SEARCH · Engineering Papers

Results for “network performance”

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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 325 records · Page 18

All sources NASA NTRS DOE OSTI

DOE OSTI2024

GraMeR: Gra ph Me ta R einforcement learning for multi-objective influence maximization

Influence maximization (IM) is a combinatorial problem of identifying a subset of seed nodes in a network (graph), which when activated, provide a maximal spread of influence in the network for a given diffusion model and a budget for seed set size. IM has numerous applications such as viral marketing, epidemic control, sensor placement and other network-related tasks. However, its practical uses are limited due to the computational complexity of current algorithms. Recently, deep reinforcement learning has been leveraged to solve IM in order to ease the computational burden. However, there are serious limitations in current approaches, including narrow IM formulation that only consider influence via spread and ignore self-activation, low scalability to large graphs, and lack of generalizability across graph families leading to a large running time for every test network. In this work, we address these limitations through a unique approach that involves: (1) Formulating a generic IM problem as a Markov decision process that handles both intrinsic and influence activations; (2)incorporating generalizability via meta-learning across graph families. There are previous works that combine deep reinforcement learning with graph neural network, but this work solves a more realistic IM problem and incorporates generalizability across graphs via meta reinforcement learning. Extensive experiments are carried out in various standard networks to validate performance of the proposed Graph Meta Reinforcement learning (GraMeR) framework. Finally, the results indicate that GraMeR is multiple orders faster and generic than conventional approaches when applied on small to medium scale graphs.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

Harnessing Neural Networks for Elucidating X-ray Absorption Structure–Spectrum Relationships in Amorphous Carbon

Improved understanding of structural and chemical properties through local experimental probes, such as X-ray absorption near-edge structure (XANES) spectroscopy, is crucial for the understanding and design of functional materials. In recent years, significant advancements have been made in the development of data science approaches for the automated interpretation of XANES structure–spectrum relationships. However, existing studies have primarily focused on crystalline solids and small molecules, while fewer efforts have been devoted to disordered systems. Thus, in this work, we demonstrate the development of neural network models for predicting and interpreting XANES spectra of amorphous carbon (a-C) from local structural descriptors. Comparison between different structural descriptors expectedly shows that the inclusion of both bond length and bond angle information is necessary for an accurate prediction of the spectra. Among the descriptors considered in this work, we find that the local many-body tensor representation yields the highest accuracy and greatest interpretability so that it can be leveraged to understand the importance of structural motifs in determining XANES spectra. Furthermore, we also discuss performance of neural network models for predicting both local structure features, such as bond lengths and bond angles, and global chemical composition, such as the sp:sp 2 :sp 3 ratio.

36 MATERIALS SCIENCE↗

DOE OSTI2025

Accelerating Hamiltonian Monte Carlo for Bayesian inference in neural networks and neural operators

Hamiltonian Monte Carlo (HMC) is a powerful and accurate method to sample from the posterior distribution in Bayesian inference. However, HMC techniques are computationally demanding for Bayesian neural networks due to the high dimensionality of the network’s parameter space and the non-convexity of their posterior distributions. Therefore, various approximation techniques, such as variational inference (VI) or stochastic gradient MCMC, are often employed to infer the posterior distribution of the network parameters. Such approximations introduce inaccuracies in the inferred distributions, resulting in unreliable uncertainty estimates. In this work, we propose a hybrid approach that combines inexpensive VI and accurate HMC methods to efficiently and accurately quantify uncertainties in neural networks and neural operators. The proposed approach leverages an initial VI training on the full network. We examine the influence of individual parameters on the prediction uncertainty, which shows that a large proportion of the parameters do not contribute substantially to uncertainty in the network predictions. This information is then used to significantly reduce the dimension of the parameter space, and HMC is performed only for the subset of network parameters that strongly influence prediction uncertainties. This yields a framework for accelerating the full batch HMC for posterior inference in neural networks. We demonstrate the efficiency and accuracy of the proposed framework on deep neural networks and operator networks, showing that inference can be performed for large networks with tens to hundreds of thousands of parameters. Finally, we show that this method can effectively learn surrogates for complex physical systems by modeling the operator that maps from upstream conditions to wall-pressure data on a cone in hypersonic flow.

Bayesian inference↗

DOE OSTI2025

Performance Evaluation of Vertical Federated Machine Learning Against Adversarial Threats on Wide-Area Control System: Preprint

Federated machine learning (FL) is gaining significant popularity to develop cybersecurity solutions in power grids because of its advanced capability to support decentralized data handing at local devices, its privacy preservation, and its low-bandwidth requirement. However, the evolving adversarial machine learning (AML) threats raise significant concerns for the cybersecurity of FL architectures. The FL-based split neural network (SplitNN) achieves high performance through the decentralized training of local neural network models while preserving data privacy across multiple entities. In this paper, we propose a methodology for evaluating the performance of a vertical FLbased anomaly detector against different types of AML attacks, including denial-of-service attacks, adversarial data injection attacks, and replay attacks on the trained local models deployed in the grid network. For a case study, we consider the modified IEEE 13-bus system, and we develop SplitNN-based binary and multiclass classification models to detect, locate, and identify different types of data integrity attacks on the volt-watt control with two pooling layers: maximum pooling and AvgPool. Our experimental results, computed through performance metrics, reveal that the severity of these AML attacks varies with the integrated pooling mechanism, the type of classification model, and the nature of the cyberattack. Further, the AML attacks negatively impacted the prediction time per sample for the pretrained SplitNN during the online testing.

adversarial threats↗

DOE OSTI2025

Dendritic Computing with Multigate Ferroelectric Field-Effect Transistors

Although inspired by neuronal systems in the brain, artificial neural networks generally employ point-neurons, which offer computational complexity far less than that of their biological counterparts. Neurons have dendritic arbors that connect to different sets of synapses and offer local nonlinear accumulation – playing a pivotal role in processing and learning. Inspired by this, we propose a novel neuron design based on a multigate ferroelectric field-effect transistor that mimics dendrites. It leverages ferroelectric nonlinearity for local computations within dendritic branches while utilizing the transistor action to generate the neuronal output. The branched architecture enables smaller crossbar arrays in hardware integration, improving efficiency. Using an experimentally calibrated device-circuit-algorithm cosimulation framework, we demonstrate that networks incorporating our dendritic neurons achieve superior performance compared to much larger networks without dendrites (∼ 17× fewer trainable weight parameters). These findings suggest that dendritic hardware can significantly improve computational efficiency and learning capacity of neuromorphic systems optimized for edge applications.

36 MATERIALS SCIENCE↗

DOE OSTI2023

Multiverse: Bayesian model selection for neural networks

Multiverse is a code repository for a set of tools for Bayesian model selection for neural networks. The goal of the tools is to provide capabilities for selecting among prior and model specifications for Bayesian neural networks. Multiverse will include tools for creating Bayesian neural networks, evaluating the Bayesian model evidence, and performing inference in Bayesian neural networks. These components are written in Python, a high-level programming language that takes advantage of the Python ecosystem of high-quality open-source packages for machine learning and signal processing.

Klein, Natalie↗

DOE OSTI2022

DistGANS- Distributed Generative Adversarial Neural Networks

DistGANs is a Python package to perform distributed training of conditional generative adversarial neural networks for multi-class labeled image data. DistGANs partitionins the training data according to data labels, and enhances scalability by performing a parallel training where multiple generators are concurrently trained, each one of them focusing on a single data label.

Lupo Pasini, Massimiliano [Oak Ridge National Lab.↗

DOE OSTI2024

The Building Business Network (B-Biz): Addressing Gaps in the High-Performance Building Technology Market, Especially for Underserved Customers

Increased market intelligence and business model innovation is needed to build contractor confidence in high-performance building technologies in order to increase the speed and scale of adoption to meet the U.S. Department of Energy's building stock decarbonization goals. Although there have been important innovations in energy efficiency, affordability, and decarbonization of building technologies, consumers are not purchasing these technologies at the necessary speed and scale partially due to a gap in institutional understanding around barriers contractors face in providing and servicing these technologies. Small businesses in this market must mitigate risk around high-performance building technologies, limiting their opportunities in the market and impacting the rate of adoption. The Building Business Network (B-Biz) aims to provide high-performance building technology solutions to underserved customers by collaborating with local small businesses that provide and service high-performance building technologies in communities with the lowest rates of adoption. This research explores the current market, the importance of business models, and the opportunity to utilize small businesses to address market gaps in underserved communities.

B-Biz↗

DOE OSTI2020

Evolving Ensembles of Spiking Neural Networks for Neuromorphic Systems

Evolutionary algorithms have been proposed as a solution to overcome many of the challenges associated with training spiking neural networks. While evolutionary optimization for spiking neural networks is very flexible, its performance has difficulty scaling to complex tasks and correspondingly complex network structures. Here we propose a method for evolving ensembles of spiking neural networks. By using ensemble learning, the flexibility of evolutionary optimization is fully preserved while scaling to more challenging tasks. We test the performance of the proposed method using handwritten digit classification. We investigate multiple strategies for constructing ensembles of spiking neural networks, and demonstrate that evolving ensembles of SNNs offers significant performance advantages over evolutionary optimization.

Elbrecht, Daniel↗

DOE OSTI2026

Complete Phase Transformation of Ir Nanowire Network into Defect-Rich Oxide Catalyst for High-Performance PEM Water Electrolysis

Iridium-based catalysts remain the most reliable option for the oxygen evolution reaction (OER) in proton exchange membrane water electrolyzers (PEMWEs). However, their high cost and limited performance represent critical barriers to the commercialization of this green hydrogen production technology. Herein, we report the creation of a metallic Ir nanowire network (IrNWN), which exhibits superior OER performance through its in-situ transition into an oxide structure with high intrinsic activity. At a low loading of 0.25 mgIr/cm2 in PEMWEs, IrNWN achieved a current density of 3.13 A/cm2 at a cell voltage of 1.8 V, outperforming the commercial Ir-based catalyst and surpassing the Department of Energy (DOE) 2026 technical target. Moreover, the high activity of IrNWN was maintained for 900 hours in a durability test at 2 A/cm2, showing a low degradation rate of 0.042 mV/hour. Structural analysis of the electrochemically oxidized IrNWN revealed the presence of mixed Ir oxidation states and a high density of surface terminal oxygen groups (μ1-O), which contributed to a reduced energy barrier for the rate-determining O-O coupling step.

oxygen evolution reaction↗

DOE OSTI2025

Discrete fracture network model benchmarks developed and applied in a DECOVALEX-2023 repository performance assessment study

This study presents newly developed benchmarks for modeling flow and transport within discrete fracture networks (DFNs) and useful methods for analyzing the results. The new benchmarks are designed to test modeling approaches for use in probabilistic performance assessment models of deep geologic repositories in fractured rock. The benchmarks simulate flow and transport through a 1 km 3 block of fractured rock. The first simulates migration of a short pulse of tracer through a simple network of four intersecting fractures. The second adds 1089 stochastically generated fractures. The third changes the pulse to a continuous point source. Evaluation of model performance relies on moment analysis and comparison of the results of different models. The expected nondimensional first moment of the conservative tracer for each benchmark is 1. The benchmarks were simulated by teams from Canada, Czechia, Germany, Korea, Sweden, Taiwan, and the United States as part of a DECOVALEX-2023 study (decovalex.org). The teams used various approaches, including explicit DFN modeling, DFN upscaling to an equivalent continuous porous medium (ECPM), and a combination of both methods. Transport mechanisms are modeled using either the advection-dispersion equation or particle tracking. Results demonstrate strong agreement among the models in breakthrough behavior up to the 75th percentile. Significant deviations in first moments and well-clustered outputs led to the identification of inaccuracies in several models. Such findings exemplify the benefit of exercising these benchmarks and using the presented methods to test DFN flow and transport models.

Benchmark↗

DOE OSTI2020

Evolutionary history, potential intermediate animal host, and cross-species analyses of SARS-CoV-2

To investigate the evolutionary history of the recent outbreak of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in China, a total of 70 genomes of virus strains from China and elsewhere with sampling dates between 24 December 2019 and 3 February 2020 were analyzed. To explore the potential intermediate animal host of the SARS-CoV-2 virus, we reanalyzed virome data sets from pangolins and representative SARS-related coronaviruses isolates from bats, with particular attention paid to the spike glycoprotein gene. We performed phylogenetic, split network, transmission network, likelihood-mapping, and comparative analyses of the genomes. Additionally, based on Bayesian time-scaled phylogenetic analysis using the tip-dating method, we estimated the time to the most recent common ancestor and evolutionary rate of SARS-CoV-2, which ranged from 22 to 24 November 2019 and 1.19 to 1.31 × 10 -3 substitutions per site per year, respectively. Our results also revealed that the BetaCoV/bat/Yunnan/RaTG13/2013 virus was more similar to the SARS-CoV-2 virus than the coronavirus obtained from the two pangolin samples (SRR10168377 and SRR10168378). We also identified a unique peptide (PRRA) insertion in the human SARS-CoV-2 virus, which may be involved in the proteolytic cleavage of the spike protein by cellular proteases, and thus could impact host range and transmissibility. Interestingly, the coronavirus carried by pangolins did not have the RRAR motif. Therefore, we concluded that the human SARS-CoV-2 virus, which is responsible for the recent outbreak of COVID-19, did not come directly from pangolins.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2023

Investigation of process history and underlying phenomena associated with the synthesis of plutonium oxides using Vector Quantizing Variational Autoencoder

Accurate, high throughput, and unbiased analysis of plutonium oxide particles is needed for analysis of the phenomenology associated with process parameters in their synthesis. Compared to qualitative and taxonomic descriptors, quantitative descriptors of particle morphology through scanning electron microscopy (SEM) have shown success in analyzing process parameters of uranium oxides. Among other candidates, a neural network called a Vector Quantizing Variational Autoencoder (VQ-VAE) has shown the ability to quantitatively describe particle morphology to attain >85% accuracy in identifying uranium oxide processing routes. We utilize a VQ-VAE to quantitatively describe plutonium dioxide (PuO 2 ) particles created in a designed experiment and investigate their phenomenology and prediction of their process parameters. PuO 2 was calcined from Pu(III) oxalates that were precipitated under varying synthetic conditions that related to concentrations, temperature, addition and digestion times, precipitant feed, and strike order; the surface morphology of the resulting PuO 2 powders were analyzed by SEM. A pipeline was developed to extract and quantify useful image representations for individual particles with the VQ-VAE, then further reduce the dimensionality of the feature space using a bottlenecking neural network fit to perform multiple classification tasks simultaneously. The reduced feature space could predict process parameters with greater than 80% accuracies for some parameters with a single particle. They also showed utility for grouping particles with similar surface morphology characteristics together. Both the clustering and classification results reveal valuable information regarding which chemical process parameters chiefly influence the PuO 2 particle morphologies: strike order and oxalic acid feedstock. Doing the same analysis with multiple particles was shown to improve the classification accuracy on each process parameter over the use of a single particle, with statistically significant results generally seen with as few as four particles in a sample.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2022

Coupling of multiscale imaging analysis and computational modeling for understanding thick cathode degradation mechanisms

Here, using a thick NMC811 (LiNi 0.8 Mn 0.1 Co 0.1 O 2 ) electrode as an example, we present a macro- to nanoscale 2D and 3D imaging analysis approach coupled with 4D (space + time) computational modeling to probe its degradation mechanism in a lithium-ion battery cell. Particle cracking increases and contact loss between particles and carbon-binder domain are observed to correlate with the cell degradation. This study unravels that the reaction heterogeneity within the thick cathode caused by the unbalanced electron conduction is the main cause of the battery degradation over cycling. The increased heterogeneity in the system will entail more cathode regions where the degree of active material utilization is uneven, leading to higher probabilities of particle cracking. These findings shed light on the crucial role of the electronic and ionic transportation networks in the performance deterioration of the thick cathode. They also provide guidance for cathode architecture optimization and performance improvement.

25 ENERGY STORAGE↗

DOE OSTI2022

Impact of Concurrent Solubilization and Fines Migration on Fracture Aperture Growth in Shales during Acidized Brine Injection

Acidic hydraulic fracturing fluid chemically and physically alters shale rock fabric during injection and shut-in, creating a “reaction-altered zone” along the fracture faces. To better characterize the variable thickness and composition of this reaction-altered zone under advective flow, we take a coupled experimental and modeling approach. A fluidic cell, with six fiducial markers, is first fabricated to keep the rock sample in place during the core floods and to allow image alignment of acquired images. Then, we conduct a series of reactive core floods in a clay-rich siliceous Wolfcamp shale sample with 10 wt % carbonate, using a synthetic fracturing fluid under no confining stress and at room temperature. High-resolution computed tomography (CT) scans are periodically conducted to observe the spatial alteration of the fracture network. We then perform scanning electron microscopy (SEM-EDS) on the two orthogonal surfaces (fracture surface and freshly cut profile face) to generate high-resolution elemental maps that show the change in mineralogy, both with distance along a given flow path along a fracture surface and with depth from the fracture surface into the shale matrix. These results are contrasted against a two-dimensional (2D) advection-diffusion reaction model developed previously for batch reactions between shale and synthetic fracturing fluids. The model simulates the geochemical interaction occurring at the fracture/matrix interface and penetrating into the shale matrix during the reactive core flood. Both model and experimental results show that the acidic brine is neutralized during the core flood, corresponding to an increase in fracture aperture as a function of fluid volume injected with the greatest change near the inlet. SEM-EDS scans reveal significant dissolution of carbonates on the fracture surface without pyrite oxidation. The reactive transport model indicates that carbonate depletion into the shale interior should be observable, yet SEM-EDS shows no discernible loss of carbonate in the orthogonal profile face. The combination of these observations suggests an additional fracture evolution mechanism in the reactive system, i.e., fines migration. We show that fines migration enhances the access of fracturing fluid to the matrix resulting in a more pronounced fracture widening. Finally, we conclude that coupled mineral dissolution and fines migration govern fracture aperture growth during acidized brine injection. In this work, we effectively show the underlying risk of relying solely on models that do not include an important (transport) process that can alter the system significantly and propose a combined chemomechanical mechanism for fracture evolution appropriate for this shale mineralogy.

58 GEOSCIENCES↗

DOE OSTI2022

On the effect of mixing-driven vaporization in a homogeneous relaxation modeling framework

The homogeneous relaxation model (HRM) is one of the most widely used models to describe the liquid–gas phase transition in multiphase flows due to the occurrence of cavitation. However, in its original formulation, the HRM does not account for the presence of ambient gas species, which generally limits its applicability to the injector's internal flow where ambient gases are negligible. In this work, a mixing-driven vaporization (MDV) model was developed to extend the capability of the HRM in handling the mixing effect in the regions external to the nozzle, where vapor–liquid equilibrium for multi-species mixtures of fuel and ambient gas is considered. Herein, to assess the model performance, simulations of the Engine Combustion Network's Spray G injector were performed with the HRM and the MDV model under both flash-boiling and evaporating conditions. It was found that the MDV model led to a better match against x-ray measurements of fuel density in the near-nozzle region. In contrast to the HRM, the MDV model was able to reproduce the vaporization process in the mixing zone at the edge of the fuel jet, which aligns with the expected physics. This resulted in substantial differences in the prediction of other flow characteristics such as mixture temperature and pressure. Furthermore, this work demonstrates that evaporation timescales have a considerable effect on the MDV model's predictions, as shown by a parametric study in which a time factor was introduced to mimic the effect of different timescales due to different phase change mechanisms.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

DOE OSTI2025

Deep inference of simulated strong lenses in ground-based surveys

The large number of strong lenses discoverable in future astronomical surveys will likely enhance the value of strong gravitational lensing as a cosmic probe of dark energy and dark matter. However, leveraging the increased statistical power of such large samples will require further development of automated lens modeling techniques. We show that deep learning and simulation-based inference (SBI) methods produce informative and reliable estimates of parameter posteriors for strong lensing systems in ground-based surveys. We present the examination and comparison of two approaches to lens parameter estimation for strong galaxy-galaxy lenses — Neural Posterior Estimation (NPE) and Bayesian Neural Networks (BNNs). We perform inference on 1-, 5-, and 12-parameter lens models for ground-based imaging data that mimics the Dark Energy Survey (DES). We find that NPE outperforms BNNs, producing posterior distributions that are more accurate, precise, and well-calibrated for most parameters. For the 12-parameter NPE model, the calibration is consistently within <10% of optimal calibration for all parameters, while the BNN is rarely within 20% of optimal calibration for any of the parameters. Similarly, residuals for most of the parameters are smaller (by up to an order of magnitude) with the NPE model than the BNN model. This work takes important steps in the systematic comparison of methods for different levels of model complexity.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

DOE OSTI2021

Relocating microseismicity from downhole monitoring of the Decatur CCS site using a modified double-difference algorithm

SUMMARY The injection of CO2 at the Decatur carbon capture and storage site has generated significant microseismic activity, which occurs in distinct spatial clusters up to approximately 2.2 km from the primary injection well. Accurate and precise event locations are vital for the characterization of the microseismicity to help understand the reservoir response to the CO2 injection, whilst enabling the identification of minor faults and fractures below the resolution of conventional active seismic imaging. However, microseismic monitoring of fluid injection sites, such as Decatur, is often performed using a network of borehole sensors often from a single well. While these downhole sensors have excellent detection capabilities, their poor azimuthal coverage limits the ability to precisely determine event locations. We have developed a modified double-difference relocation algorithm suitable for both 1-D and 3-D velocity models, and which incorporates differential back azimuth observations to allow the benefits of the original double-difference algorithm to be applicable to a downhole microseismic monitoring setting. Applying the modified double-difference algorithm to the microseismicity at Decatur, we have successfully relocated 4293 events. The relocation included over 59 million observations for 757 285 event pairs, split across seven geographic regions. Despite the majority of observations being recorded in only two boreholes, with an almost identical azimuthal coverage, the results have shown to be reliable with significantly reduced residuals and low uncertainties associated with the final locations. We have analysed the residuals in terms of their association with each geographic region, data type, station and individual events, to fully appreciate their influence in the inversion and the fit of the data to the final set of event locations. For each region, the relocated seismicity has become less diffuse with improved clustering, and with newly visible linear features often orientated in a NE–SW direction. These results show the potential improvements that can be made to microseismic event locations recorded by a borehole network with a limited and variable azimuthal distribution.

58 GEOSCIENCES↗