Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Decoder”

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 1,225 records · Page 68

LUNA Condition-Based Monitoring Update: Auto-encoders on Actuator Data [Slides]

An auto-encoder is a neural network trained to output a reconstruction of its input. There is a ‘bottleneck’ layer whose dimension is typically lower than that of the input. Because the input passes through the bottleneck, the auto-encoder must learn to compress (encode) the input into this low-dimensional form, and to decompress (decode) the compressed data into a close approximation of its original form.

97 MATHEMATICS AND COMPUTING↗

Model Coupling Through Learned Representations

Reliable climate predictions are important for making robust decisions in response to the changing climate. This project aims to reduce mis-modeling uncertainties arising from the representation of the land-atmosphere coupling in the Energy Exascale Earth System Model (E3SM) by using a machine learning approach. This approach will use an encoder-decoder architecture to represent the information that is developed in the land model and given to the atmosphere model. The simulated data will be taken from the E3SM simulation. However, the incorporation of observed data into the simulated dataset reduces mis-modeling uncertainties.

54 ENVIRONMENTAL SCIENCES↗

Resolving Lipid Structures for Threat-Agnostic Signatures

Development of threat-agnostic biosignatures is critical for the detection of and response to biological threats that go beyond the historical list-based approach. Lipids are a class of structurally diverse biomolecules that hold great promise as relevant threat-agnostic biosignatures, yet the fundamental challenge of resolving and decoding structural differences among lipids remains an obstacle in the pursuit of signature development. Structural analysis of complex lipids remains stymied due to spectral complexity and the inability to resolve low abundance lipids. To address these challenges, we integrated ozonolysis with ion mobility mass spectrometry (IMS-MS) analysis to determine the structural information for complex lipid species and (2) reduced lipid sample complexity prior to introduction into the IMS-MS using normal phase high pressure liquid chromatography (HPLC) to enable the ozonolysis of low abundance lipids. We applied these new capabilities to define the lipid structures in antimicrobial resistant (AMR) and antimicrobial susceptible (AMS) biothreat organism Yersinia pestis and cells infected with Severe Acute Respiratory Syndrome coronavirus 2 (SARS-CoV-2) variants.

59 BASIC BIOLOGICAL SCIENCES↗

Deployment of Dynamic Neural Network Optimization to Minimize Heat Rate During Ramping for Coal Power Plants (Final Technical Report)

Much success was achieved throughout the course of this project. A successful implementation of Dynamic Neural Network Optimization (D-NNO) was coupled with Adaptive Predictive Controls (APC) and a novel hardware installation comprised of an advanced sensor network (ASN) measuring mass-weighted averages of flue gas constituents above the horizontal superheater of a coal-fired utility boiler. From 2019 through 2023 (including an extension due to COVID delays), the team was able to prototype, evaluate, deploy, iterate, and ultimately finalize an advanced closed-loop control D-NNO system which demonstrated the ability to: •improve unit efficiency ~2.0% relative to unoptimized operation (represented as total fuel fired per MWh generated) •improve unit NOx emission rates 10%+ beyond static optimization baselines •improve unit temperature stability as much as 58% and on average 12% •improve operating load stability as much as 35% The culmination of this project has generated an advanced methodology of deploying specially designed recurrent neural networks (long short-term memory, gated recurrent unit, encoder-decoder networks, transformers, etc.), customized trajectory planning and closed-loop optimization modules capable of adapting to live electric grid responses and demands, self-tuning and adaptive expert controls constantly adjusting prediction parameters to real-time unit behavior, and a hardware/software package able to reliably calculate net unit heat rate (NUHR) in real-time using flue gas constituents, machine learning, and known combustion relationships. Through this real-time NUHR value, immediate feedback on system adjustments relative to operating efficiency was available, allowing for rapid improvements to system performance. In addition to development and deployment of the advanced D-NNO system, the approach methodology has been readily commercialized through the project platform Griffin Open Systems, LLC, the D-NNO software platform host. Similar methodologies to those developed by this project have already been deployed at 5 other units across the United States, with another 6 implementations scheduled, and more expected. Over the course of the project, multiple academic papers were submitted and accepted for publication within esteemed academic journals, and PhD students were trained and graduated, as well as undergraduate students becoming involved and participating to project objectives.

01 COAL, LIGNITE, AND PEAT↗

All-digital Sensor System for Distributed Downhole Pressure Monitoring in Unconventional Fields

This project developed and validated (through field tests) a new low-cost all-digital pressure sensing technology for in situ distributed downhole pressure monitoring in unconventional oil and gas (UOG) fields. The all-digital sensing technology uses a built-in non-electric analog-to-digital converter (ADC) to transform the pressure information into a combination of binary (ON/OFF) states. As such, the system does not need downhole electronics for signal conditioning and telemetry. The all-digital sensors can be remotely logged over a long distance, and many sensors can be multiplexed for distributed sensing. Based on a review of unconventional wells in the Lower 48 states, the specification of the sensor is to measure pressure up to 69 MPa (10,000 psi) and temperature up to 250°C. A sensor with a helical bourdon sensing element and a digital signal decoder of 50 mm diameter and 109 mm length was constructed. The helical bourdon sensing element was made of 304L stainless steel and filled with motor oil. The digital converter was made up of 8 digital reading pads constructed of high-temperature epoxy with conductive inserts made of stainless steel. The sensor had a linear response to pressure with an accuracy of 0.14 MPa (20 psi). To withstand the high pressure, the sensor was enclosed in a stainless-steel pressure housing with a wall thickness of 5.5 mm, a diameter of 73 mm, and a length of 724 mm. In the laboratory tests, the sensor exhibited no temperature-related effects on the results. The sensor did not show drift over a 14-day test period at elevated pressure. A field test was conducted where the sensor was deployed in a test wellbore at the Quest drilling test facility to a depth of 0 feet over three weeks. The sensor was attached to the production rods, along with a downhole reference sensor of PPS27 type, which is a permanent downhole monitoring system. During the testing phase, the test well annular blow-out preventer was closed, and the well was pressurized at the surface to 11 MPa (1600 psi). The sensor read the elevated bottom hole pressure of 1500 psi. A multiplexing unit was created for the sensor to deploy multiple sensors on one data transmission line in a distributed approach. The multiplexing unit was tested in a simulated environment of 3048 m (10,000 ft) with five sensors distributed. The sensors were pressurized at different intervals. The multiplexed sensors recorded the correct pressure, and the multiplexing did not interfere with the readings. The proposed concept of an all-digital pressure sensor for harsh downhole environments was designed, manufactured, and tested in the laboratory and tested at the field to a up to 69 MPa and 250°C. This technology has high-temperature tolerance and has potential in downhole areas outside oil and gas, such as carbon capture and storage (CCS) and geothermal wells. The sensor concept has been proven in this project, but to create a commercially viable product, manufacturing a sensor with a smaller diameter needs to be performed.

02 PETROLEUM↗

ESD: Ethernet Signal Differentiator [Poster]

Can Machine Learning Algorithms be trained to interpret and decode passively observed Automative Ethernet full-duplex signals without access to the original signals transmitted by either endpoint?

97 - MATHEMATICS AND COMPUTING↗

Intelligent Experiments Through Real-time AI: Fast Data Processing and Autonomous Detector Control for sPHENIX and Future EIC Detectors (Final Report)

The overall vision of this project was to integrate real-time artificial intelligence (AI) directly into the data acquisition and detector-control systems of nuclear physics experiments, including both fast online event selection and an autonomous detector-control feedback loop. The work carried out under the award focused on the fast online event-selection half of that vision: the efficient recording of low-momentum heavy-flavor (HF) hadron decays in proton-proton collisions at the sPHENIX experiment at the Relativistic Heavy Ion Collider (RHIC)—an observable that requires fast tracking and topological trigger selection not previously demonstrated at RHIC, and that is essential for QCD studies at future facilities such as the Electron-Ion Collider (EIC). The autonomous detector-control (GPU-based feedback) component named in the project title remained a design concept and was not implemented under this award. The Massachusetts Institute of Technology (MIT) group led the offline simulation and data processing needed to train the machine-learning (ML) models, the translation of trained models to Field-Programmable Gate Array (FPGA) firmware using the hls4ml framework, and the physics validation of heavy-flavor reconstruction. Over the award period, the team developed and hardware-tested the principal components of an AI-based heavy-flavor trigger on simulated and recorded sPHENIX tracker data: a software Bipartite Graph Attention Network (BiGAT) trigger model reaching > 95% signal efficiency at 99% background rejection; an FPGA-native hit clusterizer matching the offline clustering; smaller networks synthesized to FPGA within the required sub-10 µs latency; and an assembled decoder–clusterizer–inference firmware chain exercised on the FELIX readout board. A complete, fully integrated hardware demonstrator was not finished within the award period. This report documents the project goals, the MIT group’s contributions, the technical accomplishments, and the outlook toward applications at the future EIC ePIC detector.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Distributed video coding for arrays of remote sensing nodes : final report.

This document is the final report for the Sandia National Laboratory funded Student Fellowship position at New Mexico State University (NMSU) from 2008 to 2010. Ivan Mecimore, the PhD student in Electrical Engineering at NMSU, was conducting research into image and video processing techniques to identify features and correlations within images without requiring the decoding of the data compression. Such an analysis technique would operate on the encoded bit stream, potentially saving considerable processing time when operating on a platform that has limited computational resources. Unfortunately, the student has elected in mid-year not to continue with his research or the fellowship position. The student is unavailable to provide any details of his research for inclusion in this final report. As such, this final report serves solely to document the information provided in the previous end of year summary.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

Imaging the Wake of a Jet with Energy Correlators

In heavy ion collisions, jets are formed during the initial hard scattering event and move through the quark-gluon plasma (QGP) as it evolves. The evolution of the medium manifests itself on the jet via various jet-medium interactions, modifying its substructure as compared to vacuum QCD. Recently, the energy correlator observable has been shown to be a powerful method to decode correlations in pp collisions through measurements in ALICE, CMS and STAR. Additionally, recent studies have shown the sensitivity of this observable to medium modification due to the QGP. In this work [1], we perform the first study of the three-point energy correlator for medium modified jets, to show how it can be used to map out the shape of medium induced modifications. We illustrate this for the specific case of the Hybrid Model, where we show how higher-point correlation functions can image the wake of a jet.

Rai, Ananya [Yale University, New Haven, CT (Unite↗

Efficient Hamiltonian encoding algorithms for extracting quantum control mechanism as interfering pathway amplitudes in the Dyson series

Hamiltonian encoding is a methodology for revealing the mechanism behind the dynamics governing controlled quantum systems. In this paper, following Mitra and Rabitz \cite{abhra_1}, we define mechanism via pathways of eigenstates that describe the evolution of the system, where each pathway is associated with a complex-valued amplitude corresponding to a term in the Dyson series. The evolution of the system is determined by the constructive and destructive interference of these pathway amplitudes. Pathways with similar attributes can be grouped together into pathway classes. The amplitudes of pathway classes are computed by modulating the Hamiltonian matrix elements and decoding the subsequent evolution of the system rather than by direct computation of the individual terms in the Dyson series. The original implementation of Hamiltonian encoding was computationally intensive and became prohibitively expensive in large quantum systems. This paper presents two new encoding algorithms that calculate the amplitudes of pathway classes by using techniques from graph theory and algebraic topology to exploit patterns in the set of allowed transitions, greatly reducing the number of matrix elements that need to be modulated. These new algorithms provide an exponential decrease in both computation time and memory utilization with respect to the Hilbert space dimension of the system. To demonstrate the use of these techniques, they are applied to two illustrative state-to-state transition problems.

Abrams, Erez [Princeton University, Massachusetts ↗

Efficient Distributed Sequence Parallelism for Transformer-Based Image Segmentation

We introduce an efficient distributed sequence parallel approach for training transformer-based deep learning image segmentation models. The neural network models are comprised of a combination of a Vision Transformer encoder with a convolutional decoder to provide image segmentation mappings. The utility of the distributed sequence parallel approach is especially useful in cases where the tokenized embedding representation of image data are too large to fit into standard computing hardware memory. To demonstrate the performance and characteristics of our models trained in sequence parallel fashion compared to standard models, we evaluate our approach using a 3D MRI brain tumor segmentation dataset. We show that training with a sequence parallel approach can match standard sequential model training in terms of convergence. Furthermore, we show that our sequence parallel approach has the capability to support training of models that would not be possible on standard computing resources.

Lyngaas, Isaac↗

Transformer Neural Networks with Spatiotemporal Attention for Predictive Control and Optimization of Industrial Processes

In the context of real-time optimization and model predictive control of industrial systems, machine learning, and neural networks represent cutting-edge tools that hold promise for enhancing dynamic modeling. This work presents a novel transformer neural network architecture for real-time optimization and model predictive control. This network design includes a modified attention mechanism inspired by positional embedding attention from vision transformers and task-specific modifications to the input-output structure of the transformer’s decoder stack. Experiments were conducted using data from a 450 MW coal-fired power plant to evaluate this approach's effectiveness. The transformer neural network was compared with conventional recurrent models, including GRU and LSTM. The transformer exhibited a 6% increase in the R-squared (R2) value of predictions and an 83% reduction in mean squared error (MSE). Computation time was also reduced by 84% compared to conventional recurrent models.

Gallup, Ethan R.↗

ISiCLE: A Quantum Chemistry Pipeline for Establishing in Silico Collision Cross Section Libraries

High throughput, comprehensive, and confident identifications of metabolites and other chemicals in biological and envi-ronmental samples will revolutionize our understanding of the role of these intriguing and chemically diverse molecules in biological systems. Despite recent technological advances, a metabolomics study still results in the detection of a dispropor-tionate number of features than cannot be confidently assigned to a chemical structure. This inadequacy is driven by the sin-gle most significant limitation in metabolomics: the reliance on reference libraries constructed through the analysis of au-thentic reference chemicals and the limited commercial availability of these compounds. To this end, we have developed the in silico chemical library engine (ISiCLE), a high-performance computing-friendly cheminformatics workflow for generating chemical properties. In the instantiation described here, we predict probable three-dimensional molecular conformers (i.e. conformational isomers) using chemical identifiers as input, from which collision cross sections (CCS) are derived. The ap-proach employs state-of-the-art first-principles simulation, distinguished by use of molecular dynamics, quantum chemistry, and ion mobility calculations to generate predictions, all without training data. Importantly, optimization of ISiCLE includ-ed a refactoring of the popular MOBCAL code for trajectory-based mobility calculations, improving its computational effi-ciency by over two orders of magnitude. Calculated CCS values were validated against 1,983 experimental CCS values and compared to previously reported CCS calculation approaches. Average CCS calculation absolute error is 3.2% with standard deviation 3.1%. An online database is introduced for sharing both calculated and experimental CCS values (metabolom-ics.pnnl.gov), initially including a CCS library with over 0.8M entries. Finally, three successful applications of molecule characterization using calculated CCS are described, including the identification of an environmental degradation product, the separation of molecular isomers, and the decoding of complex blinded mixtures of exposure chemicals. This work represents a promising method to address the limitations of small molecule identification, and offers alternatives to standards-based chemical identification amenable to high-performance computation.

Colby, Sean M↗

Deep learning to generate in silico chemical property libraries and candidate molecules for small molecule identification in complex samples

Comprehensive and unambiguous identification of small molecules in complex samples will revolutionize our understanding of the role of metabolites in biological systems. Existing and emerging technologies have enabled measurement of chemical properties of molecules in complex mixtures and, in concert, are sensitive enough to resolve even stereoisomers. Despite these experimental advances, small molecule identification is inhibited by (i) chemical reference libraries (e.g. mass spectra, collision cross section, and other measurable property libraries) representing <1% of known molecules, limiting the number of possible identifications, and (ii) the lack of a method to generate candidate matches directly from experimental features (i.e. without a library). To this end, we developed a variational autoencoder (VAE) to learn a continuous numerical, or latent, representation of molecular structure to expand reference libraries for small molecule identification. We extended the VAE to include a chemical property decoder, trained as a multitask network, in order to shape the latent representation such that it assembles according to desired chemical properties. The approach is unique in its application to metabolomics and small molecule identification, with its focus on properties that can be obtained from experimental measurements (m/z, CCS) paired with its training paradigm, which involved a cascade of transfer learning iterations. First, molecular representation is learned from a large dataset of structures with m/z labels. Next, in silico property values are used to continue training, as experimental property data is limited. Finally, the network is further refined by being trained with the experimental data. This allows the network to learn as much as possible at each stage, enabling success with progressively smaller datasets without overfitting. Once trained, the network can be used to predict chemical properties directly from structure, as well as generate candidate structures with desired chemical properties. Our approach is orders of magnitude faster than first-principles simulation for CCS property prediction. Additionally, the ability to generate novel molecules along manifolds, defined by chemical property analogues, positions DarkChem as highly useful in a number of application areas, including metabolomics and small molecule identification, drug discovery and design, chemical forensics, and beyond.

Colby, Sean M↗

On Single-Cell Enzyme Assays in Marine Microbial Ecology and Biogeochemistry

Extracellular enzyme activity is a well-established parameter for evaluating microbial biogeochemical roles in marine ecosystems. The presence and activity of extracellular enzymes in seawater provide insights into the quality and quantity of organic matter being processed by the present microorganisms. A key challenge in our understanding of these processes is to decode the extracellular enzyme repertoire and activities of natural communities at the single-cell level. Current measurements are carried out on bulk or size-fractionated samples capturing activities of mixed populations. This approach – even with size-fractionation – cannot be used to trace enzymes back to their producers, nor distinguish the active microbial members, leading to a disconnect between measured activities and the producer cells. By targeting extracellular enzymes and resolving their activities at the single-cell level, we can investigate underlying phenotypic heterogeneity among clonal or closely related organisms, characterize enzyme kinetics under varying environmental conditions, and resolve spatio-temporal distribution of individual enzyme producers within natural communities. In this perspective piece, we discuss state-of-the-art technologies in the fields of microfluidic droplets and functional screening of prokaryotic cells for measuring enzyme activity in marine seawater samples, one cell at a time. We further elaborate on how this single-cell approach can be used to address research questions that cannot be answered with current methods, as pertinent to the enzymatic degradation of organic matter by marine microorganisms.

54 ENVIRONMENTAL SCIENCES↗

Ends of the line for tmRNA-SmpB

Genes for the RNA tmRNA and protein SmpB, partners in the trans-translation process that rescues stalled ribosomes, have previously been found in all bacteria and some organelles. We validate recent identification of tmRNA homologs in oomycete mitochondria by finding partner genes from oomycete nuclei that target SmpB to the mitochondrion. Exhaustive search now identifies a small number of complete, often highly derived, bacterial genomes that appear to lack a functional copy of one or the other partner gene (but not both). Three groups with reduced genomes have lost the central loop of SmpB, which is thought to improve alanylation and EF-Tu activation: Carsonella, Hodgkinia and the hemplasmas (hemotropic Mycoplasma). Carsonella has also lost the SmpB C-terminal tail, thought to stimulate the decoding center of the ribosome. Carsonella moreover exhibits gene overlap such that tmRNA maturation should produce a non-stop smpB mRNA, and one isolate exhibits complete degradation of the tmRNA gene yet its smpB shows no evidence for relaxed selective constraint. After loss of the SmpB central loop in the hemoplasmas, a subclade apparently lost tmRNA. At least some of the tmRNA/SmpB-deficient strains appear to further lack the ArfA and ArfB backup systems for ribosome rescue. The most frequent neighbors of smpB are the tmRNA gene, a ratA/rnfH unit, and the gene for RNaseR, a known physical and functional partner of tmRNA-SmpB. The tmRNA Website has moved and been updated, adding an SmpB sequence database (http://bioinformatics.sandia.gov/tmrna).

59 BASIC BIOLOGICAL SCIENCES↗

Towards interpretable Cryo-EM: disentangling latent spaces of molecular conformations

Molecules are essential building blocks of life and their different conformations (i.e., shapes) crucially determine the functional role that they play in living organisms. Cryogenic Electron Microscopy (cryo-EM) allows for acquisition of large image datasets of individual molecules. Recent advances in computational cryo-EM have made it possible to learn latent variable models of conformation landscapes. However, interpreting these latent spaces remains a challenge as their individual dimensions are often arbitrary. The key message of our work is that this interpretation challenge can be viewed as an Independent Component Analysis (ICA) problem where we seek models that have the property of identifiability. That means, they have an essentially unique solution, representing a conformational latent space that separates the different degrees of freedom a molecule is equipped with in nature. Thus, we aim to advance the computational field of cryo-EM beyond visualizations as we connect it with the theoretical framework of (nonlinear) ICA and discuss the need for identifiable models, improved metrics, and benchmarks. Moving forward, we propose future directions for enhancing the disentanglement of latent spaces in cryo-EM, refining evaluation metrics and exploring techniques that leverage physics-based decoders of biomolecular systems. Moreover, we discuss how future technological developments in time-resolved single particle imaging may enable the application of nonlinear ICA models that can discover the true conformation changes of molecules in nature. The pursuit of interpretable conformational latent spaces will empower researchers to unravel complex biological processes and facilitate targeted interventions. This has significant implications for drug discovery and structural biology more broadly. More generally, latent variable models are deployed widely across many scientific disciplines. Thus, the argument we present in this work has much broader applications in AI for science if we want to move from impressive nonlinear neural network models to mathematically grounded methods that can help us learn something new about nature.

59 BASIC BIOLOGICAL SCIENCES↗

Defining the Plant Peroxisomal Proteome: From Arabidopsis to Rice

Peroxisomes are small subcellular organelles mediating a multitude of processes in plants. Proteomics studies over the last several years have yielded much needed information on the composition of plant peroxisomes. In this review, the status of peroxisome proteomics studies in Arabidopsis and other plant species and the cumulative advances made through these studies are summarized. A reference Arabidopsis peroxisome proteome is generated, and some unique aspects of Arabidopsis peroxisomes that were uncovered through proteomics studies and hint at unanticipated peroxisomal functions are also highlighted. Knowledge gained from Arabidopsis was utilized to compile a tentative list of peroxisome proteins for the model monocot plant, rice. Differences in the peroxisomal proteome between these two model plants were drawn, and novel facets in rice were expounded upon. Finally, we discuss about the current limitations of experimental proteomics in decoding the complete and dynamic makeup of peroxisomes, and complementary and integrated approaches that would be beneficial to defining the peroxisomal metabolic and regulatory roadmaps. The synteny of genomes in the grass family makes rice an ideal model to study peroxisomes in cereal crops, in which these organelles have received much less attention, with the ultimate goal to improve crop yield.

54 ENVIRONMENTAL SCIENCES↗