Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “data inference”

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 73 records · Page 4
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Architectural Implications of Neural Network Inference for High Data-Rate, Low-Latency Scientific Applications

With more scientific fields relying on neural networks (NNs) to process data incoming at extreme throughputs and latencies, it is crucial to develop NNs with all their parameters stored on-chip. In many of these applications, there is not enough time to go off-chip and retrieve weights. Even more so, off-chip memory such as DRAM does not have the bandwidth required to process these NNs as fast as the data is being produced (e.g., every 25 ns). As such, these extreme latency and bandwidth requirements have architectural implications for the hardware intended to run these NNs: 1) all NN parameters must fit on-chip, and 2) codesigning custom/reconfigurable logic is often required to meet these latency and bandwidth constraints. In our work, we show that many scientific NN applications must run fully on chip, in the extreme case requiring a custom chip to meet such stringent constraints.

Weng, Olivia↗
DOE OSTI2022

Monitoring Operational States of a Nuclear Reactor Using Seismoacoustic Signatures and Machine Learning

Monitoring nuclear reactors is an important safety and security task with growing requirements. We explore the possibility of using seismic and acoustic data for inferring the power level of an operating reactor. Continuous data recorded at a single seismoacoustic station that is located about 50 m away from a research reactor was visualized and analyzed. The data show a clear correlation between seismoacoustic features and reactor main operational states. We designed a workflow that includes two machine learning (ML) models to classify the reactor operational states (OFF, transition, and ON) and estimate reactor power levels (10%, 30%, 50%, 70%, and 90%). We applied and compared five ML algorithms for the reactor OFF-transition-ON and four approaches for the power level classification. We also compared the performance of ML models trained with seismic-only, acoustic-only, and both types of data. Five-fold cross validations were implemented to assure a thorough evaluation of the model performances. Additionally, the results show the extreme boosting gradient algorithm worked best for the first model, whereas random forests performed best for the second model. Combining seismic and acoustic data leads to better performance than using a single type of data. Seismic data contributed more than acoustic data for both models. We reached an accuracy of 0.98 for reactor OFF and ON. The accuracies for the transition state and power levels are less optimal with a minimum accuracy of 0.66. However, our results suggest seismic and acoustic data contain useful information about the transition state as well as power levels. Seismic and acoustic data could be integrated with other observations to improve monitoring performance.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2022

Privacy policy robustness to reverse engineering

Differential privacy policies allow one to preserve data privacy while sharing and analyzing data. However, these policies are susceptible to an array of attacks. In particular, often a portion of the data desired to be privacy protected is exposed online. Access to these pre-privacy protected data samples can then be used to reverse engineer the privacy policy. With knowledge of the generating privacy policy, an attacker can use machine learning to approximate the full set of originating data. Bayesian inference is one method for reverse engineering both model and model parameters. We present a methodology for evaluating and ranking privacy policy robustness to Bayesian inference-based reverse engineering, and demonstrated this method across data with a variety of temporal trends.

Kusne, Aaron Gilad↗
DOE OSTI2025

Physics-Informed Gaussian Process Inference of Liquid Structure from Scattering Data

We present a nonparametric Bayesian framework to infer radial distribution functions from experimental scattering measurements with uncertainty quantification using nonstationary Gaussian processes. The Gaussian process prior mean and kernel functions are designed to mitigate well-known numerical challenges with the Fourier transform, including discrete measurement binning and detector windowing, while encoding fundamental yet minimal physical knowledge of the liquid structure. We demonstrate uncertainty propagation of the Gaussian process posterior to unmeasured quantities of interest. Experimental radial distribution functions of liquid argon and water with uncertainty quantification are provided as both a proof of principle for the method and a benchmark for molecular models.

Chemical structure↗
DOE OSTI2025

Better practices for inferring ecosystem water use strategy from eddy covariance data

Eddy covariance data are critical for inferring ecosystem water use strategies. Yet, such inferences are sensitive to a range of assumptions applied across studies, hindering our understanding of water use strategies within and across eddy covariance sites. A recent analysis across 151 FLUXNET2015 and AmeriFlux-FLUXNET datasets found that poor model performance was the key driver of non-robust inferences of ecosystem water use strategies. Here, we leverage this previous analysis to (i) identify the specific assumptions that improve inference model performance across most sites, (ii) explain the mechanisms behind the performance improvements, and (iii) check whether better performance improves water use inference. We find that the common practice of fitting a model to canopy conductance (G c ) derived from the evapotranspiration (ET) observations, rather than to observed ET itself, artificially amplifies data errors and degrades the model performance. Next, accounting for vegetation dynamics by applying a growing season filter or incorporating satellite LAI data improves performance, but the former practice may remove soil water stress periods. Lastly, using the leaf-to-air vapor pressure deficit (VPD l ) derived from ET observations as a model input may artificially inflate performance. Based on these results, we recommend selecting observed ET (rather than derived G c ) as the response variable, carefully accounting for vegetation dynamics, and avoiding derived VPD l as a model input; these best practices improve model performance by c. 20% and robustness by c. 80% across all eddy covariance sites. Nevertheless, the performance improvements do not always correspond to more robust inference of water use strategies, as model parameter selection and surface energy budget closure corrections still strongly influence the ecosystem water use parameter estimation in a site-specific manner.

AmeriFlux↗
DOE OSTI2026

Ripening of Rh Nanoparticle Catalysts in Reverse Water–Gas Shift via a Data-Driven Model Combining Physics, Theory, and Experiment

Degradation via sintering is an ongoing challenge that impedes the broad commercial success of supported metallic nanoparticle catalysts. To mitigate degradation via informed catalyst design and process operations, here we aim to disambiguate the underlying mechanisms of sintering by combining theory and experiment in a quantitative framework. While mechanistic sintering models exist, they only model a single sintering pathway, even though multiple sintering mechanisms can occur simultaneously or dominate at different stages of the process. Data-driven machine learning models have emerged as a means to represent complex processes through data regression. However, machine learning models have very large data needs and lack mechanistic insights due to their black-box encoding. To develop an interpretive model of catalyst degradation via sintering, we constructed a hybrid model combining mechanistic “physics-based” models and data-driven methods to obtain both reliable predictions and mechanistic insights regarding experimentally observed sintering phenomena. Focusing on nanoparticle sintering in the Rh–TiO 2 catalyst for the reverse water–gas shift (RWGS) reaction, the hybrid model couples a mechanistic term for Ostwald ripening with energy values calculated via density functional theory (DFT) with a parametric, data-driven discrepancy function term for unmodeled mechanisms. The hybrid model is trained using Bayesian inference with data collected from small-angle X-ray scattering (SAXS) in situ experiments wherein average nanoparticle diameter versus time was measured at three relevant operating temperatures. The calibrated hybrid model results show that an Ostwald ripening-only model parameterized with fixed DFT energies does not fully capture the time and temperature dependence of the SAXS-observed sintering kinetics, and that an additional functional contribution, or DFT energy calibration, is required to reconcile simulation and experiment. Analysis of the hybrid-model error confirms that the hybrid model outperforms both the purely mechanistic and purely data-driven alternatives in terms of expected predictive accuracy for time-evolving average particle sizes. Furthermore, the results support the hypothesis that the Ostwald ripening mechanism is less important for explaining the sintering phenomena as operating temperature increases under an assumed fixed DFT parameterization. This could be explained in one of two ways: either latent, unmodeled sintering mechanisms dominate at higher temperatures, or the DFT uncertainty increases with temperature. The proposed modeling approach directly links theory to experiments and simulations via a statistical hybrid modeling framework and can be extended to other catalytic systems to improve predictive models and mechanistic understanding.

Bayesian hybrid modeling↗
DOE OSTI2022

Super-Resolution Approaches in Three-Dimensions for Classification and Screening of Commercial-Off-The-Shelf Components

X-ray computed tomography is generally a primary step in characterization of defective electronic components, but is generally too slow to screen large lots of components. Super-resolution imaging approaches, in which higher-resolution data is inferred from lower-resolution images, have the potential to substantially reduce collection times for data volumes accessible via x-ray computed tomography. Here we seek to advance existing two-dimensional super-resolution approaches directly to three-dimensional computed tomography data. Multiple scan resolutions over a half order of magnitude of resolution were collected for four classes of commercial electronic components to serve as training data for a deep-learning, super-resolution network. A modular python framework for three-dimensional super-resolution of computed tomography data has been developed and trained over multiple classes of electronic components. Initial training and testing demonstrate the vast promise for these approaches, which have the potential for more than an order of magnitude reduction in collection time for electronic component screening.

47 OTHER INSTRUMENTATION↗
DOE OSTI2020

Characterization of the pressure coefficient of manganin and temperature evolution of pressure in piston-cylinder cells

Here, we report measurements of the temperature- and pressure-dependent resistance, R ( T , p ), of a manganin manometer in a 4 He-gas pressure setup from room temperature down to the solidification temperature of 4 He ( T solid ~ 50 K at 0.8 GPa) for pressures, p , between 0 GPa and ~0.8 GPa. The same manganin wire manometer was also measured in a piston-cylinder cell (PCC) from 300 K down to 1.8 K and for pressures between 0 GPa and ~2 GPa. From these data, we infer the temperature and pressure dependence of the pressure coefficient of manganin, α ( T , p ), defined by the equation R p = (1 + αp ) R 0 , where R 0 and R p are the resistances of manganin at ambient pressure and finite pressure, respectively. Our results indicate that upon cooling, α first decreases, then goes through a broad minimum at ~120 K, and increases again toward lower temperatures. In addition, we find that α is almost pressure-independent at T ≳ 60 K up to p ~ 2 GPa, but shows a pronounced p dependence at T ≲ 60 K. Using this manganin manometer, we demonstrate that p overall decreases with decreasing temperature in the PCC for the full pressure range and that the size of the pressure difference between room temperature and low temperatures ( T = 1.8 K), Δ p , decreases with increasing pressure. We also compare the pressure values inferred from the manganin manometer with the low-temperature pressure, determined from the superconducting transition temperature of elemental lead (Pb). As a result of these data and analysis, we propose a practical algorithm to infer the evolution of pressure with temperature in a PCC.

47 OTHER INSTRUMENTATION↗
DOE OSTI2022

Strong Lensing Parameter Estimation on Ground-Based Imaging Data Using Simulation-Based Inference

Current ground-based cosmological surveys, such as the Dark Energy Survey (DES), are predicted to discover thousands of galaxy-scale strong lenses, while future surveys, such as the Vera Rubin Observatory Legacy Survey of Space and Time (LSST) will increase that number by 1-2 orders of magnitude. The large number of strong lenses discoverable in future surveys will make strong lensing a highly competitive and complementary cosmic probe. To leverage the increased statistical power of the lenses that will be discovered through upcoming surveys, automated lens analysis techniques are necessary. We present two Simulation-Based Inference (SBI) approaches for lens parameter estimation of galaxy-galaxy lenses. We demonstrate the successful application of Neural Posterior Estimation (NPE) to automate the inference of a 12-parameter lens mass model for DES-like ground-based imaging data. We compare our NPE constraints to a Bayesian Neural Network (BNN) and find that it outperforms the BNN, producing posterior distributions that are for the most part both more accurate and more precise; in particular, several source-light model parameters are systematically biased in the BNN implementation.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2021

A Privacy-Preserving Distributed Control of Optimal Power Flow

Here, we consider a distributed optimal power flow formulated as an optimization problem that maximizes a nondifferentiable concave function. Solving such a problem by the existing distributed algorithms can lead to data privacy issues because the solution information exchanged within the algorithms can be utilized by an adversary to infer the data. To preserve data privacy, in this paper we propose a differentially private projected subgradient (DP-PS) algorithm that includes a solution encryption step. We show that a sequence generated by DP-PS converges in expectation, in probability, and with probability 1. Moreover, we show that the rate of convergence in expectation is affected by a target privacy level of DP-PS chosen by the user. We conduct numerical experiments that demonstrate the convergence and data privacy preservation of DP-PS.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2026

AIF for Vis (Active Inference for simulating human interpretation of data visualization) [SWR-26-084]

AIF for Vis contains the Active Inference models and analysis scripts used to study a simple visualization-interpretation task: estimating the average value of two bars in a bar chart. The work is a proof of concept for translating hypothesized cognitive strategies into executable, inspectable process models. We implement two idealized strategies inspired by dual-process accounts of visualization-aided decision making: *Fast model: a compressed, heuristic strategy that estimates the visual midpoint of the two bars and maintains a single belief over their average. *Slow model: a sequential, analytic strategy that estimates the two bar heights separately and maintains them in working memory before computing an average. Both models use a common Active-Inference-inspired framework for sequential perception, belief updating, action selection, and reporting. Their different internal representations produce distinct predicted vulnerabilities: *the Fast model is more susceptible to tick-salience bias; *the Slow model is more susceptible to working-memory decay. The repository includes the model implementations, scripts used for the experiments reported in the paper, precomputed trial-level results, and plotting scripts.

Goldwyn, Harrison [National Laboratory of the Rock↗
DOE OSTI2021

Scalable statistical inference of photometric redshift via data subsampling

Handling big data has largely been a major bottleneck in traditional statistical models. Consequently, when accurate point prediction is the primary target, machine learning models are often preferred over their statistical counterparts for bigger problems. But full probabilistic statistical models often outperform other models in quantifying uncertainties associated with model predictions. We develop a data-driven statistical modeling framework that combines the uncertainties from an ensemble of statistical models learned on smaller subsets of data carefully chosen to account for imbalances in the input space. We demonstrate this method on a photometric redshift estimation problem in cosmology, which seeks to infer a distribution of the redshift—the stretching effect in observing the light of far-away galaxies—given multivariate color information observed for an object in the sky. Our proposed method performs balanced partitioning, graph-based data subsampling across the partitions, and training of an ensemble of Gaussian process models.

data subsampling↗
DOE OSTI2020

Towards the spatial resolution of metalloprotein charge states by detailed modeling of XFEL crystallographic diffraction

Oxidation states of individual metal atoms within a metalloprotein can be assigned by examining X-ray absorption edges, which shift to higher energy for progressively more positive valence numbers. Indeed, X-ray crystallography is well suited for such a measurement, owing to its ability to spatially resolve the scattering contributions of individual metal atoms that have distinct electronic environments contributing to protein function. However, as the magnitude of the shift is quite small, about +2 eV per valence state for iron, it has only been possible to measure the effect when performed with monochromated X-ray sources at synchrotron facilities with energy resolutions in the range 2–3 × 10 −4 (Δ E / E ). This paper tests whether X-ray free-electron laser (XFEL) pulses, which have a broader bandpass (Δ E / E = 3 × 10 −3 ) when used without a monochromator, might also be useful for such studies. The program nanoBragg is used to simulate serial femtosecond crystallography (SFX) diffraction images with sufficient granularity to model the XFEL spectrum, the crystal mosaicity and the wavelength-dependent anomalous scattering factors contributed by two differently charged iron centers in the 110-amino-acid protein, ferredoxin. Bayesian methods are then used to deduce, from the simulated data, the most likely X-ray absorption curves for each metal atom in the protein, which agree well with the curves chosen for the simulation. The data analysis relies critically on the ability to measure the incident spectrum for each pulse, and also on the nanoBragg simulator to predict the size, shape and intensity profile of Bragg spots based on an underlying physical model that includes the absorption curves, which are then modified to produce the best agreement with the simulated data. This inference methodology potentially enables the use of SFX diffraction for the study of metalloenzyme mechanisms and, in general, offers a more detailed approach to Bragg spot data reduction.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2022

A Review of Bayesian Networks for Spatial Data

We report Bayesian networks are a popular class of multivariate probabilistic models as they allow for the translation of prior beliefs about conditional dependencies between variables to be easily encoded into their model structure. Due to their widespread usage, they are often applied to spatial data for inferring properties of the systems under study and also generating predictions for how these systems may behave in the future. We review published research on methodologies for representing spatial data with Bayesian networks and also summarize the application areas for which Bayesian networks are employed in the modeling of spatial data. We find that a wide variety of perspectives are taken, including a GIS-centric focus on efficiently generating geospatial predictions, a statistical focus on rigorously constructing graphical models controlling for spatial correlation, as well as a range of problem-specific heuristics for mitigating the effects of spatial correlation and dependency arising in spatial data analysis. Special attention is also paid to potential future directions for integration of Bayesian networks with spatial processes.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Variational autoencoders for at-source data reduction and anomaly detection in high energy particle detectors

Detectors in next-generation high-energy physics experiments face several daunting requirements, such as high data rates, damaging radiation exposure, and stringent constraints on power, space, and latency. To address these challenges, machine learning in readout electronics can be leveraged for smart detector designs, enabling intelligent inference and data reduction at-source. Variational autoencoders (VAEs) offer a variety of benefits for front-end readout; an on-sensor encoder can perform efficient lossy data compression while simultaneously providing a latent space representation that can be used for anomaly detection. Results are presented from low-latency and resource-efficient VAEs for front-end data processing in a futuristic silicon pixel detector. Encoder-based data compression is found to preserve good performance of off-detector analysis while significantly reducing the off-detector data rate as compared to a similarly sized data filtering approach. Furthermore, the latent space information is found to be a useful discriminator in the context of real-time sensor defect monitoring. Together, these results highlight the multifaceted utility of autoencoder-based front-end readout schemes and motivate their consideration in future detector designs.

47 OTHER INSTRUMENTATION↗
DOE OSTI2023

A Bayesian model for multivariate discrete data using spatial and expert information with application to inferring building attributes

When modeling sparsely observed multivariate data, strong prior information elicited from experts can be used to bolster predictive accuracy and counteract sampling bias. Similarly, modeling autocorrelation in space can help make use of co-occurrence patterns present in many types of spatial data. To make use of both expert prior information and spatial structure, we propose a novel graphical model for a spatial Bayesian network developed specifically to address challenges in inferring the attributes of buildings from geographically sparse observational data. This model is implemented as the sum of a spatial multivariate Gaussian random field and a tabular conditional probability function in real-valued space prior to projection onto the probability simplex. This modeling form is especially suitable for the usage of prior information in the form of sets of atomic rules obtained from experts. To perform inference with missing data, we implement a Markov chain Monte Carlo scheme composed of alternating steps of Gibbs sampling of missing entries and Hamiltonian Monte Carlo for model parameters. A case study in building attribution is presented to highlight the advantages and limitations of this approach.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Deep learning-accelerated 3D carbon storage reservoir pressure forecasting based on data assimilation using surface displacement from InSAR

Fast forecasting of the reservoir pressure distribution during geologic carbon storage (GCS) by assimilating monitoring data is a challenging problem. Due to high drilling cost, GCS projects usually have spatially sparse measurements from few wells, leading to high uncertainties in reservoir pressure prediction. To address this challenge, we use low-cost Interferometric Synthetic-Aperture Radar (InSAR) data as monitoring data to infer reservoir pressure build up. We develop a deep learning-accelerated workflow to assimilate surface displacement maps interpreted from InSAR and to forecast dynamic reservoir pressure. Employing an Ensemble Smoother Multiple Data Assimilation (ES-MDA) framework, the workflow updates three-dimensional (3D) geologic properties and predicts reservoir pressure with quantified uncertainties. We use a synthetic commercial-scale GCS model with bimodally distributed permeability and porosity to demonstrate the efficacy of the workflow. A two-step CNN-PCA approach is employed to parameterize the bimodal fields. The computational efficiency of the workflow is boosted by two residual U-Net based surrogate models for surface displacement and reservoir pressure predictions, respectively. The workflow can complete data assimilation and reservoir pressure forecasting in half an hour on a personal computer.

25 ENERGY STORAGE↗
DOE OSTI2022

Synthesis of observed and simulated rain microphysics to inform a new Bayesian statistical framework for microphysical parameterization in climate models (Final Scientific Report)

This project aimed to investigate a new approach to bulk microphysics schemes. We identified the need to move beyond the fixed structural assumptions and approximation of existing schemes. For example, most schemes assume some functional form for the rain drop size distribution (e.g. a gamma or exponential distribution). Most schemes then evolve some number of statistical moments of that distribution via various processes, such as evaporation, sedimentation, collision-coalescence, and collisional breakup. These microphysical processes, in turn, are typically some fixed functional form derived from either some other (more detailed) model, or using some single-particle rates that are then integrated over the size distribution. While these process rate functions may have some free parameters to adjust, in other cases doing so is impossible (e.g. when the process rates are analytical or piecewise solutions to some target function). Our goal was to build and test a scheme that assumed no size distribution form, and used a series of power laws as the basis for the process rates. Moments of the size distribution would be predicted, but no underlying size distribution would be specified. Any number or choice of size distribution moments could be used, and any number of power laws could be employed to model the process rates. Thus, our approach is flexible, and can seamlessly scale across levels of complexity, as demanded by the data. Bayesian inference then would provide the formalism to estimate the model parameters and structure, allowing for robust uncertainty quantification. We proposed testing this framework in idealized simulations using bin microphysical schemes as a data source. We also proposed using real data to inform our microphysics scheme, and also that we would integrate our scheme into WRF.

58 GEOSCIENCES↗