Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Continual learning”

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 973 records · Page 54

PWR loading pattern optimization with reinforcement learning

The core loading pattern optimization problem belongs to the class of combinatorial optimization problem and has been studied since the dawn of commercial nuclear energy industry. It is characterized by multiple objectives and constraints, with a very high number of candidate patterns, which makes it impossible to solve explicitly. Stochastic optimization methodologies including Genetic Algorithms and Simulated Annealing are used by different nuclear utilities and vendors to perform fuel cycle reload design. Nevertheless, hand-designed solutions continue to be the prevalent method in the industry. To improve the state-of-the-art core reload patterns, we aim to create a method as scalable as possible, that agrees with the designer's goal of performance and safety. To help in this task Deep Reinforcement Learning (DRL), in particular Proximal Policy Optimization is leveraged. DRL has recently experienced a strong impetus from its successes applied to games, sometimes even reaching 'super-human' performances. This paper lays out the foundation of this method and proposes to study the behavior of several hyper-parameters that influence the DRL algorithm. The algorithm is highly dependent on multiple factors such as an exploration/exploitation trade-off that manifests through different parameters such as the number of loading patterns seen and the number of samples collected before a policy update, but also the shape of the objective function derived for the core design. Experimental results also demonstrate the effectiveness of the method in finding high-quality solutions from scratch within a reasonable amount of time. (authors)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Freely scalable and reconfigurable optical hardware for deep learning

Abstract As deep neural network (DNN) models grow ever-larger, they can achieve higher accuracy and solve more complex problems. This trend has been enabled by an increase in available compute power; however, efforts to continue to scale electronic processors are impeded by the costs of communication, thermal management, power delivery and clocking. To improve scalability, we propose a digital optical neural network (DONN) with intralayer optical interconnects and reconfigurable input values. The path-length-independence of optical energy consumption enables information locality between a transmitter and a large number of arbitrarily arranged receivers, which allows greater flexibility in architecture design to circumvent scaling limitations. In a proof-of-concept experiment, we demonstrate optical multicast in the classification of 500 MNIST images with a 3-layer, fully-connected network. We also analyze the energy consumption of the DONN and find that digital optical data transfer is beneficial over electronics when the spacing of computational units is on the order of $$>10\,\upmu $$ > 10 μ m.

42 ENGINEERING↗

Spatial Transferability of Machine Learning Based Volume Estimation Models

High-quality traffic volume data is essential for efficient transportation planning and operations. However, such high-quality data is expensive to collect, owing primarily to the high capital cost of installing and maintaining continuous counting stations (CCSs). Recent availability of probe-based vehicle data offers a cost-effective solution for increasing the observability of traffic volumes. However, having ample ground truth traffic data is a prerequisite for developing robust volume estimation models. Though this might not be a big issue in many states, states with scarce CCS data might be able to benefit from robust volume estimation models developed in (adjacent) data-rich states. While there is a reasonable amount of spatial transferability research in the transportation domain, there is a dearth of knowledge on the spatial transferability of probe-based volume estimation models. To address this gap, this paper explores spatial transferability of volume estimation models developed from data in three states (Colorado, North Carolina, and Pennsylvania). Results indicate that it is extremely important to maintain temporal consistency when attempting spatial transferability of volume estimation models. It was also found that models trained on regions with lower peak traffic volumes will limit the performance of models transferred to states with higher peak hourly traffic volumes. Corroborating findings from existing spatial transferability research on other topics, it was found that a meta-model (developed using data from multiple states) performs better than volume estimation models developed within any one of the states.

ADVANCED PROPULSION SYSTEMS↗

An Approach to Bayesian Optimization for Design Feasibility Check on Discontinuous Black-Box Functions

The paper presents a novel approach to applying Bayesian Optimization (BO) in predicting an unknown constraint boundary, also representing the discontinuity of an unknown function, for a feasibility check on the design space, thereby representing a classification tool to discern between a feasible and infeasible region. Bayesian optimization is a low-cost black-box global optimization tool in the Sequential Design Methods where one learns and updates knowledge from prior evaluated designs, and proceeds to the selection of new designs for future evaluation. However, BO is best suited to problems with the assumption of a continuous objective function and does not guarantee true convergence when having a discontinuous design space. This is because of the insufficient knowledge of the BO about the nature of the discontinuity of the unknown true function. In this paper, we have proposed to predict the location of the discontinuity using a BO algorithm on an artificially projected continuous design space from the original discontinuous design space. The proposed approach has been implemented in a thin tube design with the risk of creep-fatigue failure under constant loading of temperature and pressure. The stated risk depends on the location of the designs in terms of safe and unsafe regions, where the discontinuities lie at the transition between those regions; therefore, the discontinuity has also been treated as an unknown creep-fatigue failure constraint. The proposed BO algorithm has been trained to maximize sampling toward the unknown transition region, to act as a high accuracy classifier between safe and unsafe designs with minimal training cost. The converged solution has been validated for different design parameters with classification error rate and function evaluations at an average of <1% and ~150, respectively. Finally, the performance of our proposed approach in terms of training cost and classification accuracy of thin tube design is shown to be better than the existing machine learning (ML) algorithms such as Support Vector Machine (SVM), Random Forest (RF), and Boosting.

Engineering↗

Genetic polymorphisms of GRIN2A and GRIN2B modify the neurobehavioral effects of low-level lead exposure in children

Highlights: • Genetic polymorphisms may increase adverse effects of Pb on neurobehavioral processes. • We studied 330 participants of the Casa Pia Dental Amalgams Trial in Children. • Adverse interactions were found with variants of GRIN2A and GRIN2B in boys. • Our results implicate genetic factors in defining susceptibility to Pb neurotoxicity. Lead (Pb) is neurotoxic and children are highly susceptible to this effect, particularly within the context of continuous low-level Pb exposure. A current major challenge is identification of children who may be uniquely susceptible to Pb toxicity because of genetic predisposition. Learning and memory are among the neurobehavioral processes that are most notably affected by Pb exposure, and modification of N-methyl-D-aspartate receptors (NMDAR) that regulate these processes during development are postulated to underlie these adverse effects of Pb. We examined the hypothesis that polymorphic variants of genes encoding glutamate receptor, ionotropic, NMDAR subunits 2A and 2B, GRIN2A and GRIN2B, exacerbate the adverse effects of Pb exposure on these processes in children. Participants were subjects who participated as children in the Casa Pia Dental Amalgam Clinical Trial and for whom baseline blood Pb concentrations and annual neurobehavioral test results over the 7 year course of the clinical trial were available. Genotyping assays were performed for variants of GRIN2A (rs727605 and rs1070503) and GRIN2B (rs7301328 and rs1806201) on biological samples acquired from 330 of the original 507 trial participants. Regression modeling strategies were employed to evaluate the association between genotype status, Pb exposure, and neurobehavioral test outcomes. Numerous significant adverse interaction effects between variants of both GRIN2A and GRIN2B, individually and in combination, and Pb exposure were observed particularly among boys, preferentially within the domains of Learning & Memory and Executive Function. In contrast, very few interaction effects were observed among similarly genotyped girls with comparable Pb exposure. These findings support observations of an essential role of GRIN2A and GRIN2B on developmental processes underlying learning and memory as well as other neurological functions in children and demonstrate, further, modification of Pb effects on these processes by specific variants of both GRIN2A and GRIN2B genes. These observations highlight the importance of genetic factors in defining susceptibility to Pb neurotoxicity and may have important public health implications for future strategies aimed at protecting children and adolescents from potential health risks associated with low-level Pb exposure.

60 APPLIED LIFE SCIENCES↗

Artificial Intelligence for Conjugated Polymers

Conjugated polymers have garnered significant attention due to their diverse applications in electronics, photonics, and energy storage. However, realizing their full potential poses a formidable challenge, as their design has historically relied on iterative adjustments and continuous inspiration from researchers. Traditional methods often struggle to efficiently navigate their vast chemical landscape. In this work, the application of artificial intelligence (AI), specifically machine learning (ML), needs to be discussed in the realm of conjugated polymers. Our paper emphasizes the importance of understanding the structure–property relationships of these polymers and how ML can facilitate property prediction and inverse-design. We delve into various chemical fingerprints, structural descriptors, and ML algorithms, showcasing their utility across a spectrum of applications, including simulations, glass transition temperature determination, photovoltaics, reorganization energy for charge transport, photocatalysts, and sensors. Finally, we give some outlooks in this filed and propose unexplored areas within the field that hold the potential to benefit from ML techniques.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Bifunctional Electrocatalysts with High-Entropy Alloys: Bridging Hydrogen Evolution and Oxygen Reduction

High-entropy alloys (HEAs) have emerged as a promising class of bifunctional electrocatalysts capable of simultaneously driving the hydrogen evolution reaction (HER) and the oxygen reduction reaction (ORR) with high activity and durability. Their near-equiatomic multicomponent compositions give rise to unique physicochemical characteristics, including lattice distortion, sluggish diffusion, high-entropy stabilization, and pronounced electronic heterogeneity, that collectively generate diverse and synergistic active sites inaccessible in conventional alloys. This review summarizes recent progress in HEA-based bifunctional electrocatalysis, with a focus on the fundamental mechanisms governing HER and ORR activity, stability, and selectivity. We discuss advances in synthesis strategies, ranging from confined growth and step-alloying to scalable continuous-flow methods, that enable precise control over composition, size, and surface structure. Complementary computational and data-driven approaches, including density functional theory, machine-learning-assisted screening, and descriptor development, are highlighted as essential tools for navigating the vast HEA design space and establishing structure−property relationships. Particular attention is paid to adsorption-energy distributions, multisite cooperativity, and environmental effects under realistic electrochemical conditions. Finally, we outline current challenges and future opportunities for integrating mechanistic understanding with AI-guided, closed-loop design frameworks to accelerate the discovery of next-generation HEA bifunctional electrocatalysts for sustainable energy conversion.

Alloys↗

Stochastic representation and conditioning of process-based geological model by deep generative and recognition networks

Accurate and realistic geological modeling is the core of oil and gas development and production. In recent years, process-based methods are developed to produce highly realistic geological models by simulating the physical processes that reproduce the sedimentary events and develop the geometry. However, the complex dynamic processes are extremely expensive to simulate, making process-based models difficult to be conditioned to field data. In this work, we propose a comprehensive generative adversarial network framework as a machine-learning-assisted approach for mimicking the outputs of process-based geological models with fast generation. The main objective of our work is to obtain a continuous parametrization of the highly realistic process-based geological models which enables us to calibrate the models and condition the models to data. Numerical results are presented to illustrate the capability of our proposed methodology.

58 GEOSCIENCES↗

Carbon Fiber Prepreg Recycling- Automated Perform Manufacturing Equipment (Final Technical Report)

Discontinuous fiber composites have a compelling business case as related to light metals such as aluminum and magnesium machined or die cast components. DFC’s have been in commercial production using carbon fiber/epoxy prepreg for both commercial aircraft and for military applications for a number of years and their usage is expanding into automotive and consumer goods components. Reasons for this include significantly lower cost as compared to continuous fiber composites easier design and understanding of the material by metal traditionalists, higher degree of part complexity with relatively simple and low cost molds when compared to high-pressure die casting, and potential for near zero waste. IACMI’s 10-year goals of reduction in carbon fiber reinforced plastics manufacturing cost by 25%, reduction in their embodied energy by 50%, and recyclability of 95% can all be advanced through robust recycling. With recycling We have shown that a 15% contribution in manufacturing cost reduction is feasible, that a significant reduction in embodied energy (over 90% for the recycled fraction) is achievable, and that when combined with primary pyrolysis recycling, the 95% target can be met. Large-scale application of advanced composites began with the aerospace industry, and CRTC’s location in Washington State provided access to a waste stream of nearly 900 tonnes/year that was going to landfill. As the aerospace industry has automated parts production, their material forms have become significantly easier to recycle, and some evidence of this exists for the future automotive and wind turbine components as well. CRTC decided to focus on re-purposing the carbon fiber/epoxy aerospace scrap streams through materials reformatting and reuse, and not on primary recycling (pyrolysis mainly) as this technology is commercially available. The main reasons for this were that it offers a route to effective use of the most widely available and consistent feedstock called edge-trim, as well as allows creation of complex and low-cost parts using rapid and cost-effective compression molding technologies. The reasons also include leading the way for potential automotive pre-preg recycling as the combination of roughly 85% virgin pre-preg and 15% recycled pre-preg directly supports IACMI’s goals and enables significant part cost reduction and a path to zero-waste composite manufacturing at the OEMs or Tier 1’s. The purpose of the project was to develop equipment and technology to transform continuous fiber pre-preg in roll forms into smaller chip-format DFC feedstock, and to determine impacts on moldability and performance based on lessons learned along the path. The availability of scrap from pre-preg manufacturers was a significant value to the program even though the aerospace pre-preg is designed to be autoclave molded thus has handling, tackiness, and processing factors that are not ideal for DFCs. The scope of the project was to develop a Phase 1 system that could identify and resolve some of the major hurdles, determine where to focus effort on building a commercial-scale Phase 2 system, and develop technology for handling a wide variety of incoming scrap materials to produce the most usable product. A chopping system was developed and reduced to practice via an experimental approach that enabled all forms of scrap coming from the pre-preg manufacturers and some forms of the scrap coming from OEM production lines to be very effectively re-purposed. The systems that were developed were not considered to be production robust but rather identified and resolved some of the major technical hurdles, and demonstrated the path towards effective full-scale commercialization.

36 MATERIALS SCIENCE↗

Building Datasets and Training Methods for ML Based Magnet Quench Detection

Detecting quenches in superconducting (SC) magnets during training is a challenging process that involves capturing physical events that occur at different frequencies and appear as various signal features. These events may be correlated across instrumentation type, thermal cycle, and ramp. These events together build a more complete picture of continuous processes occurring in the magnet, and may allow us to flag potential precursors for quench detection. We present our work on building an automatic machine learning (ML) based quench detection system. We build upon our existing work on unsupervised auto-encoders for acoustic sensors and quench antenna (QA) by first establishing a supervised ML training pipeline. We show the results of an event tagging, analysis, and simulation framework on our QA and acoustic data which are used concurrently to build a training dataset for a supervised implementation. We then show how this supervised training can be used as a prior in a semi-supervised framework and compare this to the unsupervised neural network auto-encoder performance.This allows us to have a more concrete understanding of the performance of our algorithms relative to physical events occurring in the magnet, and also provides a baseline software tool to generically evaluate our quench prediction autoencoders under completely unsupervised, supervised, and semi-supervised training conditions.

Khan, Maira [Fermilab]↗

MLCommons Science Benchmarks

Benchmarks are a cornerstone of modern machine learning practice, providing standardized eval- uations that enable reproducibility, comparison, and scientific progress. Yet, as AI systems particularly deep learning models become increasingly dynamic, traditional static benchmarking approaches are losing their relevance. Models rapidly evolve in architecture, scale, and capability; datasets shift; and deployment contexts continuously change, creating a moving target for evaluation. Without adaptive benchmarking frame- works, both scientific assessment and real-world de- ployment risk becoming misaligned with actual system behavior. Drawing on our experience from MLCommons, educa- tional initiatives, and government programs such as the DOE s Million Parameter Consortium, we identify key barriers that hinder the broader adoption and utility of benchmarking in AI. These include substantial resource demands, limited access to specialized hardware, lack of expertise in benchmark design, and uncertainty among practitioners about how to relate benchmark results to their own application domains. Moreover, current benchmarks often emphasize peak performance on leadership-class hardware, offering limited guidance for more diverse, real-world deployment scenarios. We argue that benchmarking itself must become dy- namic in order to incorporate evolving models, updated data, and heterogeneous computational platforms while maintaining transparency, reproducibility, and inter- pretability. Democratizing this process requires not only technical innovation, but also systematic educational efforts spanning undergraduate to professional levels to develop sustained expertise in benchmark design and use. Finally, benchmarks should be framed and com- municated to support application-relevant comparisons, enabling both developers and users to make informed, context-sensitive decisions. Advancing dynamic and inclusive benchmarking practices will be essential to ensure that evaluation keeps pace with the evolving AI landscape and supports responsible, reproducible, and accessible AI deployment.

Hawks, Benjamin G. [Fermilab]↗

Meteorological and dynamical requirements for MST radar networks: Waves

Studies of wave motions using the MST radar have concentrated on single station time series analyses of gravity waves and tides. Since these radars collect high time resolution data they have the potential to become a significant tool for mesoscale research. In addition, radars are operated almost continuously unattended and, consequently, data sets are available for analyzing longer period wave motions such as tides and planetary scale waves. Although there is much to learn from single station data, the possibilities of new knowledge from a network of radars is exciting. The scales of wave motions in the atmosphere cover a broad range. Consequently the choice of a radar network depends to a large extent on the types of wave motions that are studied. There are many outstanding research problems that would benefit from observations from a MST radar network. In particular, there is a strong need for measurements of gravity wave parameters and equatorial wave motions. Some of the current problems in wave dynamics are discussed.

Avery, S. K.↗

The Jupiter-Io connection - An Alfven engine in space

Much has been learned about the electromagnetic interaction between Jupiter and its satellite Io from in situ observations. Io, in its motion through the Io plasma torus at Jupiter, continuously generates an Alfven wing that carries two billion kilowatts of power into the jovian ionosphere. Concurrently, Io is acted upon by a J x B force tending to propel it out of the jovian system. The energy source for these processes is the rotation of Jupiter. This unusual planet-satellite coupling serves as an archetype for the interaction of a large moving conductor with a magnetized plasma, a problem of general space and astrophysical interest.

Belcher, John W.↗

Standards Advisor-Advanced Information Technology for Advanced Information Delivery

Developers of space systems must deal with an increasing amount of information in responding to extensive requirements and standards from numerous sources. Accessing these requirements and standards, understanding them, comparing them, negotiating them and responding to them is often an overwhelming task. There are resources to aid the space systems developer, such as lessons learned and best practices. Again, though, accessing, understanding, and using this information is often more difficult than helpful. This results in space systems that: 1. Do not meet all their requirements. 2. Do not incorporate prior engineering experience. 3. Cost more to develop. 4. Take longer to develop. The NASA Technical Standards Program (NTSP) web site at http://standards.nasa.gov has made significant improvements in making standards, lessons learned, and related material available to space systems developers agency-wide. The Standards Advisor was conceived to take the next steps beyond the current product, continuing to apply evolving information technology that continues to improve information delivery to space systems developers. This report describes the features of the Standards Advisor and suggests a technical approach to its development.

Hawker, J. Scott↗

MOPITT Mechanisms 16 Years In-Orbit Operation on TERRA

The 16th anniversary of the launch of NASA's Terra Spacecraft was marked on December 18, 2015, with the Measurements of Pollution in the Troposphere (MOPITT) instrument being a successful contributor to the NASA EOS flagship. MOPITT has been enabled by a large suite of mechanisms, allowing the instrument to perform long-duration monitoring of atmospheric carbon monoxide, providing global measurements of this important greenhouse gas for 16 years. Mechanisms have been successfully employed for scanning, cooling of detectors, and to optically modulate the gas path length within the instrument by means of pressure and gas cell length variation. The instrument utilizes these devices to perform correlation spectroscopy, enabling measurements with vertical resolution from the nadir view, and has thereby furthered understanding of source and global transport effects of carbon monoxide. Given the design requirement for a 5.25-year lifetime, the stability and performance of the majority of mechanisms have far surpassed design goals. With 16 continuously operating mechanisms in service on MOPITT, including 12 rotating mechanisms and 4 with linear drive elements, the instrument was an ambitious undertaking. The long life requirements combined with demands for cleanliness and optical stability made for difficult design choices including that of the selection of new lubrication processes. Observations and lessons learned with regards to many aspects of the mechanisms and associated monitoring devices are discussed here. Mechanism behaviors are described, including anomalies, long-term drive current/power, fill pressure, vibration and cold-tip temperature trends. The effectiveness of particular lubrication formulations and the screening method implemented is discussed in relation to continuous rotating mechanisms and stepper motors, which have exceeded 15 billon rotations and 2.5 billion steps respectively. Aspects of gas cell hermeticity, optical cleanliness, heater problems and SEU effects on accelerometers are also discussed.

Gibson, Andrew S.↗

On-Orbit Performance of MODIS Solar Diffuser Stability Monitor

Moderate Resolution Imaging Spectroradiometer (MODIS) is currently operated on both the Terra and Aqua spacecraft. It collects data in 20 reflective solar bands (RSB) and 16 thermal emissive bands. MODIS RSB calibration is reflectance based via an on-board solar diffuser (SD). On-orbit changes in the SD bidirectional reflectance factor are tracked by an onboard solar diffuser stability monitor (SDSM). The SDSM functions as an independent ratioing radiometer with nine filtered detectors, covering wavelengths in the visible (VIS) and near-infrared (NIR) spectral regions. A brief overview of SDSM design functions, on-orbit operations, and performance for both Terra and Aqua MODIS is provided. In addition to the SD on-orbit degradation at different wavelengths, the changes in SDSM detector responses and their potential impact on tracking SD on-orbit degradation are examined. After more than 12 years of on-orbit operation, Aqua MODIS SD has shown degradation varying from 0.6% at 0.94 microns to 19.0% at 0.41 microns. Due to more frequent solar exposure and longer operation time, the Terra MODIS SD has experienced a much larger degradation, varying from 2.3% at 0.94 microns to 48.0% at 0.41 microns. For both Terra and Aqua MODIS, the SD has experienced more degradation at shorter wavelengths. Meanwhile, the SDSM detector responses have also experienced wavelength-dependent degradation. The largest change in the SDSM detector responses, however, occurred at longer NIR wavelengths. Since launch, the SDSM systems on both Terra and Aqua MODIS have continued their nominal operations, enabling critical parameters to be derived in support of the RSB on-orbit calibration. The calibration strategies developed for and lessons learned from MODIS SDSM operations, and a preliminary performance comparison with the S-NPP VIIRS SDSM are discussed.

MODIS↗

Mapping National Forest Aboveground Biomass in Mexico By Integrating GEDI and Landsat Times Series Data

Mexico is one of the countries with great potential for the UN's Reducing Emissions from Deforestation and Forest Degradation (REDD+) program, a key nature-based solution for the forest sector. To monitor carbon stock changes, there is a growing demand for unbiased Monitoring Reporting Verification (MRV) systems to facilitate effective forest management and climate change mitigation strategies. Remote sensing-based national aboveground biomass density (AGBD) estimation over Mexico is scarce and often limited to one-time static mapping, leading to spatiotemporal inconsistency in inputs. As an effort under NASA's Carbon Monitoring System (CMS) program, we have developed a remote sensing-based approach to create consistent historical AGBD maps of Mexico using multi-stream remote sensing data, including spaceborne lidar GEDI and long-term Landsat time series, as well as topographic information. We employ the continuous change detection and classification (CCDC) algorithm for temporal modeling of Landsat surface reflectance, followed by the inference of forest AGBD using a random forest machine learning algorithm with the temporal information of land surface dynamics extracted by the CCDC as input. GEDI provides unprecedented forest structure and AGBD sampling datasets for model training and validation practices. In this presentation, we share the progress made in developing a spatially explicit mapping of historical AGBD changes associated with land surface changes and post-disturbance landscapes.

Taejin Park↗

Blackout Diffusion: Generative Diffusion Models in Discrete-State Spaces

Typical generative diffusion models rely on a Gaussian diffusion process for training the backward transformations, which can then be used to generate samples from Gaussian noise. However, real world data often takes place in discrete-state spaces, including many scientific applications. Here, we develop a theoretical formulation for arbitrary discrete-state Markov processes in the forward diffusion process using exact (as opposed to variational) analysis. We relate the theory to the existing continuous-state Gaussian diffusion as well as other approaches to discrete diffusion, and identify the corresponding reverse-time stochastic process and score function in the continuous-time setting, and the reverse-time mapping in the discrete-time setting. As an example of this framework, we introduce “Blackout Diffusion”, which learns to produce samples from an empty image instead of from noise. Numerical experiments on the CIFAR-10, Binarized MNIST, and CelebA datasets confirm the feasibility of our approach. Generalizing from specific (Gaussian) forward processes to discrete-state processes without a variational approximation sheds light on how to interpret diffusion models, which we discuss.

Santos, Javier E.↗