Search NASA⌕ Search

SEARCH · Search NASA

Results for “operator 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 307 records · Page 17

Karhunen–Loève deep learning method for surrogate modeling and approximate Bayesian parameter estimation

We evaluate the performance of the Karhunen-Loève Deep Neural Network (KL-DNN) framework for surrogate modeling and approximate Bayesian parameter estimation in partial differential equation models. In the surrogate model, the Karhunen-Loève (KL) expansions are used for the dimensionality reduction of the number of unknown parameters and variables, and a deep neural network is employed to relate the reduced space of parameters to that of the state variables. The KL-DNN surrogate model is used to formulate a maximum-a-posteriori-like least-squares problem, which is randomized to draw samples of the posterior distribution of the parameters. We test the proposed framework for a hypothetical unconfined aquifer via comparison with the forward MODFLOW and inverse PEST++ iterative ensemble smoother (IES) solutions as well as the state-of-the-art Fourier neural operator (FNO) and deep operator networks (DeepONets) operator learning surrogate models. Our results show that the KL-DNN surrogate model outperforms FNO and DeepONet for forward predictions. For solving inverse problems, the randomized algorithm provides the same or more accurate Bayesian predictions of the parameters than IES as evidenced by the higher log-predictive probability of both the estimated parameter field and the forecast hydraulic head. The posterior mean obtained from the randomized algorithm is closer to the reference parameter field than that obtained with FNO as the maximum a posteriori estimate.

Approximate Bayesian inference↗

Improvement of two-phase closure models in CTF using Bayesian inference

Under the Consortium for Advanced Simulation of Light Water Reactors (CASL) program, extensive capabilities have been developed in CTF to analyze light-water reactors (LWRs) for normal operating conditions, departure from nucleate boiling (DNB), and system transients. However, further improvements are required in the modeling and simulation of boiling water reactors (BWRs), which is a focus of the Nuclear Energy Advanced Modeling and Simulation (NEAMS) program. In this work, CTF validation results were used to optimize selected modeling coefficients by calibrating to experimental data using a Bayesian inference approach. Here, calibration studies were conducted to improve (vapor) void fraction prediction without worsening the two-phase pressure drop prediction, as well as to improve the two-phase pressure drop prediction. Calibration was performed for interfacial drag and wall shear models. Surrogates were developed to alleviate the computational expense required for sampling the parameter space using Markov chain Monte Carlo (MCMC). An assessment performed with calibrated models demonstrated an improvement of CTF in its prediction of key parameters such as void fraction and two-phase pressure drop.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Developing an Evolutionary Baseline Model for Humans: Jointly Inferring Purifying Selection with Population History

Building evolutionarily appropriate baseline models for natural populations is not only important for answering fundamental questions in population genetics—including quantifying the relative contributions of adaptive versus nonadaptive processes—but also essential for identifying candidate loci experiencing relatively rare and episodic forms of selection (e.g., positive or balancing selection). Here, a baseline model was developed for a human population of West African ancestry, the Yoruba, comprising processes constantly operating on the genome (i.e., purifying and background selection, population size changes, recombination rate heterogeneity, and gene conversion). Specifically, to perform joint inference of selective effects with demography, an approximate Bayesian approach was employed that utilizes the decay of background selection effects around functional elements, taking into account genomic architecture. This approach inferred a recent 6-fold population growth together with a distribution of fitness effects that is skewed towards effectively neutral mutations. Importantly, these results further suggest that, although strong and/or frequent recurrent positive selection is inconsistent with observed data, weak to moderate positive selection is consistent but unidentifiable if rare.

59 BASIC BIOLOGICAL SCIENCES↗

Predictive based monitoring of nuclear plant component degradation using support vector regression

Nuclear power plants (NPPs) are large installations comprised of many active and passive assets. Degradation monitoring of all these assets is expensive (labor cost) and highly demanding task. In this paper a framework based on Support Vector Regression (SVR) for online surveillance of critical parameter degradation of NPP components is proposed. In this case, on time replacement or maintenance of components will prevent potential plant malfunctions, and reduce the overall operational cost. In the current work, we apply SVR equipped with a Gaussian kernel function to monitor components. Monitoring includes the one-step-ahead prediction of the component’s respective operational quantity using the SVR model, while the SVR model is trained using a set of previous recorded degradation histories of similar components. Predictive capability of the model is evaluated upon arrival of a sensor measurement, which is compared to the component failure threshold. A maintenance decision is based on a fuzzy inference system that utilizes three parameters: (i) prediction evaluation in the previous steps, (ii) predicted value of the current step, (iii) and difference of current predicted value with components failure thresholds. The proposed framework will be tested on turbine blade degradation data.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Nuclear activation analysis of zirconium-90 isomeric and ground-state reactions at the OMEGA Laser Facility

Nuclear activation is a well-established technique for inferring neutron yields in laser direct-drive deuterium–tritium (DT) and deuterium–deuterium (D 2 ) implosions at the OMEGA Laser Facility. Zirconium has long been considered an excellent candidate for measuring DT neutron fusion yields by observing decays of the 90 Zr(n,2n) 89 Zr ground-state reaction. As it has a higher energy threshold than present activation detectors utilizing copper, zirconium provides a means to infer primary neutron yields that are less susceptible to being skewed due to neutron scattering within the experimental environment. However, with a 78.41-h half-life, it is not operationally practical to utilize this reaction for OMEGA experiments, which have a 1-h shot cycle. Zirconium’s 90 Zr(n,2n) 89 mZr reaction presents itself as a viable candidate to infer neutron yields within a shot cycle, given its half-life of 4.16 min. Here, we present an overview of the approach and methodology, utilizing first principles techniques, to infer the primary neutron yields from OMEGA DT fusion experiments by using both the isomeric and the ground-state reaction. Yields inferred from both reactions are compared, which are in good agreement between the two.

Activation analysis↗

Operational Detection of Sun Glints in DSCOVR EPIC Images

Satellite images often feature sun glints caused by the specular reflection of sunlight from water surfaces or from horizontally oriented ice crystals occurring in clouds. Such glints can prevent accurate retrievals of atmospheric and surface properties using existing algorithms, but the glints can also be used to infer more about the glint-causing objects—for example about the microphysical properties and radiative effects of ice clouds. This paper introduces the recently released operational glint product of the Earth Polychromatic Camera (EPIC) onboard the Deep Space Climate Observatory (DSCOVR) spacecraft. Most importantly, the paper describes the algorithm used for generating the key component of the new product: a glint mask indicating the presence of sun glint caused by the specular reflection of sunlight from ice clouds and smooth water surfaces. After describing the glint detection algorithm and glint product, the paper shows some examples of the detected glints and discusses some basic statistics of the glint population in a yearlong dataset of EPIC images. These statistics provide insights into the performance of glint detection and point toward possibilities for using the glint product to gain scientific insights about ice clouds and water surfaces.

DSCOVR↗

Human reliability analysis studies from simulator experiments using Bayesian inference

Probabilistic Safety Assessment (PSA) of complex facilities is performed to arrive at the risk posed by them. PSA also accounts for the contribution of the human errors towards the overall risk through Human Reliability Analysis (HRA) in terms of Human Error Probability (HEP). Human operators are part of the system and do not work in isolation. Their performance is influenced by the context in which the actions are performed. As a result, quantification of HEP requires operator performance data under the given context. Some good sources of operator performance data are plant‘s operation data, simulator data and expert judgement. The plant operation data pertaining to HRA is generally sparse. In this situation, a full scope plant simulator provides a good alternative for operator performance data generation. Many of the currently practiced HRA methods have been developed by combining the empirical evidence with expert judgement and contain a lot of uncertainty in their estimates. Bayesian inference is suitable for updating the prior HRA estimates with the simulator evidence to obtain the posterior HEP. Here, posterior HEP has been calculated for postulated accident scenarios in advanced reactor (first of its kind) at design stage, using plant simulator.

97 MATHEMATICS AND COMPUTING↗

A Bayesian method for using simulator data to enhance human error probabilities assigned by existing HRA methods

In the past several years, several international organizations have begun to collect data on human performance in nuclear power plant simulators. The data collected provide a valuable opportunity to improve human reliability analysis (HRA), but these improvements will not be realized without implementation of Bayesian methods. Bayesian methods are widely used to incorporate sparse data into models in many parts of probabilistic risk assessment (PRA), but Bayesian methods have not been adopted by the HRA community. In this paper, we provide a Bayesian methodology to formally use simulator data to refine the human error probabilities (HEPs) assigned by existing HRA methods. We demonstrate the methodology with a case study, wherein we use simulator data from the Halden Reactor Project to update the probability assignments from the SPAR-H method. The case study demonstrates the ability to use performance data, even sparse data, to improve existing HRA methods. Furthermore, this paper also serves as a demonstration of the value of Bayesian methods to improve the technical basis of HRA.

99 GENERAL AND MISCELLANEOUS↗

NECESSARY CONDITIONS FOR SHORT GAMMA-RAY BURST PRODUCTION IN BINARY NEUTRON STAR MERGERS

The central engine of short gamma-ray bursts (sGRBs) is hidden from direct view, operating at a scale much smaller than that probed by the emitted radiation. Thus we must infer its origin not only with respect to the formation of the trigger—the actual astrophysical configuration that is capable of powering an sGRB—but also from the consequences that follow from the various evolutionary pathways that may be involved in producing it. Considering binary neutron star mergers we critically evaluate, analytically and through numerical simulations, whether the neutrino-driven wind produced by the newly formed hyper-massive neutron star can allow the collimated relativistic outflow that follows its collapse to actually produce an sGRB or not. Upon comparison with the observed sGRB duration distribution, we find that collapse cannot be significantly delayed (≤100 ms) before the outflow is choked, thus limiting the possibility that long-lived hyper-massive remnants can account for these events. In the case of successful breakthrough of the jet through the neutrino-driven wind, the energy stored in the cocoon could contribute to the precursor and extended emission observed in sGRBs.

79 ASTRONOMY AND ASTROPHYSICS↗

Cognitive IoT and Edge Computing for Intrusion Detection with Federated TinyML

Internet of Things (IoT) and Edge Computing (EC) are rapidly becoming an integral part of the modern society. By 2030, there is estimated to be over 40 billion active and connected IoT devices [1]. This rapid progress also comes with a significant implication on cybersecurity. Back-end infrastructure and systems have a much broader attack than they did previously due to vulnerable IoT/EC devices being connected to wireless networks. This expanding attack surface is a growing concern because IoT/EC are increasingly being used in critical systems such as power grids, health care, and smart homes. To effectively address a problem of this scale, cognitive cyber methods—which can autonomously detect and react to cyber attacks as they develop—are needed. To address this, we bring Artificial Intelligence (AI) and Machine Learning (ML) to IoT/EC devices, using tinyML to monitor voluminous IoT data against cyber threats, and using Federated Learning (FL) to share local detection knowledge across the system while preserving privacy. We propose a novel three-layer architecture: (1) an IoT layer for tinyML-based inference, (2) an edge layer for ML model training, and (3) a cloud layer for FL operations. Using the publicly available 11-class N-BaIoT dataset [2], we demonstrate that this architecture mitigates resource constraints at the IoT layer while improving detection accuracy over standard two-layer designs. An outlier-resistant scaler, feature reduction, and quantization enable the tinyML model to maintain detection accuracy with a reduced model size. Additionally, federated learning that only utilizes the intersection (across heterogenous devices) of the reduced feature set achieves superior detection accuracy compared to locally trained models.

Li, Mingyan [ORNL] (ORCID:0009000569532640)↗

A GPU-Accelerated Population Generation, Sorting, and Mutation Kernel for an Optimization-Based Causal Inference Model

We develop a GPU-accelerated machine learning generative adversarial network model that can be used with observational data for the purpose of constructing causal inferences. The theoretical basis of our machine learning model is novel and is conceptualized to be operable and scalable for high performance computing platforms. Our GPU-accelerated code enables large-scale parallelization of the computation within a common and accessible computing environment. This will expand the reach of our model and empower research in new substantive domains while maintaining the underlying theoretical properties.

Cho, Wendy K. Tam↗

The Building ENergy Demand (BEND) Model: A Comprehensive Composite Building Energy Modeling Framework

The Building ENergy Demand (BEND) model was developed at Pacific Northwest National Laboratory (PNNL) to simulate climate-dependent hourly building energy demand with the ability to disaggregate down to or aggregate up to any geographic area of interest, including counties, states, electric utility service territories, and census regions. The BEND modeling framework is designed to study the effects of climate change, economics, technology advancement, and various mitigation policies on aggregate building energy demand. To make the process computationally feasible, BEND reduces granularity in building characteristics and weather forcing while maintaining enough detail to minimize systematic differences (biases) between BEND results and data from the Energy Information Administration (EIA). Based on the Commercial Building Energy Consumption Survey (CBECS) and Residential Energy Consumption Survey (RECS) databases, BEND establishes a set of representative buildings with predefined sets of building types, sizes, and vintages. These representative buildings are defined for each of the four U.S. census regions and the five climate zones used by the EIA for building surveys (hereafter “EIA climate zones”). BEND’s representative buildings are based on a statistical analysis of the CBECS and RECS databases that infers detailed parameters such as construction type, thermal properties, internal loads, equipment and mechanical systems, and operation schedules from the high-level descriptors in those databases (building type, vintage, size, and location). The representative buildings are simplified and consolidated into a set of prototype buildings for the purpose of simulating building energy demand. Combining the prototype buildings with weather files from selected locations and years, BEND creates a set of simulation-ready building models to be run using the EnergyPlus simulation engine (EnergyPlus). The base spatial unit of each BEND simulation is a subregion that is the intersection of an International Energy Conservation Code climate zone, a census region, and an EIA climate zone; analysis for each spatial unit is influenced by the weather file BEND selects to represent the climate-similar area. The major simulation outputs of interest to BEND are hourly energy consumption values for each simulated year, possibly for multiple fuel types. Based on building-stock weights derived from the CBECS/RECS data, the hourly output streams are aggregated to a single, 8760-hour load profile for each of the major building categories—residential and commercial—with the possibility of segregating by fuel types. To correct for biases inherent in the use of a limited suite of representative buildings, BEND is calibrated against observational targets of total annual building energy consumption and hourly load profiles (including peak demand) obtained for the decision-relevant geographic scales of interest (e.g., total electric load data for individual balancing authorities available from EIA). Based on available data such as population distribution, BEND is able to aggregate both up and down scales, allowing the study of building energy demand over a range of spatial scales (e.g., states or counties within the Western Electricity Coordinating Council, Eastern Interconnection, or the entire U.S.) and/or subsets of building types (e.g., office buildings with particular size ranges).

24 POWER TRANSMISSION AND DISTRIBUTION↗

The effect of perturbations of convective energy transport on the luminosity and radius of the Sun

The response of solar models to perturbations of the efficiency of convective energy transport is studied for a number of cases. Such perturbations primarily effect the shallow superadiabatic layer of the convective envelope (at depth of approx. 1000 km below the photosphere). Independent of the details of the perturbation scheme, the resulting change in the solar radius is always very small compared to the change in luminosity. This appears to be true for any physical mechanism of solar variability which operates in the outer layers of the convection zone. Changes of the solar radius have been inferred from historical observations of solar eclipses. Considering the constraints on concurrent luminosity changes, this type of solar variability must be indicative of changes in the solar structure at substantial depths below the superadiabatic layer of the convective envelope.

Endal, A. S.↗

The effect of perturbation of convective energy transport on the luminosity and radius of the sun

The response of solar models to perturbations of the efficiency of convective energy transport is studied. Such perturbations primarily affect the shallow superadiabatic layer of the convective envelope. Independent of the details of the perturbation scheme, the resulting change in the solar radius is always very small compared to the change in luminosity. This appears to be true for any physical mechanism of solar variability which operates in the outer layers of the convection zone. Changes of the solar radius have been inferred from historical observations of solar eclipses in 1715 and 1925. Considering the constraints on concurrent luminosity changes, this type of solar variability must be indicative of changes in the solar structure at substantial depths below the superadiabatic layer of the convective envelope.

Endal, A. S.↗

Magnetic Field Observations at Purcell, Oklahoma Field Campaign Report

The campaign “Magnetic Field Observations at Purcell, Oklahoma” installed a ground-based magnetometer at Purcell’s U.S. Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) Climate Research Facility boundary installation at the Kessler Atmospheric and Ecological Field Station, University of Oklahoma, to measure local magnetic field variations. It is a part of the nine stations of the Mid-continent MAgnetoseismic Chain (McMAC) placed as close to the 330° magnetic longitude as possible. This is the meridian in the world where land covers the greatest continuous range in magnetic latitude. Figure 1 shows the map of the magnetometer stations along the 330th magnetic meridian, including the Purcell (PCEL) station. The main scientific objective of the campaign is to detect the field line resonance (FLR) frequencies of the magnetic field line connected to the Purcell station. This magnetic field line extends from Purcell to the outer space at distances as far as 2 Earth radii (RE). To accurately identify FLR frequencies, however, simultaneous measurements at slightly different latitudes along the same meridian are necessary to allow the use of the cross-phase technique. This consideration explains the arrangement to operate magnetometers at the Americus (AMER) and Richardson (RICH) stations nearby. The measured resonant frequency can infer the plasma mass density along the field line through the method of normal-mode magnetoseismology. The magnetometer at the Purcell station can detect many other types of magnetic field fluctuations associated with the changes in the electric currents in the ionosphere and the magnetosphere, which by large are affected by the solar activity. In other words, the magnetic field data collected by this campaign are also useful for understanding space weather phenomena. The magnetometer was installed at Purcell’s ARM boundary facility in March 27, 2006. The construction of the triaxial fluxgate magnetometer used by the campaign, as well as the data processing and analysis, was sponsored by the National Science Foundation. Except during occasional downtimes, the magnetometer collected measurements from June 6, 2006 to July 26, 2016.

47 OTHER INSTRUMENTATION↗

Comparisons of Wilks’ and Monte Carlo Methods in Response to the 10CFR50.46(c) Proposed Rulemaking

The Nuclear Regulatory Commission (NRC) is proposing a new rulemaking on emergency core system/loss-of-coolant accident (LOCA) performance analysis. In the proposed rulemaking, designated as 10CFR50.46(c), the US NRC put forward an equivalent cladding oxidation criterion as a function of cladding pre-transient hydrogen content. The proposed rulemaking imposes more restrictive and burnup-dependent cladding embrittlement criteria; consequently nearly all the fuel rods in a reactor core need to be analyzed under LOCA conditions to demonstrate compliance to the safety limits. New analysis methods are required to provide a thorough characterization of the reactor core in order to identify the locations of the limiting rods as well as to quantify the safety margins under LOCA conditions. With the new analysis method presented in this work, the limiting transient case and the limiting rods can be easily identified to quantify the safety margins in response to the proposed new rulemaking. In this work, the best-estimate plus uncertainty (BEPU) analysis capability for large break LOCA with the new cladding embrittlement criteria using the RELAP5-3D code is established and demonstrated with a reduced set of uncertainty parameters. Both the direct Monte Carlo method and the Wilks’ nonparametric statistical method can be used to perform uncertainty quantification. Wilks’ method has become the de-facto industry standard to perform uncertainty quantification in BEPU LOCA analyses. Despite its widespread adoption by the industry, the use of small sample sizes to infer statement of compliance to the existing 10CFR50.46 rule, has been a major cause of unrealized operational margin in today’s BEPU methods. Moreover the debate on the proper interpretation of the Wilks’ theorem in the context of safety analyses is not fully resolved yet, even more than two decades after its introduction in the frame of safety analyses in the nuclear industry. This represents both a regulatory and application risk in rolling out new methods. With the 10CFR50.46(c) proposed rulemaking, the deficiencies of the Wilks’ approach are further exacerbated. The direct Monte Carlo approach offers a robust alternative to perform uncertainty quantification within the context of BEPU analyses. In this work, the Monte Carlo method is compared with the Wilks’ method in response to the NRC 10CFR50.46(c) proposed rulemaking.

10CFR50.46(c)↗

Approach for Inferring Full-Scope Human Reliability Data Based on Simplified Simulator Data

This paper proposes a method for inferring full-scope human reliability data based on the Simplified Human Error Experimental Program (SHEEP) data. It mainly focuses on the human errors observed when using simulators with different complexity levels. In the proposed method, the manner in which human error probabilities (HEPs) change as a result of increasing simulator complexity and how simulator complexity levels are quantified represent key information for inferring full-scope data. In the present study, SHEEP error data pertaining to actual professional operators using Rancor Microworld (Rancor) (i.e., a more simplified simulator) and Compact Nuclear Simulator (CNS) (i.e., a less simplified simulator) were compared with the HuREX error data. An approach to quantifying simulator complexity levels was then proposed based on information theory and acquired eye-tracker data.

99 - GENERAL AND MISCELLANEOUS↗

Predicting and Managing Risk to Bats at Commercial Wind Farms using Acoustics

Bat populations in North America face novel threats from white-nose syndrome and widespread turbine-related mortality related to the rapidly expanding wind power industry in addition to long-standing pressures from habitat loss and degradation. Bats, unlike most small mammals, are long-lived and slow to reproduce, highlighting the importance of understanding and managing anthropogenic sources of mortality. My dissertation research used acoustic bat detectors to measure bat activity at commercial wind projects, predict patterns in risk, and design strategic measures to reduce fatality rates by curtailing turbine operation during periods when bats are most active. Bats collide with wind turbines only when their rotors are spinning, and risk of turbine-related fatality is therefore a dynamic factor that can be manipulated by curtailing turbine operation when bats are active. We first measured inter-detector variation in metrics of acoustic bat activity to understand how the acoustic detection process may affect inferences related to spatial and temporal variation in bat activity. Using acoustic detectors mounted on top of wind turbines at two commercial wind farms in West Virginia, we then demonstrated that the amount of bat activity recorded when turbines were operating aligned closely with bat fatality rates on multiple scales. Accordingly, the metric of bat activity exposed to turbine operation provides a meaningful, quantitative indicator of turbine-related bat fatality risk. Further, bats responded consistently to changing wind speed and temperature at turbines in both wind farms across multiple years, enabling exposed bat activity to be predicted accurately among turbines and years. Building on these results, we simulated exposure of bats to turbine operation and energy loss for curtailment strategies recommended by state and federal agencies in the United States and Canada. By adjusting parameters such as cut-in wind speeds and temperature thresholds, we demonstrated the ability to design strategic curtailment programs that achieve equivalent or greater predicted reductions in bat activity exposure for substantially less energy-production loss. Characterizing fatality risk on a finer scale using acoustics will help regulatory agencies and the wind industry alike reduce risks of population-level impacts to vulnerable bat species while continuing to expand large-scale renewable energy generation.

17 WIND ENERGY↗