Search NASA⌕ Search

SEARCH · Search NASA

Results for “Distributed training”

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 325 records · Page 18

Modeling hadronization using machine learning

We present the first steps in the development of a new class of hadronization models utilizing machine learning techniques. We successfully implement, validate, and train a conditional sliced-Wasserstein autoencoder to replicate the Pythia generated kinematic distributions of first-hadron emissions, when the Lund string model of hadronization implemented in Pythia is restricted to the emissions of pions only. The trained models are then used to generate the full hadronization chains, with an IR cutoff energy imposed externally. The hadron multiplicities and cumulative kinematic distributions are shown to match the Pythia generated ones. We also discuss possible future generalizations of our results.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Data augmentation for disruption prediction via robust surrogate models

The goal of this work is to generate large statistically representative data sets to train machine learning models for disruption prediction provided by data from few existing discharges. Such a comprehensive training database is important to achieve satisfying and reliable prediction results in artificial neural network classifiers. Here, we aim for a robust augmentation of the training database for multivariate time series data using Student t process regression. We apply Student t process regression in a state space formulation via Bayesian filtering to tackle challenges imposed by outliers and noise in the training data set and to reduce the computational complexity. Thus, the method can also be used if the time resolution is high. We use an uncorrelated model for each dimension and impose correlations afterwards via colouring transformations. We demonstrate the efficacy of our approach on plasma diagnostics data of three different disruption classes from the DIII-D tokamak. To evaluate if the distribution of the generated data is similar to the training data, we additionally perform statistical analyses using methods from time series analysis, descriptive statistics and classic machine learning clustering algorithms.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Data augmentation for disruption prediction via robust surrogate models

The goal of this work is to generate large statistically representative datasets to train machine learning models for disruption prediction provided by data from few existing discharges. Such a comprehensive training database is important to achieve satisfying and reliable prediction results in artificial neural network classifiers. Here, we aim for a robust augmentation of the training database for multivariate time series data using Student-t process regression. We apply Student-t process regression in a state space formulation via Bayesian filtering to tackle challenges imposed by outliers and noise in the training data set and to reduce the computational complexity. Thus, the method can also be used if the time resolution is high. We use an uncorrelated model for each dimension and impose correlations afterwards via coloring transformations. We demonstrate the efficacy of our approach on plasma diagnostics data of three different disruption classes from the DIII-D tokamak. To evaluate if the distribution of the generated data is similar to the training data, we additionally perform statistical analyses using methods from time series analysis, descriptive statistics, and classic machine learning clustering algorithms.

97 MATHEMATICS AND COMPUTING↗

Early implementation of digital upgrades on simulators - 347

An important aspect of successful Nuclear Power Plant (NPP) digital control system upgrade projects is to upgrade the Operator Training Simulator (OTS) or a related Engineering Simulator (ES), prior to installation of the Distributed Control System (DCS) on the plant allowing for early testing of the digital controls/HMI and for early training of the plant operators while maintaining OTS configuration management (i.e. OTS in sync with the plant's current design). Experience has shown that early implementation of digital upgrades on the OTS and/or an ES can reduce plant startup duration (related to digital systems commissioning) and ensure operators are proficient with the new digital systems to ensure safe and reliable plant operation. Including simulators in the design process offers many benefits including early identification of design deficiencies, testing deficiency fixes, developing and validating plant procedures and supporting human factors engineering, to name a few. Implementing digital upgrades on the OTS is also a regulatory requirement driven by the ANSI/ANS-3.5-2009 standard (Nuclear Power Plant Simulators for Use in Operator Training and Examination) if the new DCS affects actions to be taken by the plant operator in the main control room. This paper presents the benefits of implementing digital upgrades on the OTS and/or ES prior to plant implementation. The various simulator implementation strategies (stimulation, emulation, simulation or a hybrid solution) and selection factors are discussed. The paper also covers four (4) case studies demonstrating how digital upgrades can be implemented on simulators. The first case study covers the implementation of a Siemens offgas Programmable Logic Controller (PLC) system on the Enrico Fermi OTS (Michigan). The second case study covers the implementation of a GE main generator excitation system on the Cernavoda OTS (Romania). The third case study covers the implementation of a Siemens main feedwater control system on the Callaway OTS and ES (Missouri). The final case study covers the implementation of the non-safety related and safety related DCS on the Nuclear New Build (NNB) Hongyanhe Phase I dual-purpose OTS/ES (China). (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

An out-of-distribution-aware autoencoder model for reduced chemical kinetics

While detailed chemical kinetic models have been successful in representing rates of chemical reactions in continuum scale computational fluid dynamics (CFD) simulations, applying the models in simulations for engineering device conditions is computationally prohibitive. To reduce the cost, data-driven methods, e.g., autoencoders, have been used to construct reduced chemical kinetic models for CFD simulations. Despite their success, data-driven methods rely heavily on training data sets and can be unreliable when used in out-of-distribution (OOD) regions (i.e., when extrapolating outside of the training set). In this paper, we present an enhanced autoencoder model for combustion chemical kinetics with uncertainty quantification to enable the detection of model usage in OOD regions, and thereby creating an OOD-aware autoencoder model that contributes to more robust CFD simulations of reacting flows.Here, we first demonstrate the effectiveness of the method in OOD detection in two well-known datasets, MNIST and Fashion-MNIST, in comparison with the deep ensemble method, and then present the OOD-aware autoencoder for reduced chemistry model in syngas combustion.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

O'Hare Airport roadway traffic prediction via data fusion and Gaussian process regression

This study proposes an approach of leveraging information gathered from multiple traffic data sources at different resolutions to obtain approximate inference on the traffic distribution of Chicago's O'Hare Airport area. Specifically, it proposes the ingestion of traffic datasets at different resolutions to build spatiotemporal models for predicting the distribution of traffic volume on the road network. Due to its good adaptability and flexibility for spatiotemporal data, the Gaussian process (GP) regression was employed to provide short-term forecasts using data collected by loop detectors (sensors) and supplemented by telematics data. The GP regression is used to make predictions of the distribution of the proportion of sensor data traffic volume represented by the telematics data for each location of the sensors. Consequently, the fitted GP model can be used to determine the approximate traffic distribution for a testing location outside of the training points. Policymakers in the transportation sector can find the results of this work helpful for making informed decisions relating to current and future transportation conditions in the area.

42 ENGINEERING↗

Integrated analysis of X-ray diffraction patterns and pair distribution functions for machine-learned phase identification

Abstract To bolster the accuracy of existing methods for automated phase identification from X-ray diffraction (XRD) patterns, we introduce a machine learning approach that uses a dual representation whereby XRD patterns are augmented with simulated pair distribution functions (PDFs). A convolutional neural network is trained directly on XRD patterns calculated using physics-informed data augmentation, which accounts for experimental artifacts such as lattice strain and crystallographic texture. A second network is trained on PDFs generated via Fourier transform of the augmented XRD patterns. At inference, these networks classify unknown samples by aggregating their predictions in a confidence-weighted sum. We show that such an integrated approach to phase identification provides enhanced accuracy by leveraging the benefits of each model’s input representation. Whereas networks trained on XRD patterns provide a reciprocal space representation and can effectively distinguish large diffraction peaks in multi-phase samples, networks trained on PDFs provide a real space representation and perform better when peaks with low intensity become important. These findings underscore the importance of using diverse input representations for machine learning models in materials science and point to new avenues for automating multi-modal characterization.

36 MATERIALS SCIENCE↗

High-Fidelity Multiphysics Modeling of a Heat Pipe Microreactor Using BlueCrab

Researchers who are actively developing nuclear microreactors are planning to employ innovative designs and features using traditional commercial modeling tools that may be inadequate for their design and licensing activities. The codes developed under the U.S. Department of Energy Office of Nuclear Energy Advanced Modeling and Simulation (NEAMS) program provide flexibility in terms of geometry modeling and multiphysics coupling and are particularly well suited for modeling novel microreactor concepts. To test the maturity of these codes, this paper introduces a conceptual heat pipe microreactor (HP-MR) designed to gather various technologies of interest to microreactor developers such as control drums, heat pipes, and hydride moderators. Here, the objective of this effort is to demonstrate NEAMS tools capability to perform high-fidelity multiphysics simulations, using coupled neutronics (via the Griffin code), heat conduction (via the BISON code), heat pipe modeling (via the Sockeye code), and hydrogen redistribution in hydride metal moderator (via the SWIFT code). Codes are coupled in-memory through the Multiphysics Object-Oriented Simulation Environment (MOOSE) framework, which permits flexible multiphysics data transfer schemes. The analysis confirmed two key aspects of the HP-MR concept: (1) its ability to follow the power load requested from the heat pipe and (2) its ability to avoid heat pipe cascading failure unless designed with high power close to operating failure limits of its heat pipes. The developed computational model was distributed publicly on the Virtual Test Bed for training purposes to accelerate adoption by industry and to provide a high-fidelity multiphysics solution for benchmarking against other tools. Additional multiphysics analyses including other transients and coupled physics were identified as necessary future work, together with a focus on validating multiphysics behavior against experiments.

Microreactor↗

Calomplification — the power of generative calorimeter models

Motivated by the high computational costs of classical simulations, machine-learned generative models can be extremely useful in particle physics and elsewhere. They become especially attractive when surrogate models can efficiently learn the underlying distribution, such that a generated sample outperforms a training sample of limited size. This kind of GANplification has been observed for simple Gaussian models. We show the same effect for a physics simulation, specifically photon showers in an electromagnetic calorimeter.

47 OTHER INSTRUMENTATION↗

Out of Distribution Detection with Neural Network Anchoring

This is code to reproduce and build on OOD detection from the paper "Out of Distribution Detection with Neural Network Anchoring". Our goal here is to exploit heteroscedastic temperature scaling as a calibration strategy for out of distribution (OOD) detection. Heteroscedasticity here refers to the fact that the optimal temperature parameter for each sample can be different, as opposed to conventional approaches that use the same value for the entire distribution. To enable this, we propose a new training strategy called anchoring that can estimate appropriate temperature values for each sample, leading to state-of-the-art OOD detection performance across several benchmarks. Using NTK theory, we show that this temperature function estimate is closely linked to the epistemic uncertainty of the classifier, which explains its behavior. In contrast to some of the best-performing OOD detection approaches, our method does not require exposure to additional outlier datasets, custom calibration objectives, or model ensembling. Through empirical studies with different OOD detection settings - far OOD, near OOD, and semantically coherent OOD - we establish a highly effective OOD detection approach.

Thiagarajan, Jayaraman↗

Machine learning applied to wind turbine blades impact detection

The significant development of wind power generation worldwide brings, together with environmental benefits, wildlife concerns, especially for volant species vulnerability to interactions with wind energy facilities. For surveying such events, an automatic system for continuous monitoring of blade collisions is critical. An onboard multi-senor system capable of providing real-time collision detection using integrated vibration sensors is developed and successfully tested. However, to detect low signal-to-noise ratio impact can be challenging; hence, an advanced impact detection method has been developed and presented in this article. A robust automated detection algorithm based on support vector machine is proposed. After a preliminary signal pre-processing, geometric features specifically selected for their sensitivity to impact signals are extracted from raw vibration signal and energy distribution graph. The predictive model is formulated by training conventional support vector machine using extracted features for impact identification. Finally, the performance of the predictive model is evaluated by accuracy, precision, and recall. Results indicate a linear regression relationship between signal-to-noise ratio and model overall performance. Here, the proposed method is much reliable on higher signal-to-noise ratio (SNR≥6), but it shows to be ineffective at lower signal-to-noise ratio (SNR<2).

17 WIND ENERGY↗

Beam loss modeling and mitigation due to intra-beam stripping

Intra-Beam Stripping (IBS) is a critical beam loss mechanism in high-intensity H- linacs and presents a significant limitation to increasing beam power. This work presents a computational framework to evaluate and mitigate IBS-induced beam loss along the Spallation Neutron Source (SNS) LINAC. Our calculation is based on an analytic theory and involves evaluation of a 9D integral using the Monte-Carlo technique. We first benchmarked our calculations against simplified, analytically solvable cases. We then applied our algorithm to Gaussian bunches with a known probability density function (PDF). We next expanded our algorithm to arbitrary bunch distributions using the Neural Spline Flow (NSF) models trained on PyORBIT tracking data. In the future, we plan to validate our algorithm experimentally and apply it to design IBS mitigation strategies.

Nln, Shivam [ORNL]↗

Survey of PFAS Treatment Technologies

Per- and polyfluoroalkyl substances (PFAS) are a group of manufactured chemicals that have caught the attention of many governmental agencies and researchers. These chemicals are highly stable and possess hydrophobic properties that lead to the widespread use of PFAS in aqueous film forming foams, and household products such as carpets, paper and non-stick cookware (Prevedouros et al., 2006). Regulatory and health advisories have focused determining a lifetime drinking water health advisory of 70 ng/L for PFOA and PFOS (USEPA, 2016). Other PFASs may be present in the environment, along with PFOA and PFOS, due to product manufacturing processes (Prevedouros et al., 2006). A variety of proprietary AFFF formulations were manufactured and sold during different timeframes, thereby complicating the signature and distribution of PFASs in groundwater beneath former firefighting training areas (McGuire et al., 2014). PFOA and PFOS are stable chemicals that persistent in the environment and are difficult to remediate. The understanding of the physicochemical properties and fate and transport of PFASs in groundwater is growing but is still limited (USEPA, 2012). Many traditional approaches such as coagulation and conventional water treatment are ineffective in treating PFOA and PFOS. Selecting the most appropriate remedial strategy for PFOA and PFOS is therefore challenging. In this survey findings from the literature on PFASs removal using various technologies were compiled.

54 ENVIRONMENTAL SCIENCES↗

Machine Learning for Neutron Resonance Evaluations [Slides]

The performance of nuclear reactors and other nuclear systems depends on a precise understanding of the neutron interaction cross sections for materials used in these systems. These cross sections exhibit a resonance structure whose shape is determined in part by the angular momentum quantum numbers of the resonances. The correct assignment of the quantum numbers of neutron resonances is therefore of paramount importance. In this presentation, we describe the application of machine learning to automate the quantum number assignments. Scikit-learn classifiers were trained on simulated resonance data whose statistical properties were chosen to mimic real data. We explored the use of several physics (and random matrix theory)-motivated features for training the classifiers, including the nearest neighbor spacing distribution, cumulative level distribution, and channel width distributions. Initial results demonstrated that we can determine resonance spin groups somewhat reliably. We are now investigating the application of our approach to 52 Cr resonance data.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Grid-Ready Energy Analytics Training with Data (“GREAT with Data”)

GridEd is a collaborative educational initiative consisting of the Electric Power Research Institute (EPRI), 5 Partner Universities (Stony Brook University, The University of Texas at Austin, University of California – Riverside, Virginia Tech, Washington State University), and participating industry sponsors. This educational initiative focuses on developing and training the next generation of power engineers so they can help shape the electric grid of the future by anticipating and fulfilling the needs of changing electric industry requirements. GridEd is leveraging electric industry research to educate a future electric grid workforce by empowering new and continuing education students, not only to become competent and well-informed engineers, but also to participate and influence major technological, social, and policy decisions that address critical global challenges. GridEd’s activities are centered around four core pillars: Enhancement of university power systems engineering curricula; Professional development and training for a diverse electric industry workforce; Stimulating students to join the movement for the next generation of power engineers, and; Improve workforce development efforts in the electric utility industry. Major accomplishments over the course of the project were: Over 50 unique professional short courses were delivered by more than 30 instructors across the GridEd network; Over 3,600 unique learners, many who took multiple courses, received more than 27,000 professional development hours (PDH) and over 1,000 certificates of completion; Approximately 2,900 unique learners undertook a course that was offered LIVE online or in-person; Approximately 700 unique learners undertook a course that was offered as computer-based training (CBT); Over 35 unique university courses were delivered by more than 30 instructors to 1,500 university students across the GridEd network; One-hundred-and-fifty-seven (157) students were funded to completed 43 student projects in topics of power systems and data science, and; Six (6) Historically Black Colleges and Universities (HBCUs) recruited as Affiliate Universities via Utility Partners. The professional training initiative and workforce development activities launched by this project will be sustained through EPRI’s collaborative business model with industry. Stimulating students to join the power engineering workforce of the future and the enhancement of tertiary training may continue to need government support.

14 SOLAR ENERGY↗

Simulation of e-cloud driven instability and its attenuation using a simulated feedback system in the CERN SPS

Electron clouds have been shown to trigger fast growing instabilities on proton beams circulating in the SPS, and a feedback system to control the single-bunch instabilities is under active development. We present the latest improvements to the WARP-POSINST simulation framework and feedback model, and its application to the self-consistent simulations of two consecutive bunches interacting with an electron cloud in the SPS. Simulations using an idealized feedback system exhibit adequate mitigation of the instability providing that the cutoff of the feedback bandwidth is at or above 450 MHz. Artifacts from numerical noise of the injected distribution of electrons in the modeling of portions of bunch trains are discussed, and benchmarking of WARP against POSINST and HEADTAIL are presented.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Application of machine learning to characterize gas hydrate reservoirs in Mackenzie Delta (Canada) and on the Alaska north slope (USA)

Here, artificial neural network-trained models were used to predict gas hydrate saturation distributions in permafrost-associated deposits in the Eileen Gas Hydrate Trend on the Alaska North Slope (ANS), USA and at the Mallik research site in the Beaufort-Mackenzie Basin, Northwest Territories, Canada. The database of Logging-While-Drilling (LWD) and wireline logs collected at five wells (Mount Elbert, Ignik Sikumi, and Kuparuk 7–11–12 wells at ANS, plus 2L-38 and 5L-38 wells at the Mallik research site) includes more than 10,000 depth points, which were used for training, validation, and testing the machine learning (ML) models. Data used in training the ML models include the well logs of density, porosity, electrical resistivity, gamma radiation, and acoustic wave velocity measurements. Combinations of two or three out of these five well logs were found to reliably predict the gas hydrate saturation with accuracy varying between 80 and 90% when compared to the gas hydrate saturations derived from Nuclear Magnetic Resonance (NMR)-based technique. The ML models trained on data from three ANS wells achieved high fidelity predictions of gas hydrate saturation at the Mallik site. The results obtained in this study indicate that ML models trained on data from one geological basin can successfully predict key reservoir parameters for permafrost-associated gas hydrate accumulations within another basin. A generalized approach for selecting a well log combination that can improve model accuracy is discussed. Overall, the study outcome supports earlier work demonstrating that ML models trained on non-NMR well logs are a viable alternative to physics-driven methods for predicting gas hydrate saturations.

58 GEOSCIENCES↗