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 541 records · Page 30

Applying Machine Learning to the Classification of DC-DC Converters. NA-22 Final Report

Tools are now available that enable measurement electromagnetic radiation (EMR) from active electronics in an item. This radiation may be intended WIFI or cellular network links, for example or unintended such as the switching noise generated by DC-to-DC converters. It would be extremely valuable to have the capability to discriminate between the low-voltage DC-to-DC converters or other digital noise prevalent in most modern electronics, versus the high-voltage DC-to-DC converters used in utility firesets. Previous work performed under a Sandia Laboratory Directed Research and Development (LDRD) project on Charge State Detection using a deep neural network has been continued in this effort. A state-of-the-art supervised machine learning algorithm has not only been extended to discriminate between low and high voltage converters but has been validated in determining a converters make and model.

42 ENGINEERING↗

Continuous conditional generative adversarial networks for data-driven solutions of poroelasticity with heterogeneous material properties

Machine learning-based data-driven modeling can allow computationally efficient time-dependent solutions of PDEs, such as those that describe subsurface multiphysical problems. In this work, our previous approach (Kadeethum et al., 2021d) of conditional generative adversarial networks (cGAN) developed for the solution of steady-state problems involving highly heterogeneous material properties is extended to time-dependent problems by adopting the concept of continuous cGAN (CcGAN). The CcGAN that can condition continuous variables is developed to incorporate the time domain through either element-wise addition or conditional batch normalization. Moreover, this framework can handle training data that contain different timestamps and then predict timestamps that do not exist in the training data. As a numerical example, the transient response of the coupled poroelastic process is studied in two different permeability fields: Zinn & Harvey transformation and a bimodal transformation. The proposed CcGAN uses heterogeneous permeability fields as input parameters while pressure and displacement fields over time are model output. Our results show that the model provides sufficient accuracy with computational speed-up. This robust framework will enable us to perform real-time reservoir management and robust uncertainty quantification in poroelastic problems.

97 MATHEMATICS AND COMPUTING↗

Integrity Assessment of Hanford 242-A Evaporator Facility E-A-1 Re-boiler Vessel via Fluorescein Liquid Tracer Test Methods - 17357

The 242-A Evaporator Facility is critical to the United States Department of Energy's (DOE) Hanford Site clean-up mission due to limited space within the existing Double-Shell Tanks (DSTs) used to store nuclear and chemical waste until future processing. The 242-A Evaporator Facility is currently the sole waste volume reduction process used at Hanford, and the E-A-1 Re-boiler is essential to the Evaporator process. Due to the potential for cross-contamination between the Re-boiler tubes through which waste passes and the tube sheets of uncontaminated steam, an assessment of the Re-boiler vessel's integrity is required for continuing operations of the facility. In order to meet these requirements, an independent integrity assessment was designed to use fluorescein liquid tracer (LT) to verify the leak-tightness of the Re-boiler vessel in accordance with Washington River Protection Solutions, LLC (WRPS) tightness test standards, as well as, the ASME Boiler and Pressure Vessel (B and PV) Code, Section III, Division 1, NB-6300. LT test methodologies were evaluated and determined to be appropriate for a vessel with restricted access, such as the E-A-1 Re-boiler, as it provided the needed sensitivity, reliability, and cost; as well as, exceeded the leak-tightness specifications required by both WRPS and ASME B and PV Code standards. The integrity test requires water laced with fluorescein be circulated through the waste processing loop of the 242-A Evaporator and samples of steam condensate be collected and analyzed for traces of fluorescein to determine leak-tightness. Liquid condensate sampling was conducted prior to and during testing and the samples were analyzed for fluorescein tracer to determine the amount of fluorescein that flowed out of the tubes of the Re-boiler into the steam side of the vessel. Given the established criteria for 'leak-tight,' test results determined that the Re-boiler passed the integrity assessment. After further scrutiny of the 2014 test and data results, it was determined that although the Re-boiler was 'leak-tight,' as established by the test criteria, there was concern of future leakage. The DOE and the DOE Office of River Protection (ORP) requested that WRPS repeat the integrity assessment in two and a half years to determine if the potential leak was growing, which could threaten continuing operations of the 242-A Evaporator. The 2017 test has been redesigned using lessons learned from the 2014 test and was modified to account for ergonomic and safety considerations, ease of repeatability, and accuracy of results. The liquid tracer methodology and test plan developed for the Re-boiler will continue to be implemented in 5 year increments to monitor current and future integrity. The method may also prove useful for other aging components across the Hanford Site, where appropriate. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Oracle-Preserving Latent Flows

A fundamental task in data science is the discovery, description, and identification of any symmetries present in the data. We developed a deep learning methodology for the simultaneous discovery of multiple non-trivial continuous symmetries across an entire labeled dataset. The symmetry transformations and the corresponding generators are modeled with fully connected neural networks trained with a specially constructed loss function, ensuring the desired symmetry properties. The two new elements in this work are the use of a reduced-dimensionality latent space and the generalization to invariant transformations with respect to high-dimensional oracles. The method is demonstrated with several examples on the MNIST digit dataset, where the oracle is provided by the 10-dimensional vector of logits of a trained classifier. We find classes of symmetries that transform each image from the dataset into new synthetic images while conserving the values of the logits. We illustrate these transformations as lines of equal probability (“flows”) in the reduced latent space. These results show that symmetries in the data can be successfully searched for and identified as interpretable non-trivial transformations in the equivalent latent space.

97 MATHEMATICS AND COMPUTING↗

Superconducting Radio-frequency Cavity Fault Classification Using Machine Learning at Jefferson Laboratory

We report on the development of machine learning models for classifying C100 superconducting radiofrequency (SRF) cavity faults in the Continuous Electron Beam Accelerator Facility (CEBAF) at Jefferson Lab. Of the 418 SRF cavities in CEBAF, 96 are designed with a digital low-level RF system configured such that a cavity fault triggers recordings of RF signals for each of eight cavities in the cryomodule. Subject matter experts analyze the collected time-series data and identify which of the eight cavities faulted first and classify the type of fault. This information is used to find trends and strategically deploy mitigations to problematic cryomodules. However, manually labeling the data is laborious and time-consuming. By leveraging machine learning, near real-time - rather than postmortem - identification of the offending cavity and classification of the fault type has been implemented. We discuss the performance of the machine learning models during a recent physics run. We also discuss efforts for further insights into fault types through unsupervised learning techniques and present preliminary work on cavity and fault prediction using data collected prior to a failure event.

Tennant, C. D.↗

Integrating very-high-resolution imagery, Sentinel-2 time-series data, and machine learning to map shrub fractional abundance across arid and semi-arid ecosystems in China

Shrub fractional abundance (SFA), the proportion of shrub cover per unit area, serves as a critical indicator of environmental aridity and ecosystem health in arid and semi-arid regions, particularly across the Mongolian steppe. However, large-scale SFA mapping in Mongolian steppe ecosystems remains challenging due to the small crown size of shrubs, their sparse distribution, and spectral overlap with coexisting low vegetation (e.g., grasses and herbs), which hinders accurate detection using coarser-resolution satellite data or traditional field surveys. To address these challenges, we developed a two-step approach that integrates very-high-resolution (VHR) imagery, time-series Sentinel-2 data, and deep learning techniques. First, we generated high-accuracy benchmark maps of individual shrub crowns from 0.5 m VHR imagery by combining manual segmentation with a hybrid deep learning framework (Dino V2 and convolutional neural networks). Second, we used these shrub crown maps as training data to build an XGBoost model for predicting SFA from 20 m Sentinel-2 time-series data, leveraging phenological information to improve estimation. We validated our approach across 70 sites (1km 2 each) in the Inner Mongolia Autonomous Region, which is representative of Mongolian steppe ecosystems. From VHR imagery, we mapped 1.31 million shrub crowns with an accuracy of R 2 = 0.92. Scaling up with Sentinel-2 data yielded regional SFA maps with an R 2 = 0.60. Further SHAP (SHapley Additive exPlanations) analysis on the developed XGBoost model revealed that phenological metrics (particularly observations in early-May, mid-July, and late-September), which distinguish shrub phenology from that of other land cover types (e.g., grasses and bare soil), were the most influential predictors of SFA. Finally, our regional SFA maps uncovered unimodal relationships between shrub distribution and climate variables, peaking at mean annual minimum temperatures near 0 °C and annual precipitation around 200 mm. Collectively, these findings demonstrate how the integration of multi-source remote sensing and machine learning can overcome historical limitations in SFA mapping, enabling accurate, spatially continuous assessments across vast Inner-Mongolian steppe ecosystems. Our framework has the potential to be applied to other steppe ecosystems and dryland ecosystems across the Mongolian steppe and beyond, offering a foundation for improved monitoring and ecological impact assessments in the face of global climate changes.

Arid and semi-arid landscapes↗

Developing a Deep Learning-Computer Vision Framework to Monitor Avian Interactions with Solar Energy Facility Infrastructure (Final Technical Report)

The project addressed an inability to monitor avian interactions with photovoltaic (PV) solar energy facilities necessary for understanding PV solar impacts on birds. In the project, machine-vision technology that continuously monitors avian activities at PV solar facilities was developed. The technology includes four machine-learning (ML) models, each of which accomplishes a specific task in detecting birds and classifying their activities in live or recorded videos—detecting and tracking moving objects, differentiating birds from other objects, detecting bird collisions with solar panels, and classifying non-collision bird activities around PV facilities. Major project outcomes include adoption by two of DOE SETO’s SolWEB projects, providing novel observational data on birds to promote co-location of PV solar development and habitat conservation, known as ecovoltaics.

14 SOLAR ENERGY↗

DeepCMB: Lensing reconstruction of the cosmic microwave background with deep neural networks

Next-generation cosmic microwave background (CMB) experiments will have lower noise and therefore increased sensitivity, enabling improved constraints on fundamental physics parameters such as the sum of neutrino masses and the tensor-to-scalar ratio r. Achieving competitive constraints on these parameters requires high signal-to-noise extraction of the projected gravitational potential from the CMB maps. Standard methods for reconstructing the lensing potential employ the quadratic estimator (QE). However, the QE performs suboptimally at the low noise levels expected in upcoming experiments. Other methods, like maximum likelihood estimators (MLE), are under active development. In this work, we demonstrate reconstruction of the CMB lensing potential with deep convolutional neural networks (CNN) - ie, a ResUNet. The network is trained and tested on simulated data, and otherwise has no physical parametrization related to the physical processes of the CMB and gravitational lensing. We show that, over a wide range of angular scales, ResUNets recover the input gravitational potential with a higher signal-to-noise ratio than the QE method, reaching levels comparable to analytic approximations of MLE methods. We demonstrate that the network outputs quantifiably different lensing maps when given input CMB maps generated with different cosmologies. We also show we can use the reconstructed lensing map for cosmological parameter estimation. This application of CNN provides a few innovations at the intersection of cosmology and machine learning. First, while training and regressing on images, we predict a continuous-variable field rather than discrete classes. Second, we are able to establish uncertainty measures for the network output that are analogous to standard methods. We expect this approach to excel in capturing hard-to-model non-Gaussian astrophysical foreground and noise contributions.

79 ASTRONOMY AND ASTROPHYSICS↗

Weak-form inference for hybrid dynamical systems in ecology

Species subject to predation and environmental threats commonly exhibit variable periods of population boom and bust over long timescales. Understanding and predicting such behaviour, especially given the inherent heterogeneity and stochasticity of exogenous driving factors over short timescales, is an ongoing challenge. A modelling paradigm gaining popularity in the ecological sciences for such multi-scale effects is to couple short-term continuous dynamics to long-term discrete updates. We develop a data-driven method utilizing weak-form equation learning to extract such hybrid governing equations for population dynamics and to estimate the requisite parameters using sparse intermittent measurements of the discrete and continuous variables. The method produces a set of short-term continuous dynamical system equations parametrized by long-term variables, and long-term discrete equations parametrized by short-term variables, allowing direct assessment of interdependencies between the two timescales. We demonstrate the utility of the method on a variety of ecological scenarios and provide extensive tests using models previously derived for epizootics experienced by the North American spongy moth ( Lymantria dispar dispar ).

54 ENVIRONMENTAL SCIENCES↗

Development of algorithms for augmenting and replacing conventional process control using reinforcement learning

Here, this work seeks to allow for the online operation and training of model-free reinforcement learning (RL) agents but limit the risk to system equipment and personnel. The parallel implementation of RL alongside more conventional process control (CPC) allows for the RL algorithm to learn from CPC. The past performance of both methods are assessed on a continuous basis allowing for a transition from CPC to RL and, if needed, transitioning back to CPC from RL. This allows for the RL algorithm to slowly and safely assume control of the process without significant degradation in control performance. It is shown that the RL can derive a near optimal policy even when coupled with a suboptimal CPC. It is also demonstrated that the coupled RL-CPC algorithm learns at a faster rate than traditional RL methods of exploration while the algorithm’s performance does not deteriorate below CPC, even when exposed to an unknown operating condition.

30 DIRECT ENERGY CONVERSION↗

Quantifying the generalization error in deep learning in terms of data distribution and neural network smoothness

We report the accuracy of deep learning, i.e., deep neural networks, can be characterized by dividing the total error into three main types: approximation error, optimization error, and generalization error. Whereas there are some satisfactory answers to the problems of approximation and optimization, much less is known about the theory of generalization. Most existing theoretical works for generalization fail to explain the performance of neural networks in practice. To derive a meaningful bound, we study the generalization error of neural networks for classification problems in terms of data distribution and neural network smoothness. We introduce the cover complexity (CC) to measure the difficulty of learning a data set and the inverse of the modulus of continuity to quantify neural network smoothness. A quantitative bound for expected accuracy/error is derived by considering both the CC and neural network smoothness. Although most of the analysis is general and not specific to neural networks, we validate our theoretical assumptions and results numerically for neural networks by several data sets of images. The numerical results confirm that the expected error of trained networks scaled with the square root of the number of classes has a linear relationship with respect to the CC. We also observe a clear consistency between test loss and neural network smoothness during the training process. In addition, we demonstrate empirically that the neural network smoothness decreases when the network size increases whereas the smoothness is insensitive to training dataset size.

97 MATHEMATICS AND COMPUTING↗

Identifying Different Classes of Seismic Noise Signals Using Unsupervised Learning

Abstract Proper classification of nontectonic seismic signals is critical for detecting microearthquakes and developing an improved understanding of ongoing weak ground motions. We use unsupervised machine learning to label five classes of nonstationary seismic noise common in continuous waveforms. Temporal and spectral features describing the data are clustered to identify separable types of emergent and impulsive waveforms. The trained clustering model is used to classify every 1 s of continuous seismic records from a dense seismic array with 10–30 m station spacing. We show that dominate noise signals can be highly localized and vary on length scales of hundreds of meters. The methodology demonstrates the complexity of weak ground motions and improves the standard of analyzing seismic waveforms with a low signal‐to‐noise ratio. Application of this technique will improve the ability to detect genuine microseismic events in noisy environments where seismic sensors record earthquake‐like signals originating from nontectonic sources.

Johnson, Christopher W.↗

3D Printing at Los Alamos

At Los Alamos National Laboratory (LANL), their additive manufacturing division is seeking to answer the question whether it is possible to 3D print on a 5th axis. The end goal is to achieve true continuous fiber 3D printing. My goal on this project was to learn about additive manufacturing, and to learn how to operate the experimental 3D printer. Each print that was done demonstrated a different pattern that would change the strength and integrity of the print. Overall prints with minimal strand separation and featured weave-like patterns proved to have the best integrity.

42 ENGINEERING↗

Bioinformatic Teaching Resources – For Educators, by Educators – Using KBase, a Free, User-Friendly, Open Source Platform

Over the past year, biology educators and staff at the U.S. Department of Energy Systems Biology Knowledgebase (KBase) initiated a collaborative effort to develop a curriculum for bioinformatics education. KBase is a free web-based platform where anyone can conduct sophisticated and reproducible bioinformatic analyses via a graphical user interface. Here, we demonstrate the utility of KBase as a platform for bioinformatics education, and present a set of modular, adaptable, and customizable instructional units for teaching concepts in Genomics, Metagenomics, Pangenomics, and Phylogenetics. Each module contains teaching resources, publicly available data, analysis tools, and Markdown capability, enabling instructors to modify the lesson as appropriate for their specific course. We present initial student survey data on the effectiveness of using KBase for teaching bioinformatic concepts, provide an example case study, and detail the utility of the platform from an instructor’s perspective. Even as in-person teaching returns, KBase will continue to work with instructors, supporting the development of new active learning curriculum modules. For anyone utilizing the platform, the growing KBase Educators Organization provides an educators network, accompanied by community-sourced guidelines, instructional templates, and peer support, for instructors wishing to use KBase within a classroom at any educational level–whether virtual or in-person.

59 BASIC BIOLOGICAL SCIENCES↗

Building brains for bodies

We describe a project to capitalize on newly available levels of computational resources in order to understand human cognition. We will build an integrated physical system including vision, sound input and output, and dextrous manipulation, all controlled by a continuously operating large scale parallel MIMD computer. The resulting system will learn to 'think' by building on its bodily experiences to accomplish progressively more abstract tasks. Past experience suggests that in attempting to build such an integrated system we will have to fundamentally change the way artificial intelligence, cognitive science, linguistics, and philosophy think about the organization of intelligence. We expect to be able to better reconcile the theories that will be developed with current work in neuroscience.

Brooks, Rodney Allen↗

Continual improvement: A bibliography with indexes, 1992-1993

This bibliography lists 606 references to reports and journal articles entered into the NASA Scientific and Technical Information Database during 1992 to 1993. Topics cover the philosophy and history of Continual Improvement (CI), basic approaches and strategies for implementation, and lessons learned from public and private sector models. Entries are arranged according to the following categories: Leadership for Quality, Information and Analysis, Strategic Planning for CI, Human Resources Utilization, Management of Process Quality, Supplier Quality, Assessing Results, Customer Focus and Satisfaction, TQM Tools and Philosophies, and Applications. Indexes include subject, personal author, corporate source, contract number, report number, and accession number.

Source record↗

Reformulating Non-Monotonic Theories for Inference and Updating

We aim to help build programs that do large-scale, expressive non-monotonic reasoning (NMR): especially, 'learning agents' that store, and revise, a body of conclusions while continually acquiring new, possibly defeasible, premise beliefs. Currently available procedures for forward inference and belief revision are exhaustive, and thus impractical: they compute the entire non-monotonic theory, then re-compute from scratch upon updating with new axioms. These methods are thus badly intractable. In most theories of interest, even backward reasoning is combinatoric (at least NP-hard). Here, we give theoretical results for prioritized circumscription that show how to reformulate default theories so as to make forward inference be selective, as well as concurrent; and to restrict belief revision to a part of the theory. We elaborate a detailed divide-and-conquer strategy. We develop concepts of structure in NM theories, by showing how to reformulate them in a particular fashion: to be conjunctively decomposed into a collection of smaller 'part' theories. We identify two well-behaved special cases that are easily recognized in terms of syntactic properties: disjoint appearances of predicates, and disjoint appearances of individuals (terms). As part of this, we also definitionally reformulate the global axioms, one by one, in addition to applying decomposition. We identify a broad class of prioritized default theories, generalizing default inheritance, for which our results especially bear fruit. For this asocially monadic class, decomposition permits reasoning to be localized to individuals (ground terms), and reduced to propositional. Our reformulation methods are implementable in polynomial time, and apply to several other NM formalisms beyond circumscription.

Grosof, Benjamin N.↗

Power Lander for Support of Long-Term Lunar Presence

Emerging industrial base and the consequent sustained manned Lunar presence will require consistent high power capacities. This paper proposes a first iteration design of a flyable electric power platform which could serve as an enabler of Lunar Development and Exploration. It is intended to support a small facility solo or an emerging industrial base as part of a grid. Lunar Missions, Habitats and Facilities stand to benefit from an expected decade of non-stop operation, the economics of scale, Commercial Off-The-Shelf (COTS) availability, standardization of design, and logistical support for Lunar encampments provided by this architecture. The unattended and unmanned vehicle design is to be man- and robotics-serviceable after delivery by current and proposed heavy-lift boosters. Design continuity within a family of systems will improve reliability through "lessons learned'' in the field. Further, various configurations of the proposed scalable architecture will provide reference platforms for the indigenous construction of similar power plant facilities from in-situ Lunar resources (ISRU). The baseline design should be directed towards those materials available on the Moon and expected to be manufacturable on-site within the first decade of operation.

Joyner, Russ↗