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 685 records · Page 38

Fault systems monitoring using machine learning

Traditional seismology methods are designed to build catalogs of seismic events that are obviously different from the remaining seismic data, even to the naked eye, and discard more than 99% of the data. Los Alamos National Lab researchers developed machine learning methods to examine massive amounts of raw seismic data. The scientists found that the continuous data discarded by cataloging methods are informative of the physical state of faults in theoretical models, in laboratory experiments, and in the field. This information may enable the detection of predictive markers of seismic events and the provide insights into the underlying physics to support earthquake hazard assessment.

47 OTHER INSTRUMENTATION↗

Lessons from Testing the First Three US HL-LHC Cryo-Assemblies at FNAL

The US High-Luminosity LHC Accelerator Upgrade Project (AUP) collaborates with CERN to deliver 10 cryo-assemblies for the High-Luminosity LHC upgrade at CERN. Those comprise the Q1/Q3 quadrupole optical elements in the accelerator and are based on Nb3Sn technology, for the first time in a large particle accelerator. So far three cryo-assemblies were tested at the Fermilab’s horizontal magnet test facility, specifically redesigned to serve those needs. This work gives an overview of the facility and its necessary evolution as the campaign progressed; main results and observations from the cryo-assembly tests are presented and discussed in the context of overall performance and expectations toward the completion of efforts. Lessons learned after multiple cryo-assembly tests by the same team are critical to ensure continuous success of the whole enterprise – those are openly shared.

Stoynev, Stoyan [Fermilab]↗

Learning from Inconsistency

This position paper argues that inconsistencies that occur during the development of a software specification offer an excellent way of learning more about the development process. We base this argument on our work on inconsistency management. Much attention has been devoted recently to the need to allow inconsistencies to occur during software development, to facilitate flexible development strategies, especially for collaborative work. Recent work has concentrated on reasoning in the presence of inconsistency, tracing inconsistencies with 'pollution markers' and supporting resolution. We argue here that one of the most important aspects of inconsistency is the learning opportunity it provides. We are therefore concerned with how to capture this learning outcome so that its significance is not lost. We present a small example of how apprentice software engineers learn from their mistakes, and outline how an inconsistency management tool could support this learning. We then argue that the approach can be used more generally as part of continuous process improvement.

Easterbrook, Steve↗

SONEX: NASA SASS Ozone and Nitrogen Oxide Experiment

This final report follows closely on our 1998 Annual Progress Report which was forwarded to NASA on November 19, 1998. Rather than reiterate the material therein we note here the continuation of the various items covered. SONEX has proN,Ided a number of opportunities to learn more about the fine-scale structure of the atmosphere. Coupled with the MOZAIC work on layers we can synthesize a new 3-D view of the fine scales that influence atmospheric chemistry. As a side issue we are also relating the features to clear air turbulence which would be, we think, a useful connection for commercial aviation.

Newell, Reginald E.↗

Cellular Mechanisms Underlying Bone-Forming Cell Proliferative Response to Hypergravity

Life on Earth has evolved under the continuous influence of gravity (1-g). As humans explore and develop space, however, we must learn to adapt to an environment with little or no gravity. Studies indicate that lack of weightbearing for vertebrates occurring with immobilization, paralysis, or in a microgravity environment may cause muscle and bone atrophy through cellular and subcellular level mechanisms. We hypothesize that gravity is needed for the efficient transduction of cell growth and survival signals from the extra-cellular matrix (ECM) (consisting of molecules such as collagen, fibronectin, and laminin) in mechanosensitive tissues. We test for the presence of gravity-sensitive pathways in bone-forming cells (osteoblasts) using hypergravity applied by a cell culture centrifuge. Stimulation of 50 times gravity (50-g) increased proliferation in primary rat osteoblasts for cells grown on collagen Type I and fibronectin, but not on laminin or uncoated surfaces. Survival was also enhanced during hypergravity stimulation by the presence of ECM. Bromodeoxyuridine incorporation in proliferating cells showed an increase in the number of actively dividing cells from about 60% at 1-g to over 90% at 25-g. Reverse transcription-polymerase chain reaction was used to test for all possible integrins. Our combined results indicate that beta1 and/or beta3 integrin subunits may be involved. These data indicate that gravity mechanostimulation of osteoblast proliferation involves specific matrix-integrin signalling pathways which are sensitive to g-level. Further research to define the mechanisms involved will provide direction so that we may better adapt and counteract bone atrophy caused by the lack of weightbearing.

Vercoutere, W.↗

Role of Cyber-Physical Testing in Developing Resilient Extraterrestrial Habitats

Extraterrestrial long-term habitat systems (henceforth referred to as habitat systems) require groundbreaking technological advances to overcome the extreme demands introduced by isolation and challenging environments. A habitat system must operate as intended under continuous disruptive conditions. Designing for the demands that challenging environments will place on habitat systems (e.g., wild temperature fluctuations, galactic cosmic rays, destructive dust, meteoroid impacts, vibrations, and solar particle events) represents one of the greatest challenges in this endeavor. This engineering problem necessitates that we design and manage habitat systems to be resilient. System resilience requires a comprehensive approach that accounts for disruptions through the design process and adapts to them in operation. As the habitat system evolves—growing in physical size, complexity, population, and connectivity—and diversifies in operations, it must continue to be safe and resilient. In this endeavor, we should take advantage of lessons learned in developing civil infrastructure responsive to catastrophic natural hazards, autonomous robotics platforms, smart buildings, cyber-physical testing, complex systems, and diagnostics and prognostics for intelligent health management. This study highlights the importance of system resilience and cyber-physical testing to address the grand challenge of developing habitat systems.

Public health and safety↗

System for Photogrammetric Imaging, Detection, and Ranging (SPIDR)

The System for Photogrammetric Imaging, Detection, and Ranging (SPIDR) project aims to advance the state-of-the-art (SOA) technology in camera tracking. SPIDR will advance this technology in the following ways: 1) perform autonomous tracking of rocket launches, 2) allow modularity for camera payloads to enable various types of imagery capture (standard video, high-speed, infrared, etc.), 3) enable additional imagery assets for increased scope of imagery analysis, and 4) serve as a viable tracking replacement for the existing Kineto Tracking Mount (KTM) system used by the Exploration Ground Systems (EGS) program. The project initially aimed to build an in-house mechanical design and control system, but realigned to a commercial off-the-shelf (COTS) mechanical design. During the FY23 CIF timeline, the SPIDR team made significant advances in building a machine-learning (ML) based object detection model for various rockets and their plumes. The team will continue advancing the model, and begin work on the pan and tilt unit (PTU) control system and tracking algorithm development.

Alden Param↗

Prediction of High-Latitude Ionospheric Electrodynamics Using the Machine Learning Based Auroral Ionospheric Electrodynamics Model

We introduce a new framework for Machine-Learning (ML) based Auroral Ionosphere Model (ML-AIM). ML-AIM solves a current continuity equation by utilizing the ML model of Field Aligned Currents (FACs) of Kunduri et al., 2020 (https://doi.org/10.1029/2020JA027908), the FAC-derived aurora conductance model of Robinson et al., 2020 (https://doi.org/10.1029/2020JA028008), and the solar irradiance conductance model of Moen & Brekke (1993). The ML-AIM inputs are 60min time histories of solar wind plasma, interplanetary magnetic fields (IMF), and geomagnetic indices, and its outputs are ionospheric electric potential, electric fields, Pederson/Hall currents, and Joule Heating. We conduct two ML-AIM simulations for a weak geomagnetic activity on 14 May 2013 and a geomagnetic storm on 7-8 September 2017. ML-AIM produces reasonable ionospheric potential patterns such as two cell convection patterns and the enhancement of electric potentials during active times. The cross polar cap potential drop from ML-AIM is also comparable to the ones from the Weimer 2005 model, Super Dual Auroral Radar Network (SuperDARN), and Defense Meteorological Satellite Program (DMSP) F17 observations. ML-AIM is unique in a sense that it predicts ionospheric responses to the time-varying solar wind and geomagnetic conditions, while other traditional empirical model like Weimer 2005 is designed to provide static ionospheric conditions under steady solar wind/IMF conditions. In future, ML-AIM will include ML-based models of aurora precipitation and ionospheric conductance, improving its performance during active times.

H. K. Connor↗

ISS Regenerative Life Support: Challenges and Success in the Quest for Long-Term Habitability in Space

This presentation will discuss the International Space Station s (ISS) Regenerative Environmental Control and Life Support System (ECLSS) operations with discussion of the on-orbit lessons learned, specifically regarding the challenges that have been faced as the system has expanded with a growing ISS crew. Over the 10 year history of the ISS, there have been numerous challenges, failures, and triumphs in the quest to keep the crew alive and comfortable. Successful operation of the ECLSS not only requires maintenance of the hardware, but also management of the station resources in case of hardware failure or missed re-supply. This involves effective communication between the primary International Partners (NASA and Roskosmos) and the secondary partners (JAXA and ESA) in order to keep a reserve of the contingency consumables and allow for re-supply of failed hardware. The ISS ECLSS utilizes consumables storage for contingency usage as well as longer-term regenerative systems, which allow for conservation of the expensive resources brought up by re-supply vehicles. This long-term hardware, and the interactions with software, was a challenge for Systems Engineers when they were designed and require multiple operational workarounds in order to function continuously. On a day-to-day basis, the ECLSS provides big challenges to the on console controllers. Main challenges involve the utilization of the resources that have been brought up by the visiting vehicles prior to undocking, balance of contributions between the International Partners for both systems and resources, and maintaining balance between the many interdependent systems, which includes providing the resources they need when they need it. The current biggest challenge for ECLSS is the Regenerative ECLSS system, which continuously recycles urine and condensate water into drinking water and oxygen. These systems were brought to full functionality on STS-126 (ULF-2) mission. Through system failures and recovery, the ECLSS console has learned how to balance the water within the systems, store and use water for contingencies, and continue to work with the International Partners for short-term failures. Through these challenges and the system failures, the most important lesson learned has been the importance of redundancy and operational workarounds. It is only because of the flexibility of the hardware and the software that flight controllers have the opportunity to continue operating the system as a whole for mission success.

Bazley, Jesse A.↗

Condition-Based Maintenance of a Circulating Water System of a Canadian Nuclear Power Plant using Machine Learning and Statistical Tools

Canada Deuterium Uranium pressurized-heavy-water reactors (PHWR) are a type of nuclear power plant that generate clean and reliable energy. The scope of this work is to automate data analysis methodologies to inform a condition-based maintenance strategy of a circulating water system (CWS) of a PHWR. The multiunit CWS provides a continuous supply of water to cool steam condensers, even during transient scenarios, thereby improving the thermal efficiency. This work aims to develop a machine learning (ML) based approach to detect anomalies in heterogeneous data of a CWS in a PHWR to help inform a predictive maintenance strategy. The heterogeneous data include textual and numeric time series data for a PHWR. Natural-language-processing (NLP)-based models are used to analyze textual data contained in work orders and operator logs and an event-timeseries correlation detection method is applied to assist anomalies diagnoses for CWS. An ML model Robust Linear Model (RLM) is also used to remove the seasonal variations in the system variable distributions based on distributions of environmental variables. A machine learning model, Density-Based Spatial Clustering of Applications with Noise (DBSCAN), trained on both original data and data without any seasonal variations will then be used to detect if an anomaly exists. Thus, by moving to an automated methodology to detect, classify, and forecast anomalies, the maintenance strategy would be based on component condition instead of a time-based schedule.

97 - MATHEMATICS AND COMPUTING↗

Fourier series of atomic radial distribution functions: A molecular fingerprint for machine learning models of quantum chemical properties

Here we introduce a fingerprint representation of molecules based on a Fourier series of atomic radial distribution functions. This fingerprint is unique (except for chirality), continuous, and differentiable with respect to atomic coordinates and nuclear charges. It is invariant with respect to translation, rotation, and nuclear permutation, and requires no preconceived knowledge about chemical bonding, topology, or electronic orbitals. As such, it meets many important criteria for a good molecular representation, suggesting its usefulness for machine learning models of molecular properties trained across chemical compound space. To assess the performance of this new descriptor, we have trained machine learning models of molecular enthalpies of atomization for training sets with up to 10 k organic molecules, drawn at random from a published set of 134 k organic molecules with an average atomization enthalpy of over 1770 kcal/mol. We validate the descriptor on all remaining molecules of the 134 k set. For a training set of 10 k molecules, the fingerprint descriptor achieves a mean absolute error of 8.0 kcal/mol. This is slightly worse than the performance attained using the Coulomb matrix, another popular alternative, reaching 6.2 kcal/mol for the same training and test sets.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Monitoring Operational States of a Nuclear Reactor Using Seismoacoustic Signatures and Machine Learning

Monitoring nuclear reactors is an important safety and security task with growing requirements. We explore the possibility of using seismic and acoustic data for inferring the power level of an operating reactor. Continuous data recorded at a single seismoacoustic station that is located about 50 m away from a research reactor was visualized and analyzed. The data show a clear correlation between seismoacoustic features and reactor main operational states. We designed a workflow that includes two machine learning (ML) models to classify the reactor operational states (OFF, transition, and ON) and estimate reactor power levels (10%, 30%, 50%, 70%, and 90%). We applied and compared five ML algorithms for the reactor OFF-transition-ON and four approaches for the power level classification. We also compared the performance of ML models trained with seismic-only, acoustic-only, and both types of data. Five-fold cross validations were implemented to assure a thorough evaluation of the model performances. Additionally, the results show the extreme boosting gradient algorithm worked best for the first model, whereas random forests performed best for the second model. Combining seismic and acoustic data leads to better performance than using a single type of data. Seismic data contributed more than acoustic data for both models. We reached an accuracy of 0.98 for reactor OFF and ON. The accuracies for the transition state and power levels are less optimal with a minimum accuracy of 0.66. However, our results suggest seismic and acoustic data contain useful information about the transition state as well as power levels. Seismic and acoustic data could be integrated with other observations to improve monitoring performance.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

FY 2022 Site Sustainability Plan

The United States Department of Energy's (DOE's) National Renewable Energy Laboratory (NREL) is a recognized leader in sustainability, as evidenced by its ongoing optimization of resources in campus operations, including water, energy, waste, and purchasing, as well as continued commitment to meeting federal mandates and goals. NREL's dedication to sustainability supports the laboratory's success by applying what is learned through research and development to campus facilities and infrastructure systems. The Site Sustainability Plan provides a road map for all site planning and development. NREL drives the adoption of these initiatives to a global audience by conducting business operations that demonstrate the incorporation of clean energy practices. NREL will continue to develop a sustainable and resilient campus while growing greater technological capabilities to advance the national renewable energy marketplace.

climate change↗

Investigation of Multiple Data Streams for Gearbox Bearing Fault Prediction Through Machine-Learning Models

Operations and maintenance (O&M) cost of wind plant accounts up to 30% of total energy cost, which can be reduced through continuous monitoring and successfully detecting incipient wind turbine failures. To accomplish this, condition monitoring and predictive maintenance systems are being implemented in wind industry to support O&M decision making. A wide range of approaches for condition monitoring and fault prediction have been developed. These approaches generally use historical data of wind turbines collected by Supervisory Control and Data Acquisition (SCADA) system to identify patterns that lead to failure. These SCADA data show the overall condition of a wind turbine and can be leveraged to detect when the turbine's performance is degrading and to identify if a fault is developing. However, it becomes challenging to predict the failure of a specific wind turbine gearbox bearing, because the SCADA data are often not directly linked to the component. To bridge the gap, we have investigated features calculated from SCADA data using physics-based models and the gearbox design over the years. The damaged metric we used in the physics domain is frictional energy. Combining these physics domain variables with SCADA data as inputs to various machine learning models for gearbox bearing fault prediction, we have demonstrated the benefits of leveraging both physics and data domain models. It was an attempt to improve frictional-energy-based damage metric by adding data domain inputs, as we had learned that the frictional-energy-based damage metric alone is not sufficient to single out failed bearings from healthy. As condition monitoring data (either vibration or oil debris data) has become available at more and more wind plants, we would like to evaluate whether by adding the condition monitoring data can help further improve the performance of frictional-energy-based damage metric for gearbox bearing fault prediction. Both cases by modeling through various machine learning algorithms are discussed in this study along with some observations.

fault prediction↗

Quantum advantage for differential equation analysis

Quantum algorithms for differential equation solving, data processing, and machine learning potentially offer an exponential speedup over all known classical algorithms. However, there also exist obstacles to obtaining this potential speedup in useful problem instances. The essential obstacle for quantum differential equation solving is that outputting useful information may require difficult postprocessing, and the essential obstacle for quantum data processing and machine learning is that inputting the data is a difficult task just by itself. In this study, we demonstrate that, when combined, these difficulties solve one another. We show how the output of quantum differential equation solving can serve as the input for quantum data processing and machine learning, allowing dynamical analysis in terms of principal components, power spectra, and wavelet decompositions. To illustrate this, we consider continuous-time Markov processes on epidemiological and social networks. These quantum algorithms provide an exponential advantage over existing classical Monte Carlo methods.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Multiclass Flight Anomaly Detection Using Sensor Fusion Based on Dempster-Shafer Theory

As aviation systems in commercial operations continue to grow in complexity, the anomalies exhibited by these systems become more elaborate and difficult to detect. To address the challenge of detecting these complex anomalies, deep learning models have been used extensively in aviation anomaly detection studies, at the expense of end-user interpretability. Aiming to maintain the same level of interpretability as traditional threshold-exceedance methods, we continue our development of prediction models using ordinal patterns and their distributions throughout the flight. Specifically, this study extends our work into multiclass anomaly detection using sensor fusion based on Dempster-Shafer theory (DST), a second-order probability theory used to combine information from different sources of evidence. Our approach uses DST to reduce the uncertainty in the class predictions of an ensemble of classifiers. These classifiers rely on the similarity between flight data and class templates to make a prediction of the state of the aircraft. Our approach aims to take advantage of simple models trained on interpretable features (ordinal patterns) to correctly predict an anomaly and identify the flight dynamics linked to the anomaly. Our results show an improvement when using DST-based sensor fusion over simple majority voting. Additionally, our results provide insight into aircraft states linked to rare high-risk anomalies.

Risk detection↗

Multiclass Flight Anomaly Detection Using Sensor Fusion Based on Dempster-Shafer Theory

As aviation systems in commercial operations continue to grow in complexity, the anomalies exhibited by these systems become more elaborate and difficult to detect. To address the challenge of detecting these complex anomalies, deep learning models have been used extensively in aviation anomaly detection studies, at the expense of end-user interpretability. Aiming to maintain the same level of interpretability as traditional threshold-exceedance methods, we continue our development of prediction models using ordinal patterns and their distributions throughout the flight. Specifically, this study extends our work into multiclass anomaly detection using sensor fusion based on Dempster-Shafer theory (DST), a second-order probability theory used to combine information from different sources of evidence. Our approach uses DST toreduce the uncertainty in the class predictions of an ensemble of classifiers. These classifiers rely on the similarity between flight data and class templates to make a prediction of the state of the aircraft. Our approach aims to take advantage of simple models trained on interpretable features (ordinal patterns) to correctly predict an anomaly and identify the flight dynamics linked to the anomaly. Our results show an improvement when using DST-based sensor fusion over simple majority voting. Additionally, our results provide insight into aircraft states linked to rare high-risk anomalies.

Risk detection↗

3D reconstruction and neural rendering for adversarial machine learning

While evasion attacks on computer vision systems have been widely studied, creating attacks that remain effective under significant changes in viewpoint continues to be challenging. Traditional approaches often rely on affine transformations of images, but these approaches degrade at larger perspective shifts and often produce unrealistic or ineffective perturbations. Recent methods use differentiable renderers to improve viewpoint robustness, but they typically depend on manually constructed 3D models. We introduce a semi-automated pipeline that generates physically printable and perspective-invariant adversarial patches using only a small set of 2D images. Our method integrates 3D reconstruction, neural rendering, adversarial patch optimization, and an object detection victim model into a unified workflow. We use 2D Gaussian Splatting for high fidelity mesh reconstruction and FlexPara for surface parameterization that produces texture maps suitable for patch editing. Together, these components form a fully differentiable pipeline in PyTorch3D that links texture modification to model outputs, enabling efficient optimization of patches that remain effective across many viewpoints. The complete process, from image capture to patch printing and physical evaluation, can be completed within a few hours. We demonstrate the effectiveness of the resulting patches through attacks on the YOLOv8 object detection model and discuss remaining challenges and opportunities for improving robustness and scalability.

Singhvi, Vivaan [ORNL] (ORCID:0009000586288221)↗