Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “network robustness”

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 631 records · Page 35

Effect of network connectivity on behavior of synthetic Broborg hillfort glasses

There is wide industrial interest in developing robust models of long-term (>100 years) glass durability. Archeological glass analogs, glasses of similar composition, and alteration conditions to those being tested for durability can be used to evaluate and inform such models. Two such analog glasses from a 1500-year-old vitrified hillfort near Uppsala, Sweden have previously been identified as potential analogs for low concentration Fe-bearing aluminosilicate nuclear waste glasses. However, open questions remain regarding the melting environment from which these historic glasses were formed and the effect of these conditions on their chemical durability. A key factor to answering the previous melting and durability questions is the redox state of Fe in the starting and final materials. Past work has shown that the melting conditions of a glass-forming melt may influence the redox ratio value (Fe +3 /ΣFe), a measure of a glass's redox state, and both melting conditions and the redox ratio may influence the glass alteration behavior. Synthetic analogs of the hillfort glasses have been produced using either fully oxidized or reduced Fe precursors to address this question. In this study, the melting behavior, glass transition temperature, oxidation state, network structure, and chemical durability of these synthesized glass analogs is presented. Resulting data suggests that the degree of network connectivity as impacted by the oxidation state of iron impacted the behavior of the glass-forming melt but in this case does not affect the chemical durability of the final glass. Glasses with a lower degree of melt connectivity were found to have a lower viscosity, resulting in a lower glass transition temperature and softening temperature, as well as in a lower temperature of foam onset and temperature of foam maximum. This lower degree of network connectivity most likely played a more significant role in accelerating the conversion of batch chemicals into glass than the presence of water vapor in the furnace's atmosphere. Future work will focus on using the results from this work with outcomes from other aspects of this project to evaluate long-term glass alteration models.

36 MATERIALS SCIENCE↗

Effective, Robust Design of Community Mitigation for Pandemic Influenza: A Systematic Examination of Proposed US Guidance

Background: The US government proposes pandemic influenza mitigation guidance that includes isolation and antiviral treatment of ill persons, voluntary household member quarantine and antiviral prophylaxis, social distancing of individuals, school closure, reduction of contacts at work, and prioritized vaccination. Is this the best strategy combination? Is choice of this strategy robust to pandemic uncertainties? What are critical enablers of community resilience? Methods and Findings: We systematically simulate a broad range of pandemic scenarios and mitigation strategies using a networked, agent-based model of a community of explicit, multiply-overlapping social contact networks. We evaluate illness and societal burden for alterations in social networks, illness parameters, or intervention implementation. For a 1918- like pandemic, the best strategy minimizes illness to ,1% of the population and combines network-based (e.g. school closure, social distancing of all with adults’ contacts at work reduced), and case-based measures (e.g. antiviral treatment of the ill and prophylaxis of household members). We find choice of this best strategy robust to removal of enhanced transmission by the young, additional complexity in contact networks, and altered influenza natural history including extended viral shedding. Administration of age-group or randomly targeted 50% effective pre-pandemic vaccine with 7% population coverage (current US H5N1 vaccine stockpile) had minimal effect on outcomes. In order, mitigation success depends on rapid strategy implementation, high compliance, regional mitigation, and rigorous rescinding criteria; these are the critical enablers for community resilience. Conclusions: Systematic evaluation of feasible, recommended pandemic influenza interventions generally confirms the US community mitigation guidance yields best strategy choices for pandemic planning that are robust to a wide range of uncertainty. The best strategy combines network- and case-based interventions; network-based interventions are paramount. Because strategies must be applied rapidly, regionally, and stringently for greatest benefit, preparation and public education is required for long-lasting, high community compliance during a pandemic.

60 APPLIED LIFE SCIENCES↗

Uncertainty quantification of the convolutional neural networks on permeability estimation from micro-CT scanned sandstone and carbonate rock images

Rock permeability is one of the most crucial properties affecting subsurface fluid flow behaviors. To accurately and robustly estimate the permeability, Digital Rock Physics, including micro-CT scanning technology and direct flow simulations on scanned images, has prevailed in recent years. Besides, machine learning techniques such as convolutional neural networks (CNNs) have been widely adopted and achieved success in permeability estimations directly from rock images. However, existing ML methods used for permeability estimation from rock images lack uncertainty quantification that causes unreliable predictions and overconfident estimations on out-of-distribution (OOD) samples. Here, in this work, we propose a PI3NN-CNN framework to address this problem. PI3NN-CNN consists of a CNN model for absolute permeability estimation and a PI3NN method to quantify the estimation uncertainty. It is able to quantify the uncertainty for in-distribution (InD) data with a desired confidence level, and identify OOD samples to avoid overconfident predictions. We demonstrate the method using micro-CT scanned images from two sandstone and two carbonate rocks. We found that PI3NN-CNN generates accurate predictions for InD samples, while producing high-quality prediction uncertainties regardless of the prediction accuracy. Meanwhile, PI3NN-CNN identifies OOD samples using its special network initialization scheme. The unique feature of PI3NN-CNN makes it applicable to more complex real-world image-based data for robust learning and predictions without overconfident estimations when the ground-truth information is unavailable.

58 GEOSCIENCES↗

In-Space Networking On NASA's SCaN Testbed

The NASA Space Communications and Navigation (SCaN) Testbed, an external payload onboard the International Space Station, is equipped with three software defined radios (SDRs) and a programmable flight computer. The purpose of the Testbed is to conduct inspace research in the areas of communication, navigation, and networking in support of NASA missions and communication infrastructure. Multiple reprogrammable elements in the end to end system, along with several communication paths and a semi-operational environment, provides a unique opportunity to explore networking concepts and protocols envisioned for the future Solar System Internet (SSI). This paper will provide a general description of the system's design and the networking protocols implemented and characterized on the testbed, including Encapsulation, IP over CCSDS, and Delay-Tolerant Networking (DTN). Due to the research nature of the implementation, flexibility and robustness are considered in the design to enable expansion for future adaptive and cognitive techniques. Following a detailed design discussion, lessons learned and suggestions for future missions and communication infrastructure elements will be provided. Plans for the evolving research on SCaN Testbed as it moves towards a more adaptive, autonomous system will be discussed.

space networks↗

MFNets: Multifidelity data-driven networks for Bayesian learning and prediction

This paper presents a multifidelity uncertainty quantification framework called MFNets. We seek to address three existing challenges that arise when experimental and simulation data from different sources are used to enhance statistical estimation and prediction with quantified uncertainty. Specifically, we demonstrate that MFNets can (1) fuse heterogeneous data sources arising from simulations with different parameterizations, e.g simulation models with different uncertain parameters or data sets collected under different environmental conditions; (2) encode known relationships among data sources to reduce data requirements; and (3) improve the robustness of existing multi-fidelity approaches to corrupted data. MFNets construct a network of latent variables (LVs) to facilitate the fusion of data from an ensemble of sources of varying credibility and cost. These LVs are posited as explanatory variables that provide the source of correlation in the observed data. Furthermore, MFNets provide a way to encode prior physical knowledge to enable efficient estimation of statistics and/or construction of surrogates via conditional independence relations on the LVs. We highlight the utility of our framework with a number of theoretical results which assess the quality of the posterior mean as a frequentist estimator and compare it to standard sampling approaches that use single fidelity, multilevel, and control variate Monte Carlo estimators. We also use the proposed framework to derive the Monte Carlo-based control variate estimator entirely from the use of Bayes rule and linear-Gaussian models -- to our knowledge the first such derivation. Finally, we demonstrate the ability to work with different uncertain parameters across different models.

97 MATHEMATICS AND COMPUTING↗

Reassessing long-standing meteorological records: an example using the national hottest day in Ireland

This analysis highlights the potential value in reanalysing early national meteorological records from around the world. These were oftentimes measured via techniques that preceded standardisation of instrumentation and methods of observation and thus could be subject to considerable biases and uncertainties. This analysis uses the techniques pioneered by WMO record assessment teams. The highest currently recognised air temperature (33.3 °C) ever recorded in the Republic of Ireland was logged at Kilkenny Castle in 1887. The original observational record however no longer exists. Given that Ireland is now the only country in Europe to have a national heat record that was set in the 19th century, a reassessment of the verity of this record is both timely and valuable. The present analysis undertakes a fundamental reassessment of the plausibility of the 1887 temperature record using methods similar to those used to assess various weather extremes under WMO auspices over recent years. Specifically, we undertake an inter-station reassessment using sparse available records and make recourse to the new and improved 20CRv3 sparse-input reanalysis product. Neither surrounding available stations nor the reanalysis offer substantive support for the Kilkenny record of 33.3 °C being correct. Moreover, recent data rescue efforts have uncovered several earlier extreme values, one of which exceeds the Kilkenny value (33.5 °C on 16 July 1876 recorded at the Phoenix Park). However, the sparsity of early observational networks, a distinct lack of synoptic support from 20CRv3 for many of the extreme heat values, and the fact that these measurements were obtained using non-standard exposures lead us to conclude that there is grossly insufficient evidence to support any of these 19th century extremes as robust national heat record candidates. Data from the early 20th century onwards benefit from a denser network of stations undertaking measurements in a more standardised manner, many under the direct auspices of Met Éireann and its predecessors, adhering to WMO guidance and protocols. This enables more robust cross-checking of records. We argue that the Met Éireann-recognised 20th century heat record from Boora in 1976 is verified as the most plausible robust national temperature record based upon the synoptic situation and comparisons with nearby neighbouring stations. This measurement of 32.5°C thus likely constitutes the highest reliably recorded temperature measurement in the Republic of Ireland. Ultimately, the formal decision on any reassessment and reassignment of the national record rests with the national meteorological service, Met Éireann.

15 GEOTHERMAL ENERGY↗

Renewable Energy Positioning System: Energy Positioning: Control and Economics

GENI Project: The University of Washington and the University of Michigan are developing an integrated system to match well-positioned energy storage facilities with precise control technologies so the electric grid can more easily include energy from renewable power sources like wind and solar. Because renewable energy sources provide intermittent power, it is difficult for the grid to efficiently allocate those resources without developing solutions to store their energy for later use. The two universities are working with utilities, regulators, and the private sector to position renewable energy storage facilities in locations that optimize their ability to provide and transmit electricity where and when it is needed most. Expanding the network of transmission lines is prohibitively expensive, so combining well-placed storage facilities with robust control systems to efficiently route their power will save consumers money and enable the widespread use of safe, renewable sources of power.

ARPA-E↗

Fracture, failure and compression behaviour of a 3D interconnected carbon aerogel (Aerographite) epoxy composite

Aerographite (AG) is a mechanically robust, lightweight synthetic cellular material, which consists of a 3D interconnected network of tubular carbon [1]. The presence of open channels in AG aids to infiltrate them with polymer matrices, thereby yielding an electrical conducting and lightweight composite. Aerographite produced with densities in the range of 7–15 mg/cm 3 was infiltrated with a low viscous epoxy resin by means of vacuum infiltration technique. Detailed morphological and structural investigations on synthesized AG and AG/epoxy composite were performed by scanning electron microscopic techniques. Our present study investigates the fracture and failure of AG/epoxy composites and its energy absorption capacity under compression. The composites displayed an extended plateau region when uni-axially compressed, which led to an increase in energy absorption of ~133% per unit volume for 1.5 wt% of AG, when compared to pure epoxy. Preliminary results on fracture toughness showed an enhancement of ~19% in K IC for AG/epoxy composites with 0.45 wt% of AG. Furthermore, our observations of fractured surfaces under scanning electron microscope gives evidence of pull-out of arms of AG tetrapod, interface and inter-graphite failure as the dominating mechanism for the toughness improvement in these composites. These observations were consistent with the results obtained from photoelasticity experiments on a thin film AG/epoxy model composite.

36 MATERIALS SCIENCE↗

Data-driven deconvolution for large eddy simulations of Kraichnan turbulence

In this article, we demonstrate the use of artificial neural networks as optimal maps which are utilized for convolution and deconvolution of coarse-grained fields to account for sub-grid scale turbulence effects. Here, we demonstrate that an effective eddy-viscosity is predicted by our purely data-driven large eddy simulation framework without explicit utilization of phenomenological arguments. In addition, our data-driven framework precludes the knowledge of true sub-grid stress information during the training phase due to its focus on estimating an effective filter and its inverse so that grid-resolved variables may be related to direct numerical simulation data statistically. The proposed predictive framework is also combined with a statistical truncation mechanism for ensuring numerical realizability in an explicit formulation. Through this, we seek to unite structural and functional modeling strategies for modeling non-linear partial differential equations using reduced degrees of freedom. Both a priori and a posteriori results are shown for a two-dimensional decaying turbulence case in addition to a detailed description of validation and testing. A hyperparameter sensitivity study also shows that the proposed dual network framework simplifies learning complexity and is viable with exceedingly simple network architectures. Our findings indicate that the proposed framework approximates a robust and stable sub-grid closure which compares favorably to the Smagorinsky and Leith hypotheses for capturing the theoretical k -3 scaling in Kraichnan turbulence.

42 ENGINEERING↗

DeepGrid: Robust Deep Reinforcement Learning-based Contingency Management

Increasing uncertainty raised by the integration of renewable energy resources requires an enormous number of simulations to be carried out for the security assessment of the power grid. However, it is challenging to assess the steady-state and dynamic security indices for different system contingency events by doing an exhaustive analysis in real-time due to the computational and communication constraints. One promising solution is using data-driven techniques along with the system models to train an intelligent contingency management framework to better handle the contingencies in real-time. Nevertheless, implementing a data-driven technique to obtain the best remedial actions necessitates to account for the effect of the measurement noise on the performance of the contingency management. To tackle these challenges, we leverage a robust deep reinforcement learning (DRL) algorithm called Double Deep Q-Network (DDQN) to design a recommender system capable of prescribing optimal control actions with the help of the real-time digital simulator (RTDS). The use of RTDS system in combination with the advanced DRL algorithm allows to explore a wide variety of system contingencies in order to derive better remedial actions. The performance of the proposed algorithm is evaluated in IEEE 9-bus system under different loading conditions, and different network configurations in presence of noisy measurements.

Ghasemkhani, Amir↗

Systematic comparison of lncRNAs with protein coding mRNAs in population expression and their response to environmental change

Background: Long non-coding RNA (lncRNA) is a class of non-coding RNA with important regulatory roles in biological process of organisms. The systematic comparison of lncRNAs with protein coding mRNAs in population expression and their response to environmental change are still poorly understood. Here we identified 17,610 lncRNAs and calculated their expression levels based on RNA-seq of 80 individuals of Miscanthus lutarioriparius from two environments, the nearly native habitats and transplanted field, respectively. Results: LncRNAs had significantly higher expression diversity and lower expression frequency in population than protein coding mRNAs in both environments, which suggested that lncRNAs may experience more relaxed selection or divergent evolution in population compared with protein coding RNAs. In addition, the increase of expression diversity for lncRNAs was always significantly higher and the magnitude of fold change of expression in new stress environment was significantly larger than protein-coding mRNAs. These results suggested that lncRNAs may be more sensitive to environmental change than protein-coding mRNAs. Analysis of environment-robust and environment-specific lncRNA-mRNA co-expression network between two environments revealed the characterization of lncRNAs in response to environmental change. Furthermore, candidate lncRNAs contributing to water use efficiency (WUE) identified based on the WUE-lncRNA-mRNA co-expression network suggested the roles of lncRNAs in response to environmental change. Conclusion: Our study provided a comprehensive understanding of expression characterization of lncRNAs in population for M. lutarioriparius under field condition, which would be useful to explore the roles of lncRNAs and could accelerate the process of adaptation in new environment for many plants.

Expression diversity↗

Irreversibility of Image Transform using Feature Descriptors

Our work in radiographic image matching has centered on the use of SURF (Speeded Up Robust Features) for feature detection, and FLANN (Fast Learning Artificial Neural Network) for feature matching. We discovered that while the SURF process does return information on location, scale, and rotation for each detected feature, they are not essential for image matching. The nature of the remaining feature detection data does not appear to contain any useful information in terms of reconstructing a useful portion of an image, and therefore is not amenable to reconstructing the original image. This led us to wonder if, in fact, we had discovered an irreversible process; the original image could not be constructed from the remaining feature data. Additional detail on the derivation of the image processing and matching algorithms and the irreversibility hypothesis are available in the final SAND Report documenting our previous LDRD work (SAND2015-9665 “Processing Radiation Images Behind an Information Barrier for Automatic Warhead Authentication” Little, Wilson, Weber and Novick, 2015).

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Electric Utility Industry Standards Landscape

The electric utility industry relies on robust communication protocols to manage complex electrical grid data. The inherent networked nature of electrical grids, coupled with the radial structure of the “last mile” portion delivering power to end-use customers, presents difficulties in describing electrical models using simple data constructs. The paper provides an overview of communication protocols that address electric utility data, including grid data, and classifies these protocols through identification key characteristics that make them suitable for different electric grid data domains. Because of their variety, grid edge devices and their associated dedicated protocols are assessed by groups: those primarily designed for energy production and storage, those related to flexible loads, and those related to electric vehicles. The intent of this report is to provide guidance for stakeholders to navigate the challenges posed by the numerous overlapping protocols available to address electric grid data. Although further industry review, refinement, and validation of the categorization of these protocols is recommended, the authors propose this categorization as a start to improve electric grid awareness and understanding. In addition, this report includes three recommended industry actions regarding protocols to improve communications on the electric grid.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

Measuring the Electronic Bandgap of Carbon Nanotube Networks in Non-Ideal p-n Diodes

The measurement of the electronic bandgap and exciton binding energy in quasi-one-dimensional materials such as carbon nanotubes is challenging due to many-body effects and strong electron–electron interactions. Unlike bulk semiconductors, where the electronic bandgap is well known, the optical resonance in low-dimensional semiconductors is dominated by excitons, making their electronic bandgap more difficult to measure. In this work, we measure the electronic bandgap of networks of polymer-wrapped semiconducting single-walled carbon nanotubes (s-SWCNTs) using non-ideal p-n diodes. We show that our s-SWCNT networks have a short minority carrier lifetime due to the presence of interface trap states, making the diodes non-ideal. We use the generation and recombination leakage currents from these non-ideal diodes to measure the electronic bandgap and excitonic levels of different polymer-wrapped s-SWCNTs with varying diameters: arc discharge (~1.55 nm), (7,5) (0.83 nm), and (6,5) (0.76 nm). Our values are consistent with theoretical predictions, providing insight into the fundamental properties of networks of s-SWCNTs. The techniques outlined here demonstrate a robust strategy that can be applied to measuring the electronic bandgaps and exciton binding energies of a broad variety of nanoscale and quantum-confined semiconductors, including the most modern nanoscale transistors that rely on nanowire geometries.

36 MATERIALS SCIENCE↗

Machine Learning for Conservative-to-Primitive in Relativistic Hydrodynamics

The numerical solution of relativistic hydrodynamics equations in conservative form requires root-finding algorithms that invert the conservative-to-primitive variables map. These algorithms employ the equation of state of the fluid and can be computationally demanding for applications involving sophisticated microphysics models, such as those required to calculate accurate gravitational wave signals in numerical relativity simulations of binary neutron stars. This work explores the use of machine learning methods to speed up the recovery of primitives in relativistic hydrodynamics. Artificial neural networks are trained to replace either the interpolations of a tabulated equation of state or directly the conservative-to-primitive map. The application of these neural networks to simple benchmark problems shows that both approaches improve over traditional root finders with tabular equation-of-state and multi-dimensional interpolations. In particular, the neural networks for the conservative-to-primitive map accelerate the variable recovery by more than an order of magnitude over standard methods while maintaining accuracy. Neural networks are thus an interesting option to improve the speed and robustness of relativistic hydrodynamics algorithms.

79 ASTRONOMY AND ASTROPHYSICS↗

Creating a Training Dataset for Semantic Segmentation of Canal Networks for Irrigation Modernization

Canal infrastructure has provided critical irrigation water to the western United States for over a century. To continue providing vital water resources to the semi-arid West, irrigation systems must undergo maintenance and modernization. Many canal companies are resource-constrained, and because funding opportunities often require detailed knowledge of existing infrastructure, they can struggle to secure financial capital. We address this problem by creating training data for a semantic segmentation deep learning model to map canal networks throughout the western United States. To create a diverse and robust training dataset, we labelled 1-m NAIP imagery with the locations of no canals, wet canals, and dry/vegetated canals. Since creating these datasets is time consuming, we first developed a preprocessing methodology to identify canals within our four study areas. We used NAIP imagery and provided canal centerline data to buffer, standardize, and cluster the imagery, automating the labeling process as much as possible. However, this still required manual cleaning and manual classification of canal type. Challenges arose when canals were interrupted (e.g., road culverts or piped sections) or when nearby features shared similar characteristics (e.g., irrigated fields, trees, and shadows). Combining automated preprocessing with manual refinement produced four detailed canal masks to be used in the semantic segmentation model developed by Richard Tapia.

13 - HYDRO ENERGY↗

Demonstration of Decentralized Physics-Driven Learning

In typical artificial neural networks, neurons adjust according to global calculations of a central processor, but in the brain, neurons and synapses self-adjust based on local information. Contrastive learning algorithms have recently been proposed to train physical systems, such as fluidic, mechanical, or electrical networks, to perform machine-learning tasks from local evolution rules. However, to date, such systems have only been implemented in silico due to the engineering challenge of creating elements that autonomously evolve based on their own response to two sets of global boundary conditions. Here, we introduce and implement a physics-driven contrastive learning scheme for a network of variable resistors, using circuitry to locally compare the response of two identical networks subjected to the two different sets of boundary conditions. Using this method, our system effectively trains itself, optimizing its resistance values without the use of a central processor or external information storage. Once the system is trained for a specified allostery, regression, or classification task, the task is subsequently performed rapidly and automatically by the physical imperative to minimize power dissipation in response to the given voltage inputs. Here, we demonstrate that, unlike typical computers, such learning systems are robust to extreme damage (and thus manufacturing defects) due to their decentralized learning. Our twin-network approach is therefore readily scalable to extremely large or nonlinear networks, where its distributed nature will be an enormous advantage; a laboratory network of only 500 edges will already outpace its in silico counterpart.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Classifying handedness in chiral nanomaterials using label error robust deep learning

Abstract High-throughput scanning electron microscopy (SEM) coupled with classification using neural networks is an ideal method to determine the morphological handedness of large populations of chiral nanoparticles. Automated labeling removes the time-consuming manual labeling of training data, but introduces label error, and subsequently classification error in the trained neural network. Here, we evaluate methods to minimize classification error when training from automated labels of SEM datasets of chiral Tellurium nanoparticles. Using the mirror relationship between images of opposite handed particles, we artificially create populations of varying label error. We analyze the impact of label error rate and training method on the classification error of neural networks on an ideal dataset and on a practical dataset. Of the three training methods considered, we find that a pretraining approach yields the most accurate results across label error rates on ideal datasets, where size and other morphological variables are held constant, but that a co-teaching approach performs the best in practical application.

36 MATERIALS SCIENCE↗