Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “layered networks”

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 505 records · Page 28
All sourcesNASA NTRSDOE OSTI
NASA NTRS2017

Delay/Disruption Tolerant Networking for the International Space Station (ISS)

Disruption Tolerant Networking (DTN) is an emerging data networking technology designed to abstract the hardware communication layer from the spacecraft/payload computing resources. DTN is specifically designed to operate in environments where link delays and disruptions are common (e.g., space-based networks). The National Aeronautics and Space Administration (NASA) has demonstrated DTN on several missions, such as the Deep Impact Networking (DINET) experiment, the Earth Observing Mission 1 (EO-1) and the Lunar Laser Communication Demonstration (LLCD). To further the maturation of DTN, NASA is implementing DTN protocols on the International Space Station (ISS). This paper explains the architecture of the ISS DTN network, the operational support for the system, the results from integrated ground testing, and the future work for DTN expansion.

Schlesinger, Adam↗
DOE OSTI2021

InversionNet3D: Efficient and Scalable Learning for 3-D Full-Waveform Inversion

Seismic full-waveform inversion (FWI) techniques aim to find a high-resolution subsurface geophysical model provided with waveform data. Some recent effort in data-driven FWI has shown some encouraging results in obtaining 2-D velocity maps. However, due to high computational complexity and large memory consumption, the reconstruction of 3-D high-resolution velocity maps via deep networks is still a great challenge. Here, in this article, we present InversionNet3D (InvNet3D), an efficient and scalable encoder–decoder network for 3-D FWI. The proposed method employs group convolution in the encoder to establish an effective hierarchy for learning information from multiple sources while cutting down unnecessary parameters and operations at the same time. The introduction of invertible layers further reduces the memory consumption of intermediate features during training and, thus, enables the development of deeper networks with more layers and higher capacity as required by different application scenarios. Experiments on the 3-D Kimberlina dataset demonstrate that InvNet3D achieves state-of-the-art reconstruction performance with lower computational cost and lower memory footprint compared to the baseline.

58 GEOSCIENCES↗
DOE OSTI2025

Triangle Method for Dense ReLU Layers [SWR-25-72]

This software is an implementation of the methods for initializing and training neural networks to be more efficient per parameter, described more fully below and in the related publication: In theory, depth should make a ReLU network EXPONENTIALLY more efficient by enabling it to produce an exponential number of piecewise linear sections in its output. This reasoning is largely based on the work of mathematicians that have hand-constructed networks that make good use of depth. In practice however, even very deep ReLU networks that have been randomly initialized will behave identically to their shallow counterparts - missing an entire exponential dimension of efficiency. The triangle method is a first attempt at realizing the exponential potential of deep networks. Instead of randomly setting weights, we force pairs of neurons in each layer learn to build triangles (i.e. functions from [0,1] -> [0,1] that look like triangles). This is a very efficient pattern for generating lots of linear pieces because composing two triangular functions doubles the number of pieces with each composition. The triangle method is more than just a different initialization, it is a new paradigm of training. Instead of making direct updates to the matrix weights, we do an extra step of backpropagation to collect the derivatives of the loss function with respect to the shapes of the triangles, training them to tilt left or right. This process essentially holds the networks hand throughout the loss landscape and forces it to always use depth effectively by producing triangular shapes internally. This can produce several orders of magnitude of improvement on convex one-dimensional regression problems. Much more theoretical work is needed to realize its full potential beyond this context, but the implementation in this repository will still work in arbitrary numbers of dimensions. The file Triangle_Method.py is a generalized form of the method that will build each neuron its own custom 1-d convex activation function (with exponential efficiency). Example usage on one dimensional problems can be found in Example_Usage.ipynb and an example of using this in a real neural network can be found in Example_VGG16_CIFAR10.ipynb.

Milkert, Max [National Renewable Energy Laboratory↗
NASA NTRS2020

Composite Piezoelectric Resonator 1D Modeling with Loss

Abstract – Maximizing the benefit of applying piezoelectric materials in composite structures, which are a mix of piezoelectric and elastic elements, requires analytical modeling. In this paper we describe two main classes of resonators and how to model using network equivalent including material losses. These include extensional devices where the layer areas are parallel to the direction of wave travel as found in a variety of thickness or length mode transducers (BC. Stress T1 = T2 at boundary) and transverse mode resonators (BC. Strain S1 = S2 at boundary) where the major layer area is perpendicular to the wave direction as is found in radial mode or length thickness mode composite resonators. In order to illustrate these models we will use Mason’s equivalent circuits with elastic, dielectric and piezoelectric loss. It is noted that in order to maintain consistency with the linear equations of piezoelectricity and the wave equation care is required when applying complex loss coefficients to the models. Although these techniques are applicable to other models we use Mason’s network equivalent circuit because it is intuitive for composite modeling due to the frequency independent turns ratio and the independence of the equations describing the elastic layers from the adjoining layer properties which aids in understanding the boundary conditions to impose on the layer boundaries. We will present network models for a variety of applications with loss and show how to calculate their impedance spectra, interface stress, strain velocity and displacement spectra.

Sherrit, Stewart↗
NASA NTRS2020

Convolutional Neural Network for Transition Modeling Based on Linear Stability Theory

Transition prediction is an important aspect of aerodynamic design because of its impact on skin friction and potential coupling with flow separation characteristics. Traditionally, the modeling of transition has relied on correlation-based empirical formulas based on integral quantities such as the shape factor of the boundary layer. However, in many applications of computational fluid dynamics, the shape factor is not straightforwardly available or not well-defined. We propose using the complete velocity profile along with other quantities (e.g., frequency, Reynolds number) to predict the perturbation amplification factor. While this can be achieved with regression models based on a classical fully connected neural network, such a model can be computationally more demanding. We propose a novel convolutional neural network inspired by the underlying physics as described by the stability equations. Specifically, convolutional layers are first used to extract integral quantities from the velocity profiles, and then fully connected layers are used to map the extracted integral quantities, along with frequency and Reynolds number, to the output (amplification ratio). Numerical tests on classical boundary layers clearly demonstrate the merits of the proposed method. More importantly, we demonstrate that, for Tollmien-Schlichting instabilities in two-dimensional, low-speed boundary layers, the proposed network encodes information in the boundary layer profiles into an integral quantity that is strongly correlated to a well-known, physically defined parameter – the shape factor.

Laminar-turbulent transition↗
DOE OSTI2021

Hierarchical effects facilitate spreading processes on synthetic and empirical multilayer networks

In this paper we consider the effects of corporate hierarchies on innovation spread across multilayer networks, modeled by an elaborated SIR framework. We show that the addition of management layers can significantly improve spreading processes on both random geometric graphs and empirical corporate networks. Additionally, we show that utilizing a more centralized working relationship network rather than a strict administrative network further increases overall innovation reach. In fact, this more centralized structure in conjunction with management layers is essential to both reaching a plurality of nodes and creating a stable adopted community in the long time horizon. Further, we show that the selection of seed nodes affects the final stability of the adopted community, and while the most influential nodes often produce the highest peak adoption, this is not always the case. In some circumstances, seeding nodes near but not in the highest positions in the graph produces larger peak adoption and more stable long-time adoption.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

AtomSets as a hierarchical transfer learning framework for small and large materials datasets

Abstract Predicting properties from a material’s composition or structure is of great interest for materials design. Deep learning has recently garnered considerable interest in materials predictive tasks with low model errors when dealing with large materials data. However, deep learning models suffer in the small data regime that is common in materials science. Here we develop the AtomSets framework, which utilizes universal compositional and structural descriptors extracted from pre-trained graph network deep learning models with standard multi-layer perceptrons to achieve consistently high model accuracy for both small compositional data (<400) and large structural data (>130,000). The AtomSets models show lower errors than the graph network models at small data limits and other non-deep-learning models at large data limits. They also transfer better in a simulated materials discovery process where the targeted materials have property values out of the training data limits. The models require minimal domain knowledge inputs and are free from feature engineering. The presented AtomSets model framework can potentially accelerate machine learning-assisted materials design and discovery with less data restriction.

Chen, Chi (ORCID:0000000180087043)↗
NASA NTRS2013

Improved Boundary Layer Depth Retrievals from MPLNET

Continuous lidar observations of the planetary boundary layer (PBL) depth have been made at the Micropulse Lidar Network (MPLNET) site in Greenbelt, MD since April 2001. However, because of issues with the operational PBL depth algorithm, the data is not reliable for determining seasonal and diurnal trends. Therefore, an improved PBL depth algorithm has been developed which uses a combination of the wavelet technique and image processing. The new algorithm is less susceptible to contamination by clouds and residual layers, and in general, produces lower PBL depths. A 2010 comparison shows the operational algorithm overestimates the daily mean PBL depth when compared to the improved algorithm (1.85 and 1.07 km, respectively). The improved MPLNET PBL depths are validated using radiosonde comparisons which suggests the algorithm performs well to determine the depth of a fully developed PBL. A comparison with the Goddard Earth Observing System-version 5 (GEOS-5) model suggests that the model may underestimate the maximum daytime PBL depth by 410 m during the spring and summer. The best agreement between MPLNET and GEOS-5 occurred during the fall and they diered the most in the winter.

Micropluse Lidar Network↗
NASA NTRS2010

Development of an e-VLBI Data Transport Software Suite with VDIF

We have developed a software library (KVTP-lib) for VLBI data transmission over the network with the VDIF (VLBI Data Interchange Format), which is the newly proposed standard VLBI data format designed for electronic data transfer over the network. The software package keeps the application layer (VDIF frame) and the transmission layer separate, so that each layer can be developed efficiently. The real-time VLBI data transmission tool sudp-send is an application tool based on the KVTP-lib library. sudp-send captures the VLBI data stream from the VSI-H interface with the K5/VSI PC-board and writes the data to file in standard Linux file format or transmits it to the network using the simple- UDP (SUDP) protocol. Another tool, sudp-recv , receives the data stream from the network and writes the data to file in a specific VLBI format (K5/VSSP, VDIF, or Mark 5B). This software system has been implemented on the Wettzell Tsukuba baseline; evaluation before operational employment is under way.

Sekido, Mamoru↗
DOE OSTI2022

CPES-QSM: A Quantitative Method Towards the Secure Operation of Cyber-Physical Energy Systems

Power systems are evolving into cyber-physical energy systems (CPES) mainly due to the integration of modern communication and Internet-of-Things (IoT) devices. CPES security evaluation is challenging since the physical and cyber layers are often not considered holistically. Existing literature focuses on only optimizing the operation of either the physical or cyber layer while ignoring the interactions between them. This paper proposes a metric, the Cyber-Physical Energy System Quantitative Security Metric (CPES-QSM), that quantifies the interaction between the cyber and physical layers across three domains: electrical, cyber-risk, and network topology. A method for incorporating the proposed cyber-metric into operational decisions is also proposed by formulating a cyber-constrained AC optimal power flow (C-ACOPF) that considers the status of all the CPES layers. The C-ACOPF considers the vulnerabilities of physical and cyber networks by incorporating factors such as voltage stability, contingencies, graph-theory, and IoT cyber risks, while using a multi-criteria decision-making technique. We note that simulation studies are conducted using standard IEEE test systems to evaluate the effectiveness of the proposed metric and the C-ACOPF formulation.

24 POWER TRANSMISSION AND DISTRIBUTION↗
NASA NTRS2001

Nodal endoplasmic reticulum, a specialized form of endoplasmic reticulum found in gravity-sensing root tip columella cells

The endoplasmic reticulum (ER) of columella root cap cells has been postulated to play a role in gravity sensing. We have re-examined the ultrastructure of columella cells in tobacco (Nicotiana tabacum) root tips preserved by high-pressure freezing/freeze-substitution techniques to gain more precise information about the organization of the ER in such cells. The most notable findings are: the identification of a specialized form of ER, termed "nodal ER," which is found exclusively in columella cells; the demonstration that the bulk of the ER is organized in the form of a tubular network that is confined to a peripheral layer under the plasma membrane; and the discovery that this ER-rich peripheral region excludes Golgi stacks, vacuoles, and amyloplasts but not mitochondria. Nodal ER domains consist of an approximately 100-nm-diameter central rod composed of oblong subunits to which usually seven sheets of rough ER are attached along their margins. These domains form patches at the interface between the peripheral ER network and the ER-free central region of the cells, and they occupy defined positions within central and flanking columella cells. Over one-half of the nodal ER domains are located along the outer tangential walls of the flanking cells. Cytochalasin D and latrunculin A cause an increase in size and a decrease in numbers of nodal ER domains. We postulate that the nodal ER membranes locally modulate the gravisensing signals produced by the sedimenting amyloplasts, and that the confinement of all ER membranes to the cell periphery serves to enhance the sedimentability of the amyloplasts in the central region of columella cells.

NASA Discipline Cell Biology↗
NASA NTRS2007

Spatial Correlation and Coherence of Boundary Layer Winds Near Cape Canaveral Florida

The spatial correlation and coherence of winds over separation distances from 8.5 to 31 km based on central Florida data from November 1999 through August 2001 are presented. The winds at altitudes from 500 to 3000 m were measured using a network of five radar wind profilers. The goal was to determine the extent to which the profilers may be considered independent data sources. Quality controlled profiles were produced every 15 minutes for up to sixty gates, each representing 101 m in altitude over the range from 130 m to 6089 m. Five levels, each containing three consecutive gates, were selected for analysis. These levels covered the range from 433 to 3059 m. The results show that the profilers are independent for features having time scales of less than one hour in the winter or two hours in the summer. This does not depend significantly on height. Because the size of the network coincides with the "spectral gap" in the boundary layer, the result also does not depend on the spacing of the profilers within the network.

Merceret, Francis J.↗
NASA NTRS1987

Network protocols for real-time applications

The Fiber Distributed Data Interface (FDDI) and the SAE AE-9B High Speed Ring Bus (HSRB) are emerging standards for high-performance token ring local area networks. FDDI was designed to be a general-purpose high-performance network. HSRB was designed specifically for military real-time applications. A workshop was conducted at NASA Ames Research Center in January, 1987 to compare and contrast these protocols with respect to their ability to support real-time applications. This report summarizes workshop presentations and includes an independent comparison of the two protocols. A conclusion reached at the workshop was that current protocols for the upper layers of the Open Systems Interconnection (OSI) network model are inadequate for real-time applications.

Johnson, Marjory J.↗
DOE OSTI2022

Network upgrade exploiting multi band: S- or E-band?

Nowadays, the fiber spectrum is only partially exploited, i.e., mainly in the C-band and more recently in the C + L-band, where the fiber attenuation profile experiences the minimum. Thus, fiber communications technology—amplifiers, switching, transceivers, etc.—and networking solutions are mature for those spectrum bands. However, the continuous increase in traffic means that capacity saturation of the current infrastructure is looming. Taking advantage of the unused portions of the spectrum (e.g., the S- and E-bands) may be an efficient solution to accommodate an increase in traffic without installing new fibers. Research is thus investigating multi-band transmission and networking to evaluate and enable such network upgrades. Some issues need to be solved or taken into account, from the enabling technology (e.g., amplifiers in the S- or E-band are still under development) to physical layer effects previously neglected, such as stimulated Raman scattering (SRS). SRS affects wideband transmission, potentially degrading active channels. The contribution of this paper is the investigation of network upgrades for C + L-band systems. In particular, upgrades exploiting the E- and S-bands are compared taking into account each band capacity and the effects of SRS on both new and already deployed channels (in both the C- and L-bands). A detailed analysis of the physical layer is provided also in the presence of guard bands between previously exploited bands and the bands used for upgrade. By leveraging the physical layer assessment, a networking analysis is carried out to evaluate the supported traffic increase and also the signal quality degradation due to SRS on active channels. Overall, the results suggest that upgrades to the E- and S-bands support a comparable increase in traffic. However, the exploitation of the E-band with 14 THz of guard band between the C- and E-bands may avoid detrimental effects to already active channels in the C + L-band, suggesting this upgrade strategy can be the most effective of the two.

47 OTHER INSTRUMENTATION↗
NASA NTRS2003

Using Neural Networks to Describe Tracer Correlations

Neural networks are ideally suited to describe the spatial and temporal dependence of tracer-tracer correlations. The neural network performs well even in regions where the correlations are less compact and normally a family of correlation curves would be required. For example, the CH4-N2O correlation can be well described using a neural network trained with the latitude, pressure, time of year, and CH4 volume mixing ratio (v.m.r.). In this study a neural network using Quickprop learning and one hidden layer with eight nodes was able to reproduce the CH4-N2O correlation with a correlation co- efficient of 0.9995. Such an accurate representation of tracer-tracer correlations allows more use to be made of long-term datasets to constrain chemical models. Such as the dataset from the Halogen Occultation Experiment (HALOE) which has continuously observed CH4, (but not N2O) from 1991 till the present. The neural network Fortran code used is available for download.

Lary, D. J.↗
NASA NTRS2005

Mapping of ice layer extent and snow accumulation in the percolation zone of the Greenland ice sheet

The Greenland ice sheet underwent record extensive melt in 2002 and prolonged melt in 2003. The severe melting created a significant and extensive ice layer over the Greenland ice sheet. An innovative approach is developed to detect the ice layer formation using data acquired by the SeaWinds scatterometer on the QuikSCAT satellite. QuikSCAT backscatter together with in situ data from automatic weather stations of the Greenland Climate Network are used to map the extent of ice layer formation. The results reveal areas of extensive ice layer formed by the 2002 melt, which is consistent with the maximum melt extent in 2002. Moreover, during freezing seasons, QuikSCAT data show a linear decrease in backscatter (in decibels or dB) that is related to the amount of snow accumulation in the ice layer formation region. This snow accumulation signature is caused by the attenuation of radar waves in the snow layer, accumulating since the last major melt event, whose thickness appears as an exponential function in relation to the backscatter signature. We use the Greenland Climate Network data to calibrate the QuikSCAT accumulation rate in order to estimate and map snow accumulation. QuikSCAT results capture the extreme snowfall in mid-April 2003, which deposited more than 0.5 m of snow in a day as measured by the automated weather station at the NASA South East site. Large-scale QuikSCAT results show an anomalous increase of snow accumulation over the southeast region of Greenland during the 2002-2003 freezing season.

snow↗
DOE OSTI2020

Efficient proactive vehicle relocation for on-demand mobility service with recurrent neural networks

One major challenge for on-demand mobility service (OMS) providers is to seamlessly match empty vehicles with trip requests so that the total vacant mileage is minimized. In this work, we develop an innovative data-driven approach for devising efficient vehicle relocation policy for OMS that (1) proactively relocates vehicles before the demand is observed and (2) reduces the inequality among drivers' income so that the proactive relocation policy is fair and is likely to be followed by drivers. Our approach represents the fusion of optimization and machine learning methods, which comprises three steps: First, we formulate the optimal proactive relocation as an optimal/stable matching problems and solve for global optimal solutions based on historical data. Second, the optimal solutions are then grouped and fed to train the deep learning models which consist of fully connected layers and long short-term memory networks. Low rank approximation is introduced to reduce the model complexity and improve the training performances. Finally, we use the trained model to predict the relocation policy which can be implemented in real time. We conduct comprehensive numerical experiments and sensitivity analyses to demonstrate the performances of the proposed method using New York City taxi data. Here, the results suggest that our method will reduce empty mileage per trip by 54-70% under optimal matching strategy, and a 25-32% reduction can also be achieved by following stable matching strategy. We also validate that the predicted relocation policies are robust in the presence of uncertain passenger demand level and passenger trip-requesting behavior.

33 ADVANCED PROPULSION SYSTEMS↗
DOE OSTI2023

Load Profile Inpainting for Missing Load Data Restoration and Baseline Estimation

This paper introduces a Generative Adversarial Nets (GAN) based, Load Profile Inpainting Network (Load-PIN) for restoring missing load data segments and estimating the baseline for a demand response event. The inputs are time series load data before and after the inpainting period together with explanatory variables (e.g., weather data). Here, we propose a Generator structure consisting of a coarse network and a fine-tuning network. The coarse network provides an initial estimation of the data segment in the inpainting period. The fine-tuning network consists of self-attention blocks and gated convolution layers for adjusting the initial estimations. Loss functions are specially designed for the fine-tuning and the discriminator networks to enhance both the point-to-point accuracy and realisticness of the results. We test the Load-PIN on three real-world data sets for two applications: patching missing data and deriving baselines of conservation voltage reduction (CVR) events. We benchmark the performance of Load-PIN with five existing deep-learning methods. Our simulation results show that, compared with the state-of-the-art methods, Load-PIN can handle varying-length missing data events and achieve 15-30% accuracy improvement.

14 SOLAR ENERGY↗