic and real case studies suggest that the proposed drought forecasting system significantly improves the seasonal drought forecasting skills and can facilitate the state drought preparation and declaration, at least three months before the official state drought declaration.

data assimilation↗

DOE OSTI2016

Higher harmonic flow coefficients of identified hadrons in Pb-Pb collisions at $\sqrt{s_{\mathrm{NN}}}=2.76 $ TeV

The elliptic, triangular, quadrangular and pentagonal anisotropic flow coefficients for π ± , K ± and p + $\bar{p}$ in Pb-Pb collisions at √s NN =2.76 TeV were measured with the ALICE detector at the Large Hadron Collider. The results were obtained with the Scalar Product method, correlating the identified hadrons with reference particles from a different pseudorapidity region. Effects not related to the common event symmetry planes (non-flow) were estimated using correlations in pp collisions and were subtracted from the measurement. The obtained flow coefficients exhibit a clear mass ordering for transverse momentum (p T ) values below ≈ 3 GeV/c. In the intermediate p T region (3 < p T < 6 GeV/c), particles group at an approximate level according to the number of constituent quarks, suggesting that coalescence might be the relevant particle production mechanism in this region. The results for p T < 3 GeV/c are described fairly well by a hydrodynamical model (iEBE-VISHNU) that uses initial conditions generated by A Multi-Phase Transport model (AMPT) and describes the expansion of the fireball using a value of 0.08 for the ratio of shear viscosity to entropy density (η/s), coupled to a hadronic cascade model (UrQMD). Finally, expectations from AMPT alone fail to quantitatively describe the measurements for all harmonics throughout the measured transverse momentum region. However, the comparison to the AMPT model highlights the importance of the late hadronic rescattering stage to the development of the observed mass ordering at low values of p T and of coalescence as a particle production mechanism for the particle type grouping at intermediate values of p T for all harmonics.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗