Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “AUTOS”

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 1,153 records · Page 64

An Inhibited Conformation for the Protein Kinase Domain of the Saccharomyces cerevisiae AMPK Homolog Snf1

AMP-activated protein kinase (AMPK) is a master metabolic regulator for controlling cellular energy homeostasis. Its homolog in yeast, SNF1, is activated in response to glucose depletion and other stresses. The catalytic ({alpha}) subunit of AMPK/SNF1 in yeast (Snf1) contains a protein Ser/Thr kinase domain (KD), an auto-inhibitory domain (AID) and a region that mediates interactions with the two regulatory ({beta} and {gamma}) subunits. Here, the crystal structure of residues 41-440 of Snf1, which include the KD and AID, is reported at 2.4 {angstrom} resolution. The AID is completely disordered in the crystal. A new inhibited conformation of the KD is observed in a DFG-out conformation and with the glycine-rich loop adopting a structure that blocks ATP binding to the active site.

59 BASIC BIOLOGICAL SCIENCES↗

Unsupervised learning approaches to characterizing heterogeneous samples using X-ray single-particle imaging

One of the outstanding analytical problems in X-ray single-particle imaging (SPI) is the classification of structural heterogeneity, which is especially difficult given the low signal-to-noise ratios of individual patterns and the fact that even identical objects can yield patterns that vary greatly when orientation is taken into consideration. Proposed here are two methods which explicitly account for this orientation-induced variation and can robustly determine the structural landscape of a sample ensemble. The first, termed common-line principal component analysis (PCA), provides a rough classification which is essentially parameter free and can be run automatically on any SPI dataset. The second method, utilizing variation auto-encoders (VAEs), can generate 3D structures of the objects at any point in the structural landscape. Both these methods are implemented in combination with the noise-tolerant expand–maximize–compress (EMC) algorithm and its utility is demonstrated by applying it to an experimental dataset from gold nanoparticles with only a few thousand photons per pattern. Both discrete structural classes and continuous deformations are recovered. These developments diverge from previous approaches of extracting reproducible subsets of patterns from a dataset and open up the possibility of moving beyond the study of homogeneous sample sets to addressing open questions on topics such as nanocrystal growth and dynamics, as well as phase transitions which have not been externally triggered.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

The structure of a contact-dependent growth-inhibition (CDI) immunity protein from Neisseria meningitidis MC58

Contact-dependent growth inhibition (CDI) is an important mechanism of intercellular competition between neighboring Gram-negative bacteria. CDI systems encode large surface-exposed CdiA effector proteins that carry a variety of C-terminal toxin domains (CdiA-CTs). All CDI + bacteria also produce CdiI immunity proteins that specifically bind to the cognate CdiA-CT and neutralize its toxin activity to prevent auto-inhibition. Here, the X-ray crystal structure of a CdiI immunity protein from Neisseria meningitidis MC58 is presented at 1.45 Å resolution. The CdiI protein has structural homology to the Whirly family of RNA-binding proteins, but appears to lack the characteristic nucleic acid-binding motif of this family. Sequence homology suggests that the cognate CdiA-CT is related to the eukaryotic EndoU family of RNA-processing enzymes. A homology model is presented of the CdiA-CT based on the structure of the XendoU nuclease from Xenopus laevis . Molecular-docking simulations predict that the CdiA-CT toxin active site is occluded upon binding to the CdiI immunity protein. Together, these observations suggest that the immunity protein neutralizes toxin activity by preventing access to RNA substrates.

59 BASIC BIOLOGICAL SCIENCES↗

A Real-Time Implementation and Validation of Federated Learning for Grid Services

Grid-edge devices are becoming increasingly important in the energy transition. Preserving privacy was not previously considered an important aspect for power grid operations, but with the increased proliferation of customer-owned assets, it is now an essential consideration. Several mechanisms have been proposed to provide privacy for non-utility owned assets in the power grid. Federated learning (FL) is one method gaining prominence in this area. Although FL has been used for other applications, such as auto-complete in phones, there has not been much investigation into whether these approaches are feasible for grid applications. In this work, we use a research platform with real-time simulators and hardware-in-the-loop capabilities to investigate how FL can be applied to grid-edge devices, and we present the potential grid services that can be derived for these devices. We discuss the computational challenges with deploying complex FL approaches, and we explore several grid services, including participation in retail electricity markets, voltage control, and resilience-driven reconfiguration.

24 POWER TRANSMISSION AND DISTRIBUTION↗

CALORIC: A readout chip for high granularity calorimeter

A very-front-end electronics has been developed to fulfil requirements for the next generation of electromagnetic calorimeters. The compactness of this kind of detector and its large number of channels (up to several millions) impose a drastic limitation of the power consumption and a high level of integration. The electronic channel proposed is first of all composed of a low-noise Charge Sensitive Amplifier (CSA) able to amplify the charge delivered by a silicon diode up to 10 pC. Next, a two-gain shaping, based on a Gated Integration (G.I.), is implemented to cover the 15 bits dynamic range required: a high gain shaper processes signals from 4 fC (charge corresponding to the MIP) up to 1 pC, and a low gain filter handles charges up to 10 pC. The G.I. performs also the analog memorization of the signal until it is digitalized. Hence, the analog-to-digital conversion is carried out through a low-power 12-bit cyclic ADC. If the signal overloads the high-gain channel dynamic range, a comparator selects the low-gain channel instead. Moreover, an auto-trigger channel has been implemented in order to select and store a valid event over the noise. The timing sequence of the channel is managed by a digital IP. It controls the G.I. switches, generates all needed clocks, drives the ADC and delivers the final result over 12 bits. The whole readout channel is power controlled, which permits to reduce the consumption according to the duty cycle of the beam collider. Simulations have been performed with Spectre simulator on the prototype chip designed with the 0.35 {mu}m CMOS technology from Austriamicrosystems. Results show a non-linearity better than 0.1% for the high-gain channel, and a non-linearity limited to 1% for the low-gain channel. The Equivalent Noise Charge referred to the input of the channel is evaluated to 0.4 fC complying with the MIP/10 limit. With the timing sequence of the International Linear Collider, which presents a duty cycle of 1%, the power consumption of the complete channel is limited to 43 {mu}W thanks to the power pulsing. The total area of the channel is 1.2 mm{sup 2} with an analog memory depth of 16. (authors)

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

A Control Scheme to Mitigate the Dead-Time Effects in a Wireless Power Transfer System

In practice, a dead-time is always provided between the complementary switching instances of the inverter phase-leg devices. At higher operating frequencies, the dead-time issues in wireless power transfer (WPT) systems become critical, especially as the power level increases. In certain operating conditions, the dead-time effect in wireless power transfer system affects the switching characteristics. Consequently, the switching losses in the power semiconductor devices increase and also impact the efficiency of the overall system. In this paper, a simple control scheme is proposed to eliminate the dead-time effect (or voltage polarity reversal) in the WPT inverter. The proposed control scheme monitors the inverter output voltage, and the switching frequency is auto-tuned to eliminate the undesired switching instances in the inverter voltage. The proposed control scheme is validated using the closed-loop simulations in PLECS, and the experimental results on a 5.6 kW WPT prototype are also presented. After eliminating the voltage-polarity-reversal at the inverter output, the inverter losses were reduced by ~40%, and the overall system losses were reduced by ~17%.

Kavimandan, Utkarsh↗

Control and Management of Multiple Converters in a Residential Smart Grid

Power electronic systems are becoming a staple building block for electric grid networks. However, these systems have been largely designed to grid integrate in ad hoc configurations with little or no coordination of control. This work proposes a central controller and power electronic system hardware and software design for a residential system (photovoltaic, energy storage and residential building). This system supports auto-integration, plug and play capabilities, and optimal energy management to meet different use cases. Modelling and multi-day testing of the system have been conducted in a controller hardware-in-the-loop (C-HIL) testbed. Multiple use cases and pricing options have been considered as part of the simulation and testing. Results are presented of these systems as a proof of principle.

Starke, Michael↗

Semi-Supervised Information Extraction for Cancer Pathology Reports

Pathology reports are a main source of data for cancer surveillance programs. Manual coding of pathology reports is labor-intensive but necessary for obtaining labeled data to train automated information extraction systems. In this study, we investigated semi-supervised deep learning, improving the performance of a multitask information extraction system for automated annotation of pathology reports. We used a set of over 374,000 pathology reports from the Louisiana Tumor Registry and a novel convolutional attention-based auto-encoder. We performed a set of experiments comparing supervised training augmented with unlabeled data at 1%, 5%, 10%, and 50% of the original data size. We also compared the impact of extending text processing to include unlabeled tokens. We find that semi-supervised training consistently improved individual performance with increased micro-averaged F-scores between 0.012 and 0.064 and increased macro-averaged F-scores of up to 0.158. This demonstrates that semantic information learned via unsupervised learning can be used to improve supervised clinical task performance.

Qiu, John X.↗

Spring Boot based REST API to Improve Data Quality Report Generation for Big Scientific Data: ARM Data Center Example

Web application technologies are growing rapidly with continuous innovation and improvements. This paper focuses on the popular Spring Boot [1] java-based framework for building web and enterprise applications and how it provides the flexibility for service-oriented architecture (SOA). One challenge with any Spring-based applications is its level of complexity with configurations. Spring Boot makes it easy to create and deploy stand-alone, production-grade Spring applications with very little Spring configuration. Example, if we consider Spring Model-View-Controller (MVC) framework [2], we need to configure dispatcher servlet, web jars, a view resolver, and component scan among other things. To solve this, Spring Boot provides several Auto Configuration options to setup the application with any needed dependencies. Another challenge is to identify the framework dependencies and associated library versions required to develop a web application. Spring Boot offers simpler dependency management by using a comprehensive, but flexible, framework and the associated libraries in one single dependency, which provides all the Spring related technology that you need for starter projects as compared to CRUD web applications. This framework provides a range of additional features that are common across many projects such as embedded server, security, metrics, health checks, and externalized configuration. Web applications are generally packaged as war and deployed to a web server, but Spring Boot application can be packaged either as war or jar file, which allows to run the application without the need to install and/or configure on the application server. In this paper, we discuss how Atmospheric Radiation Measurement (ARM) Data Center (ADC) at Oak Ridge National Laboratory, is using Spring Boot to create a SOA based REST [4] service API, that bridges the gap between frontend user interfaces and backend database. Using this REST service API, ARM scientists are now able to submit reports via a user form or a command line interface, which captures the same data quality or other important information about ARM data.

Guntupally, Kavya↗

Mitigating Catastrophic Forgetting in Deep Learning in a Streaming Setting Using Historical Summary

Recent advancements in scientific equipment and the adaptation of electronics and the Internet of Things (IoT) in our everyday lives resulted in large and complex data production at a high rate. Making meaningful and timely knowledge discovery at a modest cost from this big data is difficult for computing power and storage limitations. Training deep learning models incrementally in a streaming setting can help us with overcoming these limitations. However, in a well-known phenomenon named catastrophic forgetting, incrementally trained models increasingly perform poorly on the past data. To mitigate catastrophic forgetting in training in a streaming setting, we propose constructing a historical summary over time and use the summary with newly arrived data during incremental training. We propose various data summarization techniques such as random sampling, micro clustering, coreset computation, and Auto Encoders to counteract catastrophic forgetting. We built a pipeline for incremental training with a historical summary for training deep learning models for streaming data. We demonstrate the effectiveness of historical summary in mitigating catastrophic forgetting using three case studies involving three different deep learning applications: an Artificial Neural Network (ANN) for classification task on MNIST dataset, a language model (RNN-LM) on the WikiText2 dataset, and a Convolutional Neural Network (CNN), ResNet50 to classify the ImageNet dataset. Through the training of the models, we observe that catastrophic forgetting is evident in ANN and CNN but not in an RNN. For the first task, our method recovers up to 47.9% lost accuracy due to catastrophic forgetting. For the third task, the historical summary recovers classification accuracy by up to 25%. For the second task, though there is not proof of catastrophic forgetting, the training performance (PPL) improves by up to 26% with historical summary.

Dash, Sajal↗

Intelligent Prediction of States in Multi-port Autonomous Reconfigurable Solar power plant (MARS)

In power electronics, prediction of states may be used for identification of faults, determination of aging of components, identification of bad data measurements, among others. Prediction of states in power electronics have broadly been based on: (a) physics-based models, (b) data-driven models, and (c) hybrid models. In this paper, data-driven approaches are presented for intelligent prediction of states in multi-port autonomous reconfigurable solar power plant (MARS) and compared. The data-set needed to train the data-driven models based on artificial intelligence (AI) algorithms has been identified and the trained models are evaluated under different extrapolated normal and abnormal operating conditions. The AI algorithms include nonlinear auto-regressive exogenous model (NARX), spiking neural networks (SNN), and decision tree. The models are compared and contrasted. The best model (NARX) is evaluated under different normal and abnormal operating conditions that have indicated accurate prediction.

Debnath, Suman↗

USABC Development of 12 Volt Battery for Start-Stop Application: Preprint

Global automakers are accelerating the development of fuel efficient vehicles, as a part of meeting regional regulatory CO2 emissions requirements. The micro hybrid vehicles with auto start-stop functionality are considered economical solutions for the stringent European regulations. Flooded lead acid batteries were initially considered the most economical solution for idle-stop systems. However, the dynamic charge acceptance (DCA) at lower state-of-charge (SOC) was limiting the life of the batteries. While improved lead-acid batteries with AGM and VRLA features have improved battery longevity, they do not last the life of the vehicle. The United States Advanced Battery Consortium (or USABC, a consortium of GM, Ford, and Chrysler) analyzed energy storage needs for a micro hybrid automobile with start-stop capability, and with a single power source. USABC has analyzed the start-stop behaviors of many drivers and has developed the requirements for the start-stop batteries (Table 3). The testing procedures to validate the performance and longevity were standardized and published. The guideline for the cost estimates calculations have also been provided, in order to determine the value of the newly developed modules. The analysis effort resulted in a set of requirements which will help the battery manufacturers to develop a module to meet the automotive Original Equipment Manufacturers (OEM) micro hybrid vehicle requirements. Battery developers were invited to submit development proposals and two proposals were selected for 50% cost share with USABC/DOE.

12 V START-STOP↗

Clustering High-dimensional Toxicogenomics Data with Rare Signals

Toxicogenomics studies the gene and protein activities to drug treatments or toxic exposures. As the drugs and genes are numerous, toxicogenomics data are naturally high dimensional, with dimension sizes up to millions. In addition, the distribution of toxicogenomics data is oftentimes skewed, and they contain rare but important signals representing a cell or organism’s response to toxicity. The combination of high dimension and extremely skewed distribution of toxicogenomics data makes clustering analysis extremely challenging.We present our study of clustering toxicogenomics data using classical approaches such as principal component analysis as well as deep learning approaches such as auto-encoders. Our experiments show that these approaches fail to preserve rare signals and produce high-quality clusters. We then explore augmenting matrix factorization with deep learning techniques such as attention mechanism to produce latent representations for clustering. Our technique is able to better preserve rare signals after dimensionality reduction than prior approaches. Furthermore, we combine our augmented matrix factorization with a mechanism similar to autoencoder to balance separable clusters and low regeneration errors. Our experiments demonstrate better clustering with our proposed approach.

Cong, Guojing↗

A Step towards Energy Efficient Computing: Redesigning a Hydrodynamic Application on CPU-GPU, In: 2014 IEEE 28th International Parallel and Distributed Processing Symposium

Power and energy consumption are becoming an increasing concern in high performance computing. Compared to multi-core CPUs, GPUs have a much better performance per watt. In this paper we discuss efforts to redesign the most computation intensive parts of BLAST, an application that solves the equations for compressible hydrodynamics with high order finite elements, using GPUs BLAST, Dobrev. In order to exploit the hardware parallelism of GPUs and achieve high performance, we implemented custom linear algebra kernels. We intensively optimized our CUDA kernels by exploiting the memory hierarchy, which exceed the vendor's library routines substantially in performance. We proposed an auto tuning technique to adapt our CUDA kernels to the orders of the finite element method. Compared to a previous base implementation, our redesign and optimization lowered the energy consumption of the GPU in two aspects: 60% less time to solution and 10% less power required. Compared to the CPU-only solution, our GPU accelerated BLAST obtained a 2.5× overall speedup and 1.42× energy efficiency (green up) using 4th order (Q_4) finite elements, and a 1.9× speedup and 1.27× green up using 2nd order (Q2) finite elements.

Computer Science↗