SEARCH · Search NASA

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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 163 records · Page 9

All sources NASA NTRS DOE OSTI

NASA NTRS2002

A Summary of the NASA ISS Space Debris Collision Avoidance Program

Creating and implementing a process for the mitigation of the impact hazards due to cornets and asteroids will prove to be a complex and involved process. The closest similar program is the collision avoidance process currently used for protection of the International Space Station (ISS). This process, in operation for over three years, has many similarities to the NEG risk problem. By reviewing the ISS program, a broader perspective on the complications of and requirements for a NEO risk mitigation program might be obtained. Specifically, any lessons learned and continuing issues of concern might prove useful in the development of a NEO risk assessment and mitigation program.

Frisbee, Joseph↗

NASA NTRS2011

MODIS On-Orbit Performance and Lessons Learned

MODIS is a key instrument for the NASA's Earth Observing System (EOS) and has successfully operated for more than 11 and 9 years, respectively, on-board the Terra and Aqua spacecraft. MODIS collects data in 36 spectral bands, covering wavelengths from visible (VIS) to long-wave infrared (LWIR). To date, both Terra and Aqua MODIS have produced an unprecedented amount of data products and significantly contributed to the earth remote sensing studies and applications. MODIS was developed with stringent calibration requirements and was, consequently, designed and built with a set of on-board calibrators (OBC), which include a solar diffuser (SD), a solar diffuser stability monitor (SDSM), a blackbody (BB), and a spectroradiometric calibration assembly (SRCA). This presentation briefly reviews MODIS instrument operation and various calibration and characterization activities, It demonstrates both the instrument and the OBC on-orbit performance and discusses lessons learned, particularly focusing on on-orbit changes in sensor responses, optics degradation, and major challenging issues. As expected, Terra and Aqua MODIS on-orbit performance and lessons learned will continue to benefit the operation and calibration of future sensors, such as NPP/JPSS VIIRS and GOES-R ABI.

Xiong, Xiaoxiong↗

NASA NTRS2015

International Space Station (ISS) Environmental Control and Life Support (ECLS) System Overview of Events: 2010-2014

Nov 2, 2014 marked the completion of the 14th year of continuous human presence in space on board the International Space Station (ISS). After 42 expedition crews, over 115 assembly & utilization flights, over 180 combined Shuttle/Station, US & Russian Extravehicular Activities (EVAs), the post-Assembly-Complete ISS continues to fly and the engineering teams continue to learn from operating its systems, particularly the life support equipment. Problems with initial launch, assembly and activation of ISS elements have given way to more long term system operating trends. New issues have emerged, some with gestation periods measured in years. Major events and challenges for each U.S. Environmental Control and Life Support (ECLS) subsystem occurring during calendar years 2010 through 2014 are summarily discussed in this paper, along with look-aheads for what might be coming in the future for each U.S. ECLS subsystem.

Gentry, Gregory J.↗

DOE OSTI2012

NRC Waste Incidental to Reprocessing Program: Overview of Consultation and Monitoring Activities at the Idaho National Laboratory and the Savannah River Site - What We Have Learned - 12470

In 2005 the U.S. Nuclear Regulatory Commission (NRC) began to implement a new set of responsibilities under the Ronald W. Reagan National Defense Authorization Act (NDAA) of Fiscal Year 2005. Section 3116 of the NDAA requires the U.S. Department of Energy (DOE) to consult with the NRC for certain non-high level waste determinations and also requires NRC to monitor DOE's disposal actions related to those determinations. In Fiscal Year 2005, the NRC staff began consulting with DOE and completed reviews of draft waste determinations for salt waste at the Savannah River Site. In 2006, a second review was completed on tank waste residuals including sodium-bearing waste at the Idaho Nuclear Technology and Engineering Center Tank Farm at the Idaho National Laboratory. Monitoring Plans were developed for these activities and the NRC is actively monitoring disposal actions at both sites. NRC is currently in consultation with DOE on the F-Area Tank Farm closure and anticipates entering consultation on the H-Area Tank Farm at the Savannah River Site. This paper presents, from the NRC perspective, an overview of how the consultation and monitoring process has evolved since its conception in 2005. It addresses changes in methods and procedures used to collect and develop information used by the NRC in developing the technical evaluation report and monitoring plan under consultation and the implementation the plan under monitoring. It will address lessons learned and best practices developed throughout the process. The NDAA has presented significant challenges for the NRC and DOE. Past and current successes demonstrate that the NDAA can achieve its intended goal of facilitating tank closure at DOE legacy defense waste sites. The NRC believes many of the challenges in performing the WD reviews have been identified and addressed. Lessons learned have been collected and documented throughout the review process. Future success will be contingent on each agencies commitment to consistently apply the lessons learned and continue to create an open and collaborative work environment to maintain the process of continuous improvement. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

DOE OSTI2019

Inventory of Radioactive Waste in Canada - 19336

Since 1999, the Government of Canada has been conducting a regular exercise of gathering and compiling Canada's radioactive waste inventory and projection data from waste owners and making it available to the public in the form of a triennial Inventory Summary Report. The goal of this data collection exercise has been to support federal policy development, support the Canadian Nuclear Safety Commission's regulatory activities, provide consolidated radioactive waste information for the nuclear industry, promote transparency, and meet Canada's international commitments and growing demands internationally for radioactive waste data. The detailed national inventory report provides an overview of the production, accumulation and future projections of radioactive waste in Canada. Since the coming into force of the Joint Convention on the Safety of Spent Fuel Management and on the Safety of Radioactive Waste Management in June 2001, the data collection and reporting cycle has been aligned to provide the waste inventory data required for Canada's triennial National Report for the Joint Convention. This paper presents Canada's radioactive waste inventory and projection data as of the end of December 2016, and summarizes key findings from the analysis of the data. For this reporting cycle, the waste data collected was expanded to include more detailed information on the decommissioning schedules and associated waste volume projections for nuclear power plants, and research and prototype reactors, as well as plans for transitioning waste inventories from storage to long-term management and disposal facilities, based on current schedules for establishing these facilities in Canada. With this additional information, the Government of Canada and Canadian stakeholders are better able to understand the waste disposal capacity and schedules required to accommodate all nuclear fuel waste and radioactive waste that will be produced by Canada's existing fleet of nuclear power plants and research reactors. The collection and analysis of projected waste volumes to be emplaced in long-term management facilities is in line with and will support the proposed methodology of presenting waste inventories in terms of management strategy and disposal routes by the Organization for Economic Cooperation and Development Nuclear Energy Agency Expert Group on Waste Inventorying and Reporting Methodology. The paper describes some of the uses that are made of the waste inventory information, as well as Canada's approach to national inventory data collection and its experience and lessons learned. Canada continues to modernize the format of the Inventory Summary Report with the goal of disseminating the technical information into an attractive and public-friendly report. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

NASA NTRS2022

Operational maize yield forecasts for Sub-Saharan Africa using Earth observation data and machine learning

Food insecurity continues to grow in Sub-Saharan Africa (SSA). In 2019, chronically malnourished people numbered nearly 240 million, or 20% of the population in SSA. Globally, numerous efforts have been made to anticipate potential droughts, crop conditions, and food shortages to foster improved food insecurity early warning and risk management. To support this goal, we develop an Earth Observation (EO) and machine-learning-based operational, subnational maize yield forecast system and evaluate its out-of-sample forecast skills during the growing seasons for Kenya, Somalia, Malawi, and Burkina Faso. In general, forecast skills improve substantially during the vegetative growth period (VP) and gradually during the reproductive development period (RP). Thus, mid-season assessment can provide effective early warning months before harvest. Skillful forecasts (Nash Sutcliffe Efficiency (NSE) > 0.6 and Mean Absolute Percentage Error (MAPE) < 20%) appear approximately two dekads after the VP; for example, they appear in May in Kenya and Somalia, January in Malawi, and July in Burkina Faso.

Donghoon Lee↗

DOE OSTI2021

Real-Time Evolution and Deployment of Neuromorphic Computing at The Edge

Extremely low power neuromorphic systems are well-suited for deployment to the edge for many applications. In many use cases of neuromorphic computing for control, a spiking neural network is trained off-line using a simulation and then deployed to a neuromorphic system at the edge, where it will operate without ongoing training or learning. However, it may be desirable to continue training or learning at the edge to refine or adapt to the real-world system. In this work, we propose an approach for performing real-time evolutionary optimization for spiking neural networks for neuromorphic deployment at the edge. In particular, we propose a combination of simulation and real-world evaluations, along with feedback from the real-world environment, to train spiking neural networks for continuous deployment to the edge. We show that the real-time evolution at the edge approach achieves comparable performance to an evolution approach that requires constant evaluation in the realworld environment.

Schuman, Catherine↗

NASA NTRS1998

The Process of Science Communications at NASA/Marshall Space Flight Center

The communication of new scientific knowledge and understanding is an integral component of science research, essential for its continued survival. Like any learning- based activity, science cannot continue without communication between and among peers so that skeptical inquiry and learning can take place. This communication provides necessary organic support to maintain the development of new knowledge and technology. However, communication beyond the peer-community is becoming equally critical for science to survive as an enterprise into the 21st century. Therefore, scientists not only have a 'noble responsibility' to advance and communicate scientific knowledge and understanding to audiences within and beyond the peer-community, but their fulfillment of this responsibility is necessary to maintain the survival of the science enterprise. Despite the critical importance of communication to the viability of science, the skills required to perform effective science communications historically have not been taught as a part of the training of scientist, and the culture of science is often averse to significant communication beyond the peer community. Thus scientists can find themselves ill equipped and uncomfortable with the requirements of their job in the new millennium. At NASA/Marshall Space Flight Center, we have developed and implemented an integrated science communications process, providing an institutional capability to help scientist accurately convey the content and meaning of new scientific knowledge to a wide variety of audiences, adding intrinsic value to the research itself through communication, while still maintaining the integrity of the peer-review process. The process utilizes initial communication through the world-wide web at the site http://science.nasa.gov to strategically leverage other communications vehicles and to reach a wide-variety of audiences. Here we present and discuss the basic design of the science communications process, now in operation for nearly two years. Serving scientists in Earth Science, Microgravity Science, and Space Science. Critical features of the design are illustrated, and essential skills required to operate the process are defined. Measures of success will also be presented.

Horack, John M.↗

DOE OSTI2020

Phase behavior of continuous-space systems: A supervised machine learning approach

The phase behavior of complex fluids is a challenging problem for molecular simulations. Supervised machine learning (ML) methods have shown potential for identifying the phase boundaries of lattice models. In this work, we extend these ML methods to continuous-space systems. We propose a convolutional neural network model that utilizes grid-interpolated coordinates of molecules as input data of ML and optimizes the search for phase transitions with different filter sizes. We test the method for the phase diagram of two off-lattice models, namely, the Widom–Rowlinson model and a symmetric freely jointed polymer blend, for which results are available from standard molecular simulations techniques. The ML results show good agreement with results of previous simulation studies with the added advantage that there is no critical slowing down. We find that understanding intermediate structures near a phase transition and including them in the training set is important to obtain the phase boundary near the critical point. The method is quite general and easy to implement and could find wide application to study the phase behavior of complex fluids.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2023

In vitro continuous protein evolution empowered by machine learning and automation

Directed evolution has become one of the most successful and powerful tools for protein engineering. However, the efforts required for designing, constructing, and screening a large library of variants can be laborious, time-consuming, and costly. With the recent advent of machine learning (ML) in the directed evolution of proteins, researchers can now evaluate variants in silico and guide a more efficient directed evolution campaign. Furthermore, recent advancements in laboratory automation have enabled the rapid execution of long, complex experiments for high-throughput data acquisition in both industrial and academic settings, thus providing the means to collect a large quantity of data required to develop ML models for protein engineering. In this perspective, here we propose a closed-loop in vitro continuous protein evolution framework that leverages the best of both worlds, ML and automation, and provide a brief overview of the recent developments in the field.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2022

MULTISTEP AND CONTINUOUS PHYSICS-INFORMED NEURAL NETWORK METHODS FOR LEARNING GOVERNING EQUATIONS AND CONSTITUTIVE RELATIONS

In this work, we investigate the applicability and relative merit of discrete and continuous versions of physics-informed neural network (PINN) methods for learning unknown governing equations or constitutive relations in a nonlinear dynamical system. In the case of unknown dynamics, entire right-hand-side (RHS) equations of the ordinary differential equations are unknown. In the case of unknown constitutive relations, however, the RHS equations are known up to the specification of constitutive relations (that may depend on the state of the system). We use a deep neural network to model unknown governing equations or constitutive relations. The discrete PINN approach combines classical multistep discretization methods for dynamical systems with neural-network-based machine learning methods. On the other hand, the continuous versions utilize deep neural networks to minimize the residual function for the continuous governing equations. We use the case of a fedbatch bioreactor system to study the effectiveness of these approaches and discuss conditions for their applicability. Our results indicate that the accuracy of the trained neural network models is much higher for the cases where we only have to learn a constitutive relation instead of all dynamics. This finding corroborates the well-known fact from scientific computing that building as much structural information as is available into an algorithm can enhance its efficiency and/or accuracy.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2021

COVID-19-associated school closures and related efforts to sustain education and subsidized meal programs, United States, February 18-June 30, 2020

Pre-emptive school closures are frontline community mitigation measures recommended by the US Centers for Disease Control and Prevention (CDC) for implementation during severe pandemics. This study describes the spatiotemporal patterns of publicly announced school closures implemented in response to the coronavirus disease 2019 (COVID-19) pandemic and assesses how public K-12 districts adjusted their methods of education delivery and provision of subsidized meals. During February 18–June 30, 2020, we used daily systematic media searches to identify publicly announced COVID-19–related school closures lasting ≥1 day in the United States (US). We also collected statewide school closure policies from state government websites. Data on distance learning and subsidized meal programs were collected from a stratified sample of 600 school districts. The first COVID-19–associated school closure occurred on February 27, 2020 in Washington state. By March 30, 2020, all but one US public school districts were closed, representing the first-ever nearly synchronous nationwide closure of public K-12 schools in the US. Approximately 100,000 public schools were closed for ≥8 weeks because of COVID-19, affecting >50 million K-12 students. Of 600 districts sampled, the vast majority offered distance learning (91.0%) and continued provision of subsidized meal programs (78.8%) during the closures. Despite the sudden and prolonged nature of COVID-19–associated school closures, schools demonstrated flexibility by implementing distance learning and alternate methods to continue subsidized meal programs.

60 APPLIED LIFE SCIENCES↗

NASA NTRS1998

The Process of Science Communications at NASA/Marshall Space Flight Center

Horack, John M.↗

DOE OSTI2026

Speedup of UEDGE Parameter Scans Using Machine-Learning Optimized OpenMP Parallelization and a Continuation Solver

This article presents the OpenMP parallelization of the preconditioning Jacobian assembly and right‐hand side residual evaluation in UEDGE. A continuation algorithm, utilizing the internal NKSOL implicit Jacobian‐Free Newton‐Krylov solver to efficiently scan physical parameters, is also presented. The implemented parallelization reduces the computational time for a benchmark scan run on 32 threads by compared to the serial version when using trained random forest regression models to identify the optimal decomposition of the system of equations. Random forest regression models applied to the UEDGE time‐dependent and continuation solver algorithms did not yield meaningful improvement in computational performance. A benchmark DIII‐D gas injection rate scan in the 0.35–0.75 kA interval, performed on a test cluster using the parallelized code and continuation solver, produced 1066 steady‐state solutions with a 22 s average wall‐clock computational time per steady‐state solution.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2020

Reinforcement learning based automated history matching for improved hydrocarbon production forecast

History matching aims to find a numerical reservoir model that can be used to predict the reservoir performance. An engineer and model calibration (data inversion) method are required to adjust various parameters/properties of the numerical model in order to match the reservoir production history. In this study, we develop deep neural networks within the reinforcement learning framework to achieve automated history matching that will reduce engineers’ efforts, human bias, automatically and intelligently explore the parameter space, and remove the need of large set of labeled training data. To that end, a fast-marching-based reservoir simulator is encapsulated as an environment for the proposed reinforcement learning. The deep neural-network-based learning agent interacts with the reservoir simulator within reinforcement learning framework to achieve the automated history matching. Reinforcement learning techniques, such as discrete Deep Q Network and continuous Deep Deterministic Policy Gradients, are used toth, used to train the learning agents. The continuous actions enable the Deep Deterministic Policy Gradients to explore more states at each iteration in a a learning episode; consequently, a better history matching is achieved using this algorithm as compared to Deep Q Network. For simplified dual-target composite reservoir models, the best history-matching performances of the discrete and continuous learning methods in terms of normalized root mean square errors are 0.0447 and 0.0038, respectively. Furthermore, our study shows that continuous action space achieved by the deep deterministic policy gradient drastically outperforms deep Q network.

42 ENGINEERING↗

DOE OSTI2025

Identifying Vehicle Signals in Continuous Seismic Data Using Unsupervised Machine-Learning Techniques

Seismic sensors deployed near roadways effectively capture ground vibrations generated by passing vehicles. Although both traditional and machine‐learning algorithms have been utilized for analyzing such signals, independent validation of detected vehicle events remains limited. We applied two unsupervised machine‐learning algorithms, uniform manifold approximation and projection for dimension reduction, and hierarchical density‐based spatial clustering of applications with noise, to continuous seismic data collected along a road on the main campus of Oak Ridge National Laboratory. The algorithms identified seven distinct cluster labels across the entire dataset. By comparing these cluster labels with precipitation records from a nearby weather station and image‐derived labels from a local camera system, we identified one cluster associated with rainfall and another with vehicle activity. Our algorithms identified a greater number of vehicle‐related labels compared to the camera‐derived labels because seismic data are unaffected by poor lighting conditions. The arrival times of the newly detected vehicle signals corresponded well with the road’s speed limit, supporting our findings. Our algorithm outperformed the short‐term average/long‐term average method and k‐means clustering. Our results suggest that seismic data, when analyzed with machine‐learning algorithms, can complement existing vehicle monitoring systems, particularly under challenging environmental conditions.

Chai, Chengping [Oak Ridge National Laboratory (OR↗

DOE OSTI2021

Machine learning framework for quantum sampling of highly constrained, continuous optimization problems

In recent years, there is growing interest in using quantum computers for solving combinatorial optimization problems. In this work, we developed a generic, machine learning-based framework for mapping continuous-space inverse design problems into surrogate quadratic unconstrained binary optimization (QUBO) problems by employing a binary variational autoencoder and a factorization machine. The factorization machine is trained as a low-dimensional, binary surrogate model for the continuous design space and sampled using various QUBO samplers. Using the D-Wave Advantage hybrid sampler and simulated annealing, we demonstrate that by repeated resampling and retraining of the factorization machine, our framework finds designs that exhibit figures of merit exceeding those of its training set. We showcase the framework’s performance on two inverse design problems by optimizing (i) thermal emitter topologies for thermophotovoltaic applications and (ii) diffractive meta-gratings for highly efficient beam steering. This technique can be further scaled to leverage future developments in quantum optimization to solve advanced inverse design problems for science and engineering applications.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

NASA NTRS1972

Effect of continuous gamma-ray exposure on performance of learned tasks and effect of subsequent fractionated exposures on blood-forming tissue

Sixteen monkeys trained to perform continuous and discrete-avoidance and fixed-ratio tasks with visual and auditory cues were performance-tested before, during, and after 10-day gamma-ray exposures totaling 0, 500, 750, and 1000 rads. Approximately 14 months after the performance-test exposures, surviving animals were exposed to 100-rad gamma-ray fractions at 56-day intervals to observe injury and recovery patterns of blood-forming tissues. The fixed-ratio, food-reward task performance showed a transient decline in all dose groups within 24 hours of the start of gamma-ray exposure, followed by recovery to normal food-consumption levels within 48 to 72 hours. Avoidance tasks were performed successfully by all groups during the 10-day exposure, but reaction times of the two higher dose-rate groups in which animals received 3 and 4 rads per hour or total doses of 750 and 1000 rads, respectively, were somewhat slower.

Spalding, J. F.↗