Search NASA⌕ Search

SEARCH · Search NASA

Results for “predictive distribution”

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 37 records · Page 2
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Challenges in Markov Chain Monte Carlo for Bayesian Neural Networks

Markov chain Monte Carlo (MCMC) methods have not been broadly adopted in Bayesian neural networks (BNNs). This paper initially reviews the main challenges in sampling from the parameter posterior of a neural network via MCMC. Such challenges culminate to lack of convergence to the parameter posterior. Nevertheless, this paper shows that a nonconverged Markov chain, generated via MCMC sampling from the parameter space of a neural network, can yield via Bayesian marginalization a valuable posterior predictive distribution of the output of the neural network. Further, classification examples based on multilayer perceptrons showcase highly accurate posterior predictive distributions. The postulate of limited scope for MCMC developments in BNNs is partially valid; an asymptotically exact parameter posterior seems less plausible, yet an accurate posterior predictive distribution is a tenable research avenue.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Models, data, and scripts associated with “Prediction of Distributed River Sediment Respiration Rates using Community-Generated Data and Machine Learning”

This data package is associated with the publication “Prediction of Distributed River Sediment Respiration Rates using Community-Generated Data and Machine Learning’’ submitted to the Journal of Geophysical Research: Machine Learning and Computation (Scheibe et al. 2024). River sediment respiration observations are expensive and labor intensive to obtain and there is no physical model for predicting this quantity. The Worldwide Hydrobiogeochemisty Observation Network for Dynamic River Systems (WHONDRS) observational data set (Goldman et al.; 2020) is used to train machine learning (ML) models to predict respiration rates at unsampled sites. This repository archives training data, ML models, predictions, and model evaluation results for the purposes of reproducibility of the results in the associated manuscript and community reuse of the ML models trained in this project. One of the key challenges in this work was to find an optimum configuration for machine learning models to work with this feature-rich (i.e. 100+ possible input variables) data set. Here, we used a two-tiered approach to managing the analysis of this complex data set: 1) a stacked ensemble of ML models that can automatically optimize hyperparameters to accelerate the process of model selection and tuning and 2) feature permutation importance to iteratively select the most important features (i.e. inputs) to the ML models. The major elements of this ML workflow are modular, portable, open, and cloud-based, thus making this implementation a potential template for other applications. This data package is associated with the GitHub repository found at Please see the file level metadata (flmd; “sl-archive-whondrs_flmd.csv”) for a list of all files contained in this data package and descriptions for each. Please see the data dictionary (dd; “sl-archive-whondrs_dd.csv”) for a list of all column headers contained within comma separated value (csv) files in this data package and descriptions for each. The GitHub repository is organized into five top-level directories: (1) “input_data” holds the training data for the ML models; (2) “ml_models” holds machine learning models trained on the data in “input_data”; (3) “scripts” contains data preprocessing and postprocessing scripts and intermediate results specific to this data set that bookend the ML workflow; (4) “examples” contains the visualization of the results in this repository including plotting scripts for the manuscript (e.g., model evaluation, FPI results) and scripts for running predictions with the ML models (i.e., reusing the trained ML models); (5) “output_data” holds the overall results of the ML model on that branch. Each trained ML model resides on its own branch in the repository; this means that inputs and outputs can be different branch-to-branch. Furthermore, depending on the number of features used to train the ML models, the preprocessing and postprocessing scripts, and their intermediate results, can also be different branch-to-branch. The “main-*” branches are meant to be starting points (i.e. trunks) for each model branch (i.e. sprouts). Please see the Branch Navigation section in the top-level README.md in the GitHub repository for more details. There is also one hidden directory “.github/workflows”. This hidden directory contains information for how to run the ML workflow as an end-to-end automated GitHub Action but it is not needed for reusing the ML models archived here. Please the top-level README.md in the GitHub repository for more details on the automation.

13C↗
DOE OSTI2022

Distributed model predictive control for coordinated, grid-interactive buildings

Continued focus on reducing carbon emissions and improving energy efficiency requires buildings to become grid-interactive and not just behave as static consumers. A distributed model predictive control (DMPC) algorithm known as Limited-Communication (LC) DMPC is modified to enable grid-interactive buildings. A grid-aggregator subsystem is added that allows for a bulk grid power reference signal to be followed while the individual building subsystems also achieve their local comfort objectives. The LC-DMPC algorithm is applied for the first time to systems with multiple buildings. Adequate power tracking is shown for different simulation scenarios involving heterogeneous buildings, and next steps are discussed.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2018

Active Power Control for Wind Farms Using Distributed Model Predictive Control and Nearest Neighbor Communication: Preprint

Wind plant control strategies, including axial induction and wake steering control, aim to improve the performance of wind farms, including increasing energy production and decreasing turbine loads. This paper presents a linear model of wake characteristics for use with a distributed model predictive control method for the purpose of optimizing axial induction and yaw misalignment setpoints. In particular, we use an iterative, distributed control method with nearest neighbor communication to coordinate turbine control actions that account for wake interactions between turbines. Simulations of the model and controller are performed on a 2x3 array of turbines using a modified version of the FLOw Redirection and Induction in Steady-state (FLORIS) model to dynamically track the relevant wake parameters. Preliminary results show the controller's ability to follow an arbitrary wind farm power reference signal for the purpose of providing active power control (APC) ancillary services for power grid stability. This efficient distributed control strategy can enable real-time wind farm optimization and control, even for very large scale farms.

17 WIND ENERGY↗
DOE OSTI2021

Streamflow simulation in data-scarce basins using Bayesian and physics-informed machine learning models

Hydrologic predictions at rural watersheds are important but also challenging due to data shortage. Long short-term memory (LSTM) networks are a promising machine learning approach and have demonstrated good performance in streamflow predictions. However, due to its data-hungry nature, most LSTM applications focus on well-monitored catchments with abundant and high-quality observations. In this work, we investigate predictive capabilities of LSTM in poorly monitored watersheds with short observation records. To address three main challenges of LSTM applications in data-scarce locations, i.e., overfitting, uncertainty quantification (UQ), and out-of-distribution prediction, we evaluate different regularization techniques to prevent overfitting, apply a Bayesian LSTM for UQ, and introduce a physics-informed hybrid LSTM to enhance out-of-distribution prediction. Through case studies in two diverse sets of catchments with and without snow influence, we demonstrate that 1) when hydrologic variability in the prediction period is similar to the calibration period, LSTM models can reasonably predict daily streamflow with Nash–Sutcliffe efficiency above 0.8, even with only 2 years of calibration data; 2) when the hydrologic variability in the prediction and calibration periods is dramatically different, LSTM alone does not predict well, but the hybrid model can improve the out-of-distribution prediction with acceptable generalization accuracy; 3) L2 norm penalty and dropout can mitigate overfitting, and Bayesian and hybrid LSTM have no overfitting; and 4) Bayesian LSTM provides useful uncertainty information to improve prediction understanding and credibility. In conclusion, these insights have vital implications for streamflow simulation in watersheds where data quality and availability are a critical issue.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Machine Learning-Based Prediction of Distribution Network Voltage and Sensors Allocation: Preprint

Increasing penetration of fast-varying energy resources may negatively affect power systems' operation. At the same time, sensor deployment throughout distribution networks improves system awareness and enables the development of new and advanced voltage control solutions. Such control techniques rely on accurate prediction in anticipation to voltage violation scenarios. This paper analyzes various approaches for voltage prediction in a distribution system; it is shown that combining multiple techniques into a single regressor improves its predictive power. Moreover, a two-step regressor is proposed, where initial predictions based on a global regressor are refined by local regressors; in this case, prediction errors decrease significantly. Additionally, a clustering approach is employed for performing sensor allocation, so that only the most influential buses are selected for monitoring without diminishing prediction accuracy.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2020

Machine Learning-Based Prediction of Distribution Network Voltage and Sensor Allocation

Increasing penetration of fast-varying energy resources may negatively affect power systems' operation. At the same time, sensor deployment throughout distribution networks improves system awareness and enables the development of new and advanced voltage control solutions. Such control techniques rely on accurate prediction in anticipation to voltage violation scenarios. This paper analyzes various approaches for voltage prediction in a distribution system; it is shown that combining multiple techniques into a single regressor improves its predictive power. Moreover, a two-step regressor is proposed, where initial predictions based on a global regressor are refined by local regressors; in this case, prediction errors decrease significantly. Additionally, a clustering approach is employed for performing sensor allocation, so that only the most influential buses are selected for monitoring without diminishing prediction accuracy.

61 RADIATION PROTECTION AND DOSIMETRY↗
DOE OSTI2020

Machine Learning-Based Prediction of Distribution Network Voltage and Sensor Allocation

Increasing penetration levels of fast-varying energy resources might negatively affect power system operation. At the same time, sensor deployment throughout distribution networks improves system awareness and enables the development of new and advanced voltage control solutions. Such control techniques rely on accurate prediction in anticipation of voltage violation scenarios. This paper analyzes various approaches to voltage prediction in a distribution system, and it is shown that combining multiple techniques into a single regressor improves its predictive power. Moreover, a two-step regressor is proposed in which initial predictions based on a global regressor are refined by local regressors; in this case, prediction errors decrease significantly. Additionally, a clustering approach is employed to perform sensor allocation so that only the most influential buses are selected for monitoring without diminishing prediction accuracy.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2015

SU-E-T-798: Verification of 3DVH Dose Distribution Before Clinical Implementation for Patient-Specific IMRT QA

Purpose: In recent years patient-specific IMRT QA has transitioned from film and chamber measurements to beam-by-beam 2D array measurements. 3DVH takes this transition a step further by estimating the 3D dose distribution delivered using 2D per beam diode array measurements. In this study, the 3D dose distribution generated by 3DVH is compared to film and chamber measurements. In addition, the accuracy ROI volume and error detection is investigated. Methods: Composite film and ion chamber measurements in a solid water phantom were performed for 9 IMRT PINNACLE patient plans for 4 treatment sites. The film and chamber measurements were compared to the dose distribution predicted by 3DVH using MAPCHECK2 per beam measurements. The absolute point dose measurement (CAX) was extracted from the predicted 3DVH and PINNACLE dose distribution and was compared by taking the ratio of measured to predicted doses. The dose distribution measured with film was compared to the distribution in the corresponding plane (AX, SAG, COR) extracted from predicted dose distribution by 3DVH and PINNACLE using a 2D gamma analysis. Gamma analysis was performed with 2% dose, 2 mm DTA, 20% threshold, and global normalization. In addition, the percent difference between 3DVH and PINNACLE ROI volumes was calculated. Results: The average ratio of the measured point dose vs the 3DVH predicted dose was 1.017 (σ=0.011). The average gamma passing rate for measured vs 3DVH dose distributions was 95.1% (σ=2.53%). The average percent difference of 3DVH vs PINNACLE ROI volume was 2.29% (σ=2.5%). Conclusion: The dose distributions predicted by 3DVH using MAPCHECK2 measurements are the same as the distributions that would have been obtained using film and chamber. The ROI volumes used in 3DVH are not an exact match to those in PINNACLE; the effect requires more investigation. The accuracy of error detection by 3DVH is currently being investigated.

60 APPLIED LIFE SCIENCES↗
NASA NTRS1988

Measured and predicted pressure distributions on the AFTI/F-111 mission adaptive wing

Flight tests have been conducted using an F-111 aircraft modified with a mission adaptive wing (MAW). The MAW has variable-camber leading and trailing edge surfaces that can change the wing camber in flight, while preserving smooth upper surface contours. This paper contains wing surface pressure measurements obtained during flight tests at Dryden Flight Research Facility of NASA Ames Research Center. Upper and lower surface steady pressure distributions were measured along four streamwise rows of static pressure orifices on the right wing for a leading-edge sweep angle of 26 deg. The airplane, wing, instrumentation, and test conditions are discussed. Steady pressure results are presented for selected wing camber deflections flown at subsonic Mach numbers up to 0.90 and an angle-of-attack range of 5 to 12 deg. The Reynolds number was 26 million, based on the mean aerodynamic chord. The MAW flight data are compared to MAW wind tunnel data, transonic aircraft technology (TACT) flight data, and predicted pressure distributions. The results provide a unique database for a smooth, variable-camber, advanced supercritical wing.

Webb, Lannie D.↗
DOE OSTI2016

MO-G-201-04: Knowledge-Based Planning for Single-Isocenter Stereotactic Radiosurgery to Multiple Brain Metastases

Purpose: Single-isocenter, linac-based SRS for multiple brain metastases (multi-mets) can deliver highly conformal radiation doses and reduce overall patient treatment time compared to other therapy techniques. This study aims to quantify the dosimetric benefits of knowledge-based planning (KBP) for multi-met treatments. Methods: Using a previously-published KBP methodology (an artificial neural network (ANN) trained on single-target linac-based SRS plans), 3D dose distribution predictions for multi-met patients were obtained by treating each brain lesion as a solitary target and subsequently combining individual predictions into a single distribution using a dose-weighted geometric averaging to obtain the best results in the inter-target space. 17 previously-treated multi-met plans, with target numbers ranging from N=2–5, were used to validate the ANN predictions and subsequent KBP auto-planning routine. The fully-deliverable KBP plans were developed by converting dose distribution predictions into patient-specific optimization objectives while maintaining identical target normalizations (typically PTV V100%=D98%). Plan quality improvements were quantified by the difference between SRS quality metrics (QMs): δdQM=QM(clinical)-QM(KBP). QMs of interest were: gradient measure (GM), conformity index (CI), brain V10 and V5, brainstem D0.1cc and heterogeneity index (HI). Finally, overall plan quality was judged via blinded plan comparison by SRS-specializing physicians. Results: Two clinical plans were found to be significant outliers wherein plan quality was dramatically worse than KBP. Despite indicating KBP superiority, these were removed from the QM analysis to prevent skewing the results. In the remaining cases, clinical and KBP QMs were nearly identical with modest improvements in the KBP sample: δGM=0.12±0.56mm, δCI=−0.01±0.04, Brain δV10=0.8±2.6cc, brain δV5=6.3 ±10.7cc, brainstem δD0.1cc=0.06±1.19Gy and δHI= −0.04±0.05. Ultimately, 13/17 KBP plans were deemed superior to the manual plans in blinded physician review. Conclusion: The results demonstrate that KBP-driven automated planning in linac-based single-isocenter treatments for multiple brain metastases is indistinguishable from, or even better than, traditional manual planning. J. Hattangadi: Research Grant; Varian Medical Systems; K.L. Moore: Research Grant: Varian Medical Systems.

60 APPLIED LIFE SCIENCES↗
NASA NTRSDate not supplied

A Data-Driven Reduced Order Model of an Isolated Rotor

There are numerous conceptual design stage rotorcraft analysis tasks which demand a high-fidelity and low cost method for rotor load distribution predictions. Considering Urban Air Mobility (UAM) vehicles aim to operate in close proximity to buildings and with unique rotor configurations, there is a significant challenge in quickly and accurately modeling rotors operating in complex, turbulent flow fields. One potential path for deriving a high-fidelity, low cost rotor model is with data-driven surrogate modeling. In this study, an initial investigation is taken to apply a proper orthogonal decomposition (POD) based reduced order model (ROM) for the purpose of pressure distribution prediction. In this study, a POD ROM was derived to produce distributed pressure predictions on rotor blades subjected to topology change due to variation in twist and taper ratio. Rotor twist was varied between 0◦, 10◦, 20◦, and 30◦ while taper ratio was varied between 1.0, 0.9, 0.8, and 0.7. All rotors consisted of a single blade. The POD ROM was validated for three demonstration cases; a high thrust rotor in hover, a low thrust rotor in hover, and a rotor in forward flight with a flight speed of M = 0.1. Results showed highly accurate distributed load predictions could be achieved at minimal computational cost. Computational cost for hovering blade surface pressure modeling was reduced from 12 hours on 440 cores to 10−5 seconds on a single core. For blade in forward flight cost was reduced from 20 hours on 440 cores to 0.6 seconds on a single core. For cases of high thrust and low thrust rotors, POD ROM was used to undergo a design optimization of the rotor such that figure of merit was maximized. Total optimization time for each case was 1 minute.

Data-Driven↗
DOE OSTI2024

Colloidal Stability of PFSA-Ionomer Dispersions. Part I. Single-Ion Electrostatic Interaction Potential Energies

Charged colloidal particles neutralized by a single counterion are increasingly important for many emerging technologies. Attention here is paid specifically to hydrogen fuel cells and water electrolyzers whose catalyst layers are manufactured from a perfluorinated sulfonic acid polymer (PFSA) suspended in aqueous/alcohol solutions. Partially dissolved PFSA aggregates, known collectively as ionomers, are stabilized by the electrostatic repulsion of overlapping diffuse double layers consisting of only protons dissociated from the suspended polymer. We denote such double layers containing no added electrolyte as "single ion". Size-distribution predictions build upon interparticle interaction potential energies from the Derjaguin-Landau-Verwey-Overbeek (DLVO) formalism. However, when only a single counterion is present in solution, classical DLVO electrostatic potential energies no longer apply. Accordingly, here a new formulation is proposed to describe how single-counterion diffuse double layers interact in colloidal suspensions. Part II (Srivastav, H.; Weber, A. Z.; Radke, C. J. Langmuir 2024 DOI: 10.1021/acs.langmuir.3c03904) of this contribution uses the new single-ion interaction energies to predict aggregated size distributions and the resulting solution pH of PFSA in mixtures of n-propanol and water. A single-counterion diffuse layer cannot reach an electrically neutral concentration far from a charged particle. Consequently, nowhere in the dispersion is the solvent neutral, and the diffuse layer emanating from one particle always experiences the presence of other particles (or walls). Thus, in addition to an intervening interparticle repulsive force, a backside osmotic force is always present. With this new construction, we establish that single-ion repulsive pair interaction energies are much larger than those of classical DLVO electrostatic potentials. The proposed single-ion electrostatic pair potential governs dramatic new dispersion behavior, including dispersions that are stable at a low volume fraction but unstable at a high volume fraction and finite volume-fraction dispersions that are unstable with fine particles but stable with coarse particles. Finally, the proposed single-counterion electrostatic pair potential provides a general expression for predicting colloidal behavior for any charged particle dispersion in ionizing solvents with no added electrolyte.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2019

Outer rim simulation: Predicting matter distribution in a universe with billions of galaxies

Anticipating the completion of the Large Synoptic Survey Telescope (LSST) in 2021 in Chile, Argonne scientists have created one of the largest simulations ever produced, the Outer Rim simulation. As it follows the evolution of matter distribution in the universe and the influence of dark energy, the Outer Rim simulation helps researchers make predictions on what they should expect to see when pointing a next generation telescope like the LSST in its direction. Argonne is helping to develop advanced supercomputing and data analysis tools in support of the LSST.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2017

Can Kohn–Sham density functional theory predict accurate charge distributions for both single-reference and multi-reference molecules?

Dipole moments are the first moment of electron density and are fundamental quantities that are often available from experiments. An exchange–correlation functional that leads to an accurate representation of the charge distribution of a molecule should accurately predict the dipole moments of the molecule. It is well known that Kohn–Sham density functional theory (DFT) is more accurate for the energetics of single-reference systems than for the energetics of multi-reference ones, but there has been less study of charge distributions. Here, in this work, we benchmark 48 density functionals chosen with various combinations of ingredients, against accurate experimental data for dipole moments of 78 molecules, in particular 55 single-reference molecules and 23 multi-reference ones. We chose both organic and inorganic molecules, and within the category of inorganic molecules there are both main-group and transition-metal-containing molecules, with some of them being multi-reference. As one would expect, the multi-reference molecules are not as well described by single-reference DFT, and the functionals tested in this work do show larger mean unsigned errors (MUEs) for the 23 multi-reference molecules than the single-reference ones. Five of the 78 molecules have relatively large experimental error bars and were therefore not included in calculating the overall MUEs. For the 73 molecules not excluded, we find that three of the hybrid functionals, B97-1, PBE0, and TPSSh (each with less than or equal to 25% Hartree–Fock (HF) exchange), the range-separated hybrid functional, HSE06 (with HF exchange decreasing from 25% to 0 as interelectronic distance increases), and the hybrid functional, PW6B95 (with 28% HF exchange) are the best performing functionals with each yielding an MUE of 0.18 D. Perhaps the most significant finding of this study is that there exists great similarity among the success rate of various functionals in predicting dipole moments. In particular, of 39 functionals designed as general-purpose functionals and that do not have a global value of 100% HF exchange, the average MUE is 0.23 D, with a standard deviation of only 0.04 D. Among gradient approximations, which are especially interesting because of their speed and portability, the best overall performance is by PBE, HCTH/407, OLYP, OreLYP, and GAM, each with MUE of 0.22 D.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2024

Light Water Reactor Sustainability Program: Use of Time Distributions to Predict Operator Procedure Performance in Dynamic Human Reliability Analysis

The Human Unimodel for Nuclear Technology to Enhance Reliability (HUNTER) framework affords software capable of conducting human reliability analysis (HRA) using a dynamic approach built around operating procedures (OPs) from nuclear power plants (NPPs). Previous HUNTER reports document the development of this software tool, the coupling of HUNTER to the simulator code, the collection of operator performance data by using simulators to calibrate HUNTER models, and linking HUNTER to probabilistic risk assessment (PRA) software. The present report largely addresses two topics. The first is a new function in HUNTER called the HUNTER Procedure Performance Predictor (P3). HUNTER P3 uses HUNTER’s built in Monte Carlo tools featuring human performance variability to identify potential error traps in procedures. The second topic is time distribution analysis to generate time inputs for dynamic HRA. The current analysis was performed to investigate time distributions for task primitives, which are the minimum task unit of analysis used in dynamic HRA modeling. Using the time distribution data, the elapsed time for human actions in an extended loss of AC power (ELAP) scenario is then investigated. Time data and prediction are essential for modeling procedure performance.

99 GENERAL AND MISCELLANEOUS↗
DOE OSTI2023

Validation of EDGDE2D-EIRENE predicted 2D distributions of electron temperature and density against divertor Thomson scattering measurements in the low-field side divertor leg in DIII-D

Comparisons of profiles of the electron temperature (T e ), density (n e ) and pressure (p e ) measured with Divertor Thomson Scattering in DIII-D low-confinement mode discharges to predictions from the edge fluid code EDGE2D-EIRENE show that the models implemented in EDGE2D-EIRENE predict the measurements within their collective uncertainties if the T e at the separatrix (T e,sep ) is 10 eV, or higher. The simulations do not predict, however, the peaked T e and ne profiles measured adjacent to the target plate when T e,sep , is below 10 eV, i.e., for the plasma downstream from the region of ionization of deuterium atoms. Inclusion of cross-field drifts and a five-fold reduction of radial transport cannot reconcile the discrepancy between the measurements and predictions.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗