Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “network reduction”

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 181 records · Page 10
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Integration of Decentralized Graph-Based Multi-Agent Reinforcement Learning with Digital Twin for Traffic Signal Optimization

Machine learning (ML) methods, particularly Reinforcement Learning (RL), have gained widespread attention for optimizing traffic signal control in intelligent transportation systems. However, existing ML approaches often exhibit limitations in scalability and adaptability, particularly within large traffic networks. This paper introduces an innovative solution by integrating decentralized graph-based multi-agent reinforcement learning (DGMARL) with a Digital Twin to enhance traffic signal optimization, targeting the reduction of traffic congestion and network-wide fuel consumption associated with vehicle stops and stop delays. In this approach, DGMARL agents are employed to learn traffic state patterns and make informed decisions regarding traffic signal control. The integration with a Digital Twin module further facilitates this process by simulating and replicating the real-time asymmetric traffic behaviors of a complex traffic network. The evaluation of this proposed methodology utilized PTV-Vissim, a traffic simulation software, which also serves as the simulation engine for the Digital Twin. The study focused on the Martin Luther King (MLK) Smart Corridor in Chattanooga, Tennessee, USA, by considering symmetric and asymmetric road layouts and traffic conditions. Comparative analysis against an actuated signal control baseline approach revealed significant improvements. Experiment results demonstrate a remarkable 55.38% reduction in Eco_PI, a developed performance measure capturing the cumulative impact of stops and penalized stop delays on fuel consumption, over a 24 h scenario. In a PM-peak-hour scenario, the average reduction in Eco_PI reached 38.94%, indicating the substantial improvement achieved in optimizing traffic flow and reducing fuel consumption during high-demand periods. These findings underscore the effectiveness of the integrated DGMARL and Digital Twin approach in optimizing traffic signals, contributing to a more sustainable and efficient traffic management system.

42 ENGINEERING↗
DOE OSTI2021

Model fusion with physics-guided machine learning: Projection-based reduced-order modeling

The unprecedented amount of data generated from experiments, field observations, and large-scale numerical simulations at a wide range of spatiotemporal scales has enabled the rapid advancement of data-driven and especially deep learning models in the field of fluid mechanics. Although these methods are proven successful for many applications, there is a grand challenge of improving their generalizability. This is particularly essential when data-driven models are employed within outer-loop applications like optimization. In this work, we put forth a physics-guided machine learning (PGML) framework that leverages the interpretable physics-based model with a deep learning model. Leveraging a concatenated neural network design from multi-modal data sources, the PGML framework is capable of enhancing the generalizability of data-driven models and effectively protects against or inform about the inaccurate predictions resulting from extrapolation. We apply the PGML framework as a novel model fusion approach combining the physics-based Galerkin projection model and long- to short-term memory (LSTM) network for parametric model order reduction of fluid flows. We demonstrate the improved generalizability of the PGML framework against a purely data-driven approach through the injection of physics features into intermediate LSTM layers. Our quantitative analysis shows that the overall model uncertainty can be reduced through the PGML approach, especially for test data coming from a distribution different than the training data. Moreover, we demonstrate that our approach can be used as an inverse diagnostic tool providing a confidence score associated with models and observations. The proposed framework also allows for multi-fidelity computing by making use of low-fidelity models in the online deployment of quantified data-driven models.

42 ENGINEERING↗
DOE OSTI2018

Autoencoder node saliency: Selecting relevant latent representations

The autoencoder is an artificial neural network that performs nonlinear dimension reduction and learns hidden representations of unlabeled data. With a linear transfer function it is similar to the principal component analysis (PCA). While both methods use weight vectors for linear transformations, the autoencoder does not come with any indication similar to the eigenvalues in PCA that are paired with eigenvectors. In this work, we propose a novel autoencoder node saliency method that examines whether the features constructed by autoencoders exhibit properties related to known class labels. The supervised node saliency ranks the nodes based on their capability of performing a learning task. It is coupled with the normalized entropy difference (NED). We establish a property for NED values to verify classifying behaviors among the top ranked nodes. Lastly, by applying our methods to real datasets, we demonstrate their ability to provide indications on the performing nodes and explain the learned tasks in autoencoders.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2015

Elucidating the sole contribution from electromagnetic near-fields in plasmon-enhanced Cu 2 O photocathodes

Despite many promising reports of plasmon-enhanced photocatalysis, the inability to identify the individual contributions from multiple enhancement mechanisms has delayed the development of general design rules for engineering efficient plasmonic photocatalysts. Herein, we construct a plasmonic photocathode comprised of Au@SiO 2 (core@shell) nanoparticles embedded within a Cu 2 O nanowire network to exclusively examine the contribution from one such mechanism: electromagnetic near-field enhancement. The influence of the local electromagnetic field intensity is correlated with the overall light-harvesting efficiency of the device through variation of the SiO 2 shell thickness (5—22 nm) to systematically tailor the distance between the plasmonic Au nanoparticles and the Cu 2 O nanowires. A three-fold increase in device photocurrent is achieved upon integrating the Au@SiO 2 nanoparticles into the Cu 2 O nanowire network, further enabling a ~40% reduction in semiconductor film thickness while maintaining photocathode performance. Photoelectrochemical results are further correlated with photoluminescence studies and optical simulations to confirm that the near-field enhancement is the sole mechanism responsible for increased light absorption in the plasmonic photocathode.

77 NANOSCIENCE AND NANOTECHNOLOGY↗
DOE OSTI2024

Insights on the structure and properties of sodium iron phosphate glasses from molecular dynamics simulations

Iron phosphate glasses are promising nuclear waste forms while more detailed understanding of their structures and structure-property relations are still needed to better design waste glass compositions. In this work we report studies of three series of sodium iron phosphate (NFP) glasses: 60P 2 O 5 -(40-x)Fe 2 O 3 -xNa 2 O (x = 0→40), (100–2x)P 2 O 5 -xFe 2 O 3 -xNa 2 O (x = 5→17.5) and one with different iron redox ratio, to understand the composition as well as the iron redox effects on the structure and properties of these glasses using molecular dynamics simulations with effective two-body and three-body potentials. Structural analyses, including pair distribution function, bond angle distribution, Q n distribution, and polyhedral connectivity, were performed to obtain in-depth information on short-range and medium-range structural features. The P-O pair distributions showed a first peak splitting with phosphorus-bridging and non-bridging oxygen contributions. This and the average P-O and other cation-oxygen bond distances are in excellent agreement with experiments. The coordination number of P 5+ remained four while that of Fe 3+ increased from 4.30 to 4.72 with decreasing Fe/Na ratio. Polyhedral linkage analysis showed [PO 4 ] units linked with [PO 4 ] and [FeO x ] through corner-sharing while the [PO 4 ]-[FeO x ] linkages become dominant for compositions with Fe 2 O 3 larger than 15 mol%. The effect of iron redox ratio on the structure of NFP glasses was also studied and it was found that bond lengths and coordination numbers were not strongly affected, while the reduction of iron introduced higher network distortions, as evident by O-P-O bond angle and Q n distribution. The glass transition temperature (T g ) showed a monotonic increase with Fe 2 O 3 in the first series, in good agreement with experiments, while those of the second series showed a maximum at P 2 O 5 = 82 mol%. Here, calculated elastic moduli were found to increase with Fe 2 O 3 in the first glass series, which was be explained by the increase of network connectivity, while those of the second series decrease with Fe 2 O 3 due to decrease of P 2 O 5 .

36 MATERIALS SCIENCE↗
DOE OSTI2013

Role of oxygen hole centres in the photodarkening of ytterbium-doped phosphosilicate fibre

We have studied the photodarkening in active fibres with an ytterbium-doped phosphosilicate glass core under IR irradiation with a pump source (920 nm) and UV irradiation (193 nm). Analysis of absorption and luminescence spectra suggests that such irradiations produce phosphorus – oxygen – hole centres (P-OHCs) in the core glass network and lead to the reduction of the ytterbium ions to a divalent state (Yb{sup 2+}). The photoinduced optical loss in the fibres in the visible range (400 – 700 nm) is mainly due to absorption by the P-OHCs. A quantum-mechanical model is proposed for P-OHC and Yb{sup 2+} formation. (nonlinear optical phenomena)

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2016

Time-of-flight mass measurements of neutron-rich chromium isotopes up to N = 40 and implications for the accreted neutron star crust

Here, we present the mass excesses of 59-64 Cr, obtained from recent time-of-flight nuclear mass measurements at the National Superconducting Cyclotron Laboratory at Michigan State University. The mass of 64Cr is determined for the first time, with an atomic mass excess of -33.48(44) MeV. We find a significantly different two-neutron separation energy S 2n trend for neutron-rich isotopes of chromium, removing the previously observed enhancement in binding at N = 38. Additionally, we extend the S 2n trend for chromium to N = 40, revealing behavior consistent with the previously identified island of inversion in this region. We compare our results to state-of-the-art shell-model calculations performed with a modified Lenzi-Nowacki-Poves-Sieja interaction in the fp shell, including the g 9/2 and d 5/2 orbits for the neutron valence space. We employ our result for the mass of 64 Cr in accreted neutron star crust network calculations and find a reduction in the strength and depth of electron-capture heating from the A = 64 isobaric chain, resulting in a cooler than expected accreted neutron star crust. This reduced heating is found to be due to the >1-MeV reduction in binding for 64 Cr with respect to values from commonly used global mass models.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2023

A General Spatiotemporal Imputation Framework for Missing Sensor Data

Many applications from precision agriculture, environmental monitoring and transportation networks rely on data collected across space and time over a large geographic area. Missing data poses a significant challenge for any data-driven inference and control tasks. Data imputation or the estimation of missing data can help fill these gaps by utilizing inherent spatial relationships and temporal patterns. A variety of spatiotemporal imputation models have been developed to address missing data in spatiotemporal datasets. However, these classical methods rely on the assumption that the underlying data follows a smooth trend and fail to provide accurate estimates when there is a large number of missing points in the data. Even though there are machine learning driven tensor completion approaches such as convolutional neural network based tensor completion (CoSTCo) that capture the non-linear relationships in the dataset, the transductive nature makes the algorithm less scalable. Thus, existing approaches for estimating the missing information do not effectively capture all dimensions of the spatiotemporal data structure, resulting in erroneous predictions and poor performance. The main contributions of this paper are: (1) We propose a novel inductive framework (G-LSTM) for missing data imputation that integrates a graph neural network with LSTMs to effectively capture both spatial and temporal dependencies. (2) Experimental results on a traffic dataset demonstrate that the proposed GNN integrated with an LSTM framework achieves improved imputation and maintains steady performance even when there are extreme missing conditions in comparison with the state-of-the-art imputation framework (i.e, CoSTCo). (3) The simulation results on a traffic network show up to 69% reduction in mean absolute error and 61% reduction in root mean square error when compared to CoSTCo.

Tharzeen, Aabila↗
DOE OSTI2023

Gray-Box Modeling for Distribution Systems with Inverter-Based Resources

In this paper, we develop a novel gray-box modeling approach for distribution systems with inverter-based resources (IBRs). The proposed gray-box modeling method aims to improve estimation accuracy by taking advantages of both physics-based (white-box) and data-driven (black-box) modeling approaches. To this end, the gray-box modeling framework is constructed by encoding prior physical knowledge of the system into a whitebox model and then embedding the output variables of the white-box model into the input vector of a black-box model. Especially, the white-box modeling component is constructed alongside the equivalent network model simplified with Kron reduction. Furthermore, case studies demonstrate that our graybox modeling approach effectively improves estimation accuracy compared to purely physics-based or data-driven methods.

Zhang, Junhui↗
DOE OSTI2020

Estimating Lossy Compressibility of Scientific Data Using Deep Neural Networks

Simulation based scientific applications generate increasingly large amounts of data on high-performance computing (HPC) systems. To allow data to be stored and analyzed efficiently, data compression is often utilized to reduce the volume and velocity of data. However, a question often raised by domain scientists is the level of compression that can be expected so that they can make more informed decisions, balancing between accuracy and performance. In this letter, we propose a deep neural network based approach for estimating the compressibility of scientific data. To train the neural network, we build both general features as well as compressor-specific features so that the characteristics of both data and lossy compressors are captured in training. Our approach is demonstrated to outperform a prior analytical model as well as a sampling based approach in the case of a biased estimation, i.e., for SZ. However, for the unbiased estimation (i.e., ZFP), the sampling based approach yields the best accuracy, despite the high overhead involved in sampling the target dataset.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2015

Critical biogeochemical functions in the subsurface are associated with bacteria from new phyla and little studied lineages

Nitrogen, sulfur and carbon fluxes in the terrestrial subsurface are determined by the intersecting activities of microbial community members, yet the organisms responsible are largely unknown. Metagenomic methods can identify organisms and functions, but genome recovery is often precluded by data complexity. To address this limitation, we developed subsampling assembly methods to re-construct high-quality draft genomes from complex samples. Here, we applied these methods to evaluate the interlinked roles of the most abundant organisms in biogeochemical cycling in the aquifer sediment. Community proteomics confirmed these activities. The eight most abundant organisms belong to novel lineages, and two represent phyla with no previously sequenced genome. Four organisms are predicted to fix carbon via the Calvin Benson Bassham, Wood Ljungdahl or 3-hydroxyproprionate/4-hydroxybutarate pathways. The profiled organisms are involved in the network of denitrification, dissimilatory nitrate reduction to ammonia, ammonia oxidation and sulfate reduction/oxidation, and require substrates supplied by other community members. An ammonium-oxidizing Thaumarchaeote is the most abundant community member, despite low ammonium concentrations in the groundwater. Finally, this organism likely benefits from two other relatively abundant organisms capable of producing ammonium from nitrate, which is abundant in the groundwater. Overall, dominant members of the microbial community are interconnected through exchange of geochemical resources.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2023

Model simulations of Plum Island Ecosystems LTER low marsh site using ELM-PFLOTRAN

Model simulations using the E3SM Land Model (ELM) coupled to the PFLOTRAN reactive transport model via the Alquimia interface. The simulations were conducted for a tidal salt marsh at the Plum Island Ecosystems LTER near Rowley, Massachusetts, USA. Model simulations were forced using site-specific tidal cycles and salinity, and the simulations used a biogeochemical reaction network including aerobic decomposition, sulfate reduction, iron reduction, and methanogenesis. Model outputs include simulated carbon stocks, carbon dioxide and methane fluxes, and porewater concentrations of key solutes related to sulfur, iron, and carbon cycling. The model simulations included a saline simulation (with tidal sulfate inputs), a fresh simulation (with low salinity and low sulfate inputs), and a saline simulation with lower vegetation productivity to represent the effect of salinity on vegetation. These simulations were conducted to demonstrate that a new model framework incorporating subsurface redox and biogeochemical interactions into a land surface model could reproduce measured surface greenhouse gas fluxes and biogeochemical dynamics in tidal marsh ecosystems, and to test whether including redox interactions in a land surface model would allow the model to resolve contrasts in biogeochemical cycling and greenhouse gas production between saline and freshwater wetlands.The data package includes gzipped tar archives (which can be expanded using standard tar and gzip utilities) of model outputs from three model configurations: saline subsurface and reduced vegetation productivity related to salinity; saline subsurface with vegetation productivity not reduced; and freshwater. Also included are code for the modified E3SM model, Alquimia interface, and PFLOTRAN reactive transport simulator in gzipped tar format; plain text parameter and configuration files; python code files for visualizing model output and defining model configurations; and model output, tide and salinity forcing, and configuration files in netCDF format. See the README.md file in the data package for a detailed description of all files contained in the package. All files are in netCDF (.nc), gzipped tar archive (.tar.gz or .tgz), or text (all other files).Updated: May 13, 2024. Model output, E3SM code, PFLOTRAN input files, and python codes for visualizing results were updated to reflect changes made for the manuscript revision. The updated archive reflects the code and model output from the final accepted manuscript. Changes included updated reaction parameters reflecting improved parameterization and additional comparisons with field measurements. E3SM code changes included better support for multiple grid cells and improved flow and transport parameterization.

54 ENVIRONMENTAL SCIENCES↗
NASA NTRS1988

Non-seasonal changes in total column ozone from satellite observations, 1970-86

Analyses of nonseasonal changes in total column ozone from satellite observations during the period 1970-86 suggest, when combined with analyses of Dobson network data, that the largest reduction in global ozone since 1959 occurred between 1978 and 1986. The existence of an Arctic region of enhanced ozone depletion with similar but less pronounced characteristics of the Antarctic region is described. Before 1983 large rates of decrease were confined to high latitudes, subsequently regions of large rates of ozone depletion have appeared at midlatitudes.

Heath, Donald F.↗
NASA NTRS2013

SHAO Analysis Center 2012 Annual Report

The Shanghai Astronomical Observatory (SHAO) Analysis Center in 2012 continued routine VLBI data analysis and produced earth orientation parameter (EOP), terrestrial reference frame (TRF), and celestial reference frame (CRF) information, which was submitted to the IVS quarterly. The activities of SHAO also consisted of data reduction of the Chinese VLBI Network (CVN), spacecraft navigation using the VLBI technique, and some research activities.

Wang, Guangli↗
DOE OSTI2019

MicroBooNE's Search for Photon-Like Low Energy Excess Under a Neutral Current $\Delta \to N_{\gamma}$ Hypothesis

MicroBooNE is a Liquid Argon Time Projection Chamber which has been taking neutrino data at Fermilab’s Booster Neutrino Beamline since October 2015. One of its primary goals is to investigate the Low Energy Excess of neutrino events observed by the MiniBooNE experiment, for which one candidate interpretation is neutrino neutral current (NC) resonant ∆ production with subsequent radiative decay. This is a standard model (SM) source of low energy single-photon events that has never been directly measured in neutrino scattering. If found to have a higher than SM-predicted rate it could be a sizable contribution to the Low Energy Excess. In MicroBooNE it would be identified with an event signature of one detached photon shower and some number of proton tracks associated with a reconstructed neutrino vertex. This poster will describe the analysis developed to search for NC ∆ radiative events in MicroBooNE, consisting of a Boosted Decision Tree based event selection wit h enhanced background reduction using a Semantic Segmentation Network to target the dominant NC π0 background.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
NASA NTRS1997

Simplifying silicon burning: Application of quasi-equilibrium to (alpha) network nucleosynthesis

While the need for accurate calculation of nucleosynthesis and the resulting rate of thermonuclear energy release within hydrodynamic models of stars and supernovae is clear, the computational expense of these nucleosynthesis calculations often force a compromise in accuracy to reduce the computational cost. To redress this trade-off of accuracy for speed, the authors present an improved nuclear network which takes advantage of quasi- equilibrium in order to reduce the number of independent nuclei, and hence the computational cost of nucleosynthesis, without significant reduction in accuracy. In this paper they will discuss the first application of this method, the further reduction in size of the minimal alpha network. The resultant QSE- reduced alpha network is twice as fast as the conventional alpha network it replaces and requires the tracking of half as many abundance variables, while accurately estimating the rate of energy generation. Such reduction in cost is particularly necessary for future generation of multi-dimensional models for supernovae.

Hix, W. R.↗