Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “ROSES”

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 1,369 records · Page 76
All sourcesNASA NTRSDOE OSTI
DOE OSTI2018

Ramifications of Evolving Misbehaving Convolutional Neural Network Kernel and Batch Sizes

Deep-learners have many hyper-parameters including learning rate, batch size, kernel size - all playing a significant role toward estimating high quality models. Discovering useful hyper-parameter guidelines is an active area of research, though the state of the art generally uses a brute force, uniform grid approach or random search for finding ideal settings. We share the preliminary results of using an alternative approach to deep learner hyper-parameter tuning that uses an evolutionary algorithm to improve the accuracy of a deep-learner models used in satellite imagery building footprint detection. We found that the kernel and batch size hyper-parameters surprisingly differed from sizes arrived at via a brute force uniform grid approach. These differences suggest a novel role for evolutionary algorithms in determining the number of convolution layers, as well as smaller batch sizes in improving deep-learner models.

Coletti, Mark↗
DOE OSTI2016

Synchrotron X-Ray Characterization of Alkali Elements at Grain Boundaries in Cu(In, Ga)Se2 Solar Cells

It is well known that the addition of alkali elements such as Na and K during and after growth of Cu(In, Ga)Se2 (CIGS) has beneficial effects on the electronic properties of bulk material, improving device performance significantly. While the device level effects have been measured and reported, a direct observations of the localization of Na including its chemical nature are missing, and the impact of Na on elemental and phase segregation during CIGS growth is not fully understood. We investigate these aspects to shine light on the role of Na in CIGS solar cells with the ultimate goal of increasing their conversion efficiency. Utilizing a suite of synchrotron based x-ray characterization techniques, we discuss the challenges and advantages of these techniques for investigating segregation of main constituents of CIGS, Na distribution, chemical bonding of Na, and collection efficiency in CIGS as well as their correlations.

14 SOLAR ENERGY↗
DOE OSTI2018

The Impact of a Faster Thermal Cycling Profile on PV Cell Interconnect Cycles to Failure

The thermal cycling tests used in the photovoltaic industry are slow, requiring over a month for 200 cycles and five months for 800 cycles. A faster alternative would reduce the cost of delay for module qualification. This study investigates the correlation between a baseline thermal cycling profile and profiles that are three and six times faster. Interconnect failure data for each profile was gathered across eight splits of varying material combinations. Models were applied to the failure data and factors which impacted interconnect lifetime were identified. Here, the results show that more cycles were required in the faster profiles for equivalency to the baseline profile. Furthermore, the equivalent number of cycles varied by a factor of two across the splits. Despite these limitations, the faster profiles offer substantial qualification time savings if the underlying failure mechanisms are understood.

14 SOLAR ENERGY↗
DOE OSTI2018

167-PFlops Deep Learning for Electron Microscopy: From Learning Physics to Atomic Manipulation

An artificial intelligence system called MENNDL, which used 25,200 Nvidia Volta GPUs on Oak Ridge National Laboratory’s Summit machine, automatically designed an optimal deep learning network in order to extract structural information from raw atomic-resolution microscopy data. In a few hours, MENNDL creates and evaluates millions of networks using a scalable, parallel, asynchronous genetic algorithm augmented with a support vector machine to automatically find a superior deep learning network topology and hyper-parameter set than a human expert can find in months. For the application of electron microscopy, the system furthers the goal of improving our understanding of the electron-beam-matter interactions and real-time image-based feedback, which enables a huge step beyond human capacity toward nanofabricating materials automatically. Finally, MENNDL has been scaled to the 4,200 available nodes of Summit achieving a measured 152.5 PFlops, with an estimated sustained performance of 167 PFlops when the entire machine is available.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2019

A Twin Memristor Synapse for Spike Timing Dependent Learning in Neuromorphic Systems

Neuromorphic systems consist of a framework of spiking neurons interconnected via plastic synaptic junctures. The discovery of a two terminal passive nanoscale memristive device has spurred great interest in the realization of memristive plastic synapses in neural networks. In this work, a synapse structure is presented that utilizes a pair of memristors, to implement both positive and negative weights. The working scheme of this synapse as an electrical interlink between neurons is explained, and the relative timing of their spiking events is analyzed, which leads to a modulation of the synaptic weight in accordance with the spike-timing-dependent plasticity (STDP) rule. A digital pulse width modulation technique is proposed to achieve these variable changes to the synaptic weight. The synapse architecture presented is shown to have high accuracy when used in neural networks for classification tasks. Lastly, the energy requirement of the system during various phases of operation is presented.

Adnan, Md Musabbir↗
DOE OSTI2020

Onset of Fogging and Degradation in Polyvinyl Toluene-Based Scintillators

Plastic-based gamma-ray detector materials such as polyvinyl toluene (PVT) occasionally suffer from loss of optical light transmission due to internal defects or “fogging” when exposed to some environmental conditions over time. Fogging results in reduced scintillation light propagated to the photomultiplier tube for collection. Investigations of the physical cause of these defects and their associated environmental conditions are ongoing. The goal of this work was to characterize the signs and effects of fogging over a simulated 22-year lifespan of the PVT in outdoor environments. Here, two potential mitigation methods were also evaluated, and one successfully mitigated the phenomena.

36 MATERIALS SCIENCE↗
DOE OSTI2025

X-ray and Neutron Radiography for Quantitative Material Reconstructions

Radiography is a powerful tool to determine the interior structure of objects. X-ray radiography is widely used and provides high-resolution images, though X-rays have limited transmission through materials of high atomic number (Z) and density. In contrast, neutrons can penetrate many materials that are heavily attenuating to X-rays, such as metals, providing contrast in the inner layers of highly attenuating items. Past work has shown the value in using both X-ray and neutron radiography for estimating material thicknesses, though that work was limited to simulated data. Here, we demonstrate quantitative material reconstructions using experimental X-ray and neutron radiography data from lab-based systems, accurately modeling radiography system responses to within a few percent to enable quantitative measures of material thickness. We demonstrate the utility of neutron radiography and X-ray radiography for these quantitative reconstructions and introduce methods for using their complementarity to improve image quality and optimize experimental design.

Gilbert, Andrew J.↗
DOE OSTI2014

Parallel Operation of Multiple Closely Spaced Small Aspect Ratio Rod Pinches

A series of simulations and experiments to resolve questions about the operation of arrays of closely spaced small aspect ratio rod pinches has been performed. Design and post-shot analysis of the experimental results are supported by 3D particle-in-cell simulations. Both simulations and experiments support these conclusions. Penetration of current to the interior of the array appears to be efficient, as the current on the center rods is essentially equal to the current on the outer rods. Current loss in the feed due to the formation of magnetic nulls was avoided in these experiments by design of the feed surface of the cathode and control of the gap to keep the electric fields on the cathode below the emission threshold. Some asymmetry in the electron flow to the rod was observed, but the flow appeared to symmetrize as it reached the end of the rod. Interaction between the rod pinches can be controlled to allow the stable and consistent operation of arrays of rod pinches.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2022

Real-World Subsynchronous Oscillation Events in Power Grids with High Penetrations of Inverter-Based Resources

This paper presents a survey of real-world sub-synchronous oscillation events associated with inverter-based resources (IBR) over the past decade. The focus is on those oscillations in the subsynchronous frequency range known to be influenced by power grid characteristics, e.g., series compensation or low system strength. A brief overview of the historical events is presented followed by detailed descriptions of a series of events. This paper also examines causation mechanisms and proposes future research directions to meet grid needs worldwide.

24 POWER TRANSMISSION AND DISTRIBUTION↗