Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “learned priors”

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 487 records · Page 27
All sourcesNASA NTRSDOE OSTI
NASA NTRS2020

MER Opportunity dust-storm recovery operations and implications for future Mars surface missions

In June 2018, NASA’s Mars Exploration Rover Opportunity became engulfed in the most intense global dust storm observed in its 14-year mission and in Mars’ recorded history. Sapped of life-giving solar energy, Opportunity fell silent to ground operators on June 11, in what would be its final call home. Over the course of the next eight months, the MER team employed numerous recovery efforts and radiated over one thousand commands to wake the silent rover. Although unsuccessful, MER’s dust-storm recovery team changed the paradigm of dust-storm operations from Opportunity’s survival of a previous global dust storm in 2007. In this paper, the authors offer a glimpse into MER’s recovery efforts and lessons learned from Opportunity dust-storm operations for future solar-powered Mars surface missions. In the first section, the authors discuss the indicators of the approaching dust storm and the actions the team employed to reduce Opportunity’s power consumption and preserve available battery charge prior to loss of contact. The second section discusses the team’s recovery efforts until Opportunity’s declared End-of-Mission, the steps the team took to re-establish contact with Opportunity, the assumptions made during each step of the recovery process, and the commanding actions employed. Finally, the authors discuss the operational impacts of global dust storms on the safe and successful operation of solar-powered spacecraft on the Martian surface, and offer design recommendations for future solar-powered missions from the lessons learned during Opportunity’s 14-year mission and experience through two global dust storms.

Nelson, Robert W.↗
DOE OSTI2022

Design of a neutral thermal scattering (NeTS) module for hydrogen in light water

The accurate representation of thermal scattering law (TSL) data is integral to the design and characterization of many modern nuclear systems, particularly those using light water as a moderator/coolant. As a material-dependent distribution over energy-momentum phase space, the TSL may exhibit a variety of unique and relevant conditional (e.g., temperature, pressure) and compositional (e.g., porosity, stoichiometry, radiation damage) dependencies. Currently, there are various approaches to incorporate temperature dependence, which is especially important in coupled neutronic-thermal hydraulic simulations. In each approach, there is an inherent tradeoff between memory consumption and accuracy. Some techniques require tens to hundreds of MBs or more, while others fail to reproduce the underlying data to within 10% error over the considered input domain, despite having a reduced storage burden. This work aims to address both sides of the tradeoff simultaneously by implementing a novel deep learning (DL) approach to TSL representation. The neural thermal scattering (NeTS) concept, which is amenable to an arbitrary number of dependencies, is demonstrated via the inclusion of temperature dependence into a highly compact, highly accurate functional form of the multi- variate TSL (i.e., S(α, β, T)) for hydrogen in light water. Resulting storage requirements are on the order of 100 kB, and median and maximum percent deviations are on the order of 0.1% and 1%, respectively. These measures represent a step improvement over previous techniques. Notably, the developed neural network and feature methodology build on those employed in prior work on beryllium oxide. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2021

Enhanced Data Efficiency Using Deep Neural Networks and Gaussian Processes for Aerodynamic Design Optimization

Adjoint-based optimization methods are attractive for aerodynamic shape design primarily due to their computational costs being independent of the dimensionality of the input space and their ability to generate high-fidelity gradients that can then be used in a gradient-based optimizer. This makes them very well suited for high-fidelity simulation based aerodynamic shape optimization of highly parametrized geometries such as aircraft wings. However, the development of adjoint-based solvers involve careful mathematical treatment and their implementation require detailed software development. Furthermore, they can become prohibitively expensive when multiple optimization problems are being solved, each requiring multiple restarts to circumvent local optima. In this work, we propose a machine learning enabled, surrogate-based framework that replaces the expensive adjoint solver, without compromising on predicting predictive accuracy. Specifically, we first train a deep neural network (DNN) from training data generated from evaluating the high-fidelity simulation model on a model-agnostic design of experiments on the geometry shape parameters. The optimum shape may then be computed by using a gradient-based optimizer coupled with the trained DNN. Subsequently, we also perform a gradient-free Bayesian optimization, where the trained DNN is used as the prior mean. We observe that the latter framework (DNN-BO) improves upon the DNN-only based optimization strategy for the same computational cost. Overall, this framework predicts the true optimum with very high accuracy, while requiring far fewer high-fidelity function calls compared to the adjoint-based method. Furthermore, we show that multiple optimization problems can be solved with the same machine learning model with high accuracy, to amortize the offline costs associated with constructing our models. Our methodology finds applications in the early stages of aerospace design. (C) 2021 Published by Elsevier Masson SAS.

Renganathan, S. Ashwin↗
DOE OSTI2024

Autonomous anomaly detection of proliferation in the AGN-201 nuclear reactor digital twin

The expansion of global nuclear power necessitates advanced methods for analyzing proliferation indicators. This study introduces a novel application of the Isolation Forest Machine Learning (IFML) algorithm within a digital twin (DT) of the AGN-201 nuclear reactor to autonomously detect anomalies. Leveraging real-time operational data from the AGN-201 DT, the IFML algorithm identifies outliers without prior data labeling and operates as a lightweight, complementary approach to traditional physics-based anomaly detection methods for nuclear safeguards. In a simulated Red vs. Blue team exercise, the IFML algorithm successfully detected six significant unseen anomalies related to reactivity changes, achieving an accuracy of 99% for identifying operational deviationxs. These anomalies, caused by deliberate perturbations, were detected alongside known physics-based models, underscoring the potential of IFML to enhance real-time monitoring without displacing traditional methods. Further, this study highlights the applicability of IFML in nuclear environments by providing an additional, redundant layer of anomaly detection to improve safeguards and operational safety in complex systems.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2023

More Than 30 000-fold Field Enhancement of Terahertz Nanoresonators Enabled by Rapid Inverse Design

The rapid development of 6G communications using terahertz (THz) electromagnetic waves has created a demand for highly sensitive THz nanoresonators capable of detecting these waves. Among the potential candidates, THz nanogap loop arrays show promising characteristics but require significant computational resources for accurate simulation. This requirement arises because their unit cells are 10 times smaller than millimeter wavelengths, with nanogap regions that are 1 000 000 times smaller. To address this challenge, we propose a rapid inverse design method using physics-informed machine learning, employing double deep Q-learning with an analytical model of the THz nanogap loop array. In ~39 h on a middle-level personal computer, our approach identifies the optimal structure through 200 000 iterations, achieving an experimental electric field enhancement of 32 000 at 0.2 THz, 300% stronger than prior results. In conclusion, our analytical model-based approach significantly reduces the amount of computational resources required, offering a practical alternative to numerical simulation-based inverse design for THz nanodevices.

21 terahertz time-domain spectroscopy↗
NASA NTRS2017

Online Multi-Modal Learning and Adaptive Information Trajectory Planning for Autonomous Exploration

In robotic information gathering missions, scientists are typically interested in understanding variables which require proxy measurements from specialized sensor suites to estimate. However, energy and time constraints limit how often these sensors can be used in a mission. Robots are also equipped with cheaper to use navigation sensors such as cameras. In this paper, we explore a challenging planning problem in which a robot is required to learn about a scientific variable of interest in an initially unknown environment by planning informative paths and deciding when and where to use its sensors. To tackle this we present two innovations: a Bayesian generative model framework to automatically learn correlations between expensive science sensors and cheaper to use navigation sensors online, and a sampling based approach to plan for multiple sensors while handling long horizons and budget constraints. Our approach does not grow in complexity with data and is anytime making it highly applicable to field robotics. We tested our approach extensively in simulation and validated it with real data collected during the 2014 Mojave Volatiles Prospector Mission. Our planning algorithm performs statistically significantly better than myopic approaches and at least as well as a coverage-based algorithm in an initially unknown environment while having added advantages of being able to exploit prior knowledge and handle other intricacies of the real world without further algorithmic modifications.

learning↗
NASA NTRSDate not supplied

Thermal Performance Comparison and Lessons Learned for the Thermal Infrared Sensor Instruments 1 & 2

The Thermal Infrared Sensor (TIRS-1) is one of two instruments on the Landsat-8 Mission, which launched in February of 2013 and remains operational. The TIRS-2 instrument was developed for the Landsat-9 Mission, which is scheduled to launch in September of 2021. The TIRS-2 instrument design was adjusted to account for requirements changes and to include key lessons learned from its predecessor. The overall thermal subsystem design of the TIRS Sensor Unit remained comparable from mission to mission, but there were some areas that needed modifications. The general design is comprised of five thermal zones which range in temperature from less than 43 Kelvin to 320 Kelvin. Most zones are proportional heater controlled. A two-stage cryocooler provided by Ball Aerospace is used to cool the cryogenic subsystem, and excess power dissipation is rejected via ammonia transport heat pipes to a dedicated cryocooler radiator with embedded ammonia heat pipes. The cryogenic subsystem includes a series of shells used to radiatively and conductively isolate the cold stage from the warmer surroundings. The Optical System (telescope) is passively cooled to 180-195 Kelvin using a dedicated radiator with embedded dual-bore ethane heat pipes. The warmer end of the instrument includes a Scene Select Mechanism, an on-board Black Body Calibrator with a dedicated radiator, and a Focal Plane Electronics Box, all of which are attached to the Sensor Unit primary structure. The TIRS-1 thermal subsystem design was successfully verified during an extended testing campaign and during more than 8years of on-orbit operations. The TIRS-2 thermal subsystem design was successfully verified during months of ground testing prior to delivery in August of 2019and at the Observatory level in the Spring of 2021. A detailed comparison of the thermal performance of the two instruments has been made with a focus on key lessons learned during each instrument development, and with the benefit of on-orbit data acquired for the TIRS-1 instrument.

Veronica Otero↗
DOE OSTI2022

Scale-up of microwave-assisted, continuous flow, liquid phase reactors: Application to 5-Hydroxymethylfurfural production

Microwave (MW) technology can be powerful for the electrification and intensification of chemical manufacturing. However, very few examples of scaled MW processes exist. In this work, we build a scaled, MW-assisted, continuous flow reactor for the processing of liquid phase chemistries with specific application to 5-hydroxymethylfurfural (HMF) production. Here, we demonstrate a corresponding computational fluid dynamics model to simulate the reactor’s temperature profile and performance, both of which are in good agreement with experiments. We construct a surrogate model to relate operating parameters to performance for guiding experiments, without demanding simulations, via active learning. Heat recirculation is demonstrated in this scaled reactor, further extending its scale and reducing the energy demand. An HMF yield of ~55 % and a productivity of 0.1 kg/hr, 8x higher than any other reactor, at a flowrate >20x than prior work, are demonstrated while maintaining energy efficiency of >98 %. A basic economic analysis estimates a $1.85/kg cost of HMF from fructose and >60 % reduction in CO 2 emissions than a conventional system without considering the impact of green electricity. We demonstrate that MW technology is well-applied to systems larger than the laboratory scale, and strategies to further reduction in cost and CO 2 are discussed.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2022

Pseudonymization at Scale: OLCF’s Summit Usage Data Case Study

The analysis of vast amounts of data and the processing of complex computational jobs have traditionally relied upon high performance computing (HPC) systems, which offer reliable and efficient management of large-scale computational and data resources. Understanding these analyses’ needs is paramount for designing solutions that can lead to better science, and similarly, understanding the characteristics of the user behavior on those systems is important for improving user experiences on HPC systems. A common approach to gathering data about user behavior is to extract workload characteristics from system log data available only to system administrators. Recently at Oak Ridge Leadership Computing Facility (OLCF), however, we unveiled user behavior about the Summit supercomputer by collecting data from a user’s point of view with ordinary Unix commands.In this paper, we discuss the process, challenges, and lessons learned while preparing this dataset for publication and submission to an open data challenge. The original dataset contains personal identifiable information (PII) about the users of OLCF which needed be masked prior to publication, and we determined that anonymization, which scrubs PII completely, destroyed too much of the structure of the data to be interesting for the data challenge. We instead chose to pseudonymize the dataset, which reduced the linkability of the dataset to the users’ identities. Pseudonymization is significantly more computationally expensive than anonymization, and the size of our dataset, which is approximately 175 million lines of raw text, necessitated the development of a parallelized workflow that could be reused on different HPC machines. We demonstrate the scaling behavior of the workflow on two leadership class HPC systems at OLCF, and we show that we were able to bring the overall makespan time from an impractical 20+ hours on a single node down to around 2 hours. As a result of this work, we release the entire pseudonymized dataset and make the workflows and source code publicly available.

Maheshwari, Ketan↗
DOE OSTI2023

BitGNN: Unlocking the Performance Potential of Binary Graph Neural Networks on GPUs

Graph Neural Networks (GNNs) have shown compelling results in many graph-based learning tasks. They are, however, time-consuming. Recent work has shown a promising direction in improving GNN speed and shrinking the size — network binarization, which binarizes network values and operations. Prior work, however, mainly focused on algorithm designs, leaving it open on how to fully materialize the performance potential. This work fills the gap by proposing techniques to best map binary GNNs and their computations to fit the nature of bit manipulations, optimizations and algorithms to maximize BSpMM kernel efficiency, and solutions to other factors influencing the end-to-end time on GPUs. Results on real-world graphs show that the proposed techniques outperform state of-the-art binary GNN implementations by 21-67× with little accuracy loss.

Chen, Jou-An↗
DOE OSTI2021

pnnl/mol_dgnn

Molecular dynamics simulations often rely on parameterized atomistic potentials, which can be inaccurate compared to ab initio simulation methods. However, dynamic ab initio simulations come with high computational cost. Deep learning techniques offer a method to obtain both high accuracy and low cost predictions of dynamics. Our method gives a prediction of the structural dynamics of a molecule accounting for the prior dynamics, giving high-accuracy predictions at low computational cost

Ashby, Michael↗
NASA NTRS2011

Markov Chain Monte Carlo Bayesian Learning for Neural Networks

Conventional training methods for neural networks involve starting al a random location in the solution space of the network weights, navigating an error hyper surface to reach a minimum, and sometime stochastic based techniques (e.g., genetic algorithms) to avoid entrapment in a local minimum. It is further typically necessary to preprocess the data (e.g., normalization) to keep the training algorithm on course. Conversely, Bayesian based learning is an epistemological approach concerned with formally updating the plausibility of competing candidate hypotheses thereby obtaining a posterior distribution for the network weights conditioned on the available data and a prior distribution. In this paper, we developed a powerful methodology for estimating the full residual uncertainty in network weights and therefore network predictions by using a modified Jeffery's prior combined with a Metropolis Markov Chain Monte Carlo method.

Goodrich, Michael S.↗
NASA NTRS2022

National Campaign (NC)-1 Strategic Conflict Management Simulation (X4) Final Report

Urban Air Mobility (UAM) enables highly automated, cooperative, passenger or cargo-carrying air transportation services in and around urban areas. UAM is a subset of the Advanced Air Mobility (AAM) concept under development by the National Aeronautics and Space Administration (NASA), the Federal Aviation Administration (FAA), and industry. The Strategic Conflict Management (SCM) Simulation, dubbed “X4”, was conducted between July 2021 and June 2022 by NASA with the FAA and industry to evolve the Provider of Services for UAM (PSU) that will be needed to ensure initial UAM operations can scale in the National Airspace System (NAS). The FAA UAM Concept of Operations (ConOps) v1 [1] served as an initial guiding document for the X4 airspace management system design to ensure the architecture supports testing of services provided by third-party service providers. To that end, the X4 architecture leveraged concepts and technologies developed for Unmanned Aircraft System (UAS) Traffic Management (UTM) while also developing and testing new capabilities and services needed for UAM. The architecture included an initial prototype of the FAA-Industry Exchange Protocol (FIDXP), third-party services such as the PSUs, and Discovery and Synchronization Service (DSS), and other new services such as Demand-Capacity Balancing (DCB) to facilitate UAM strategic conflict management. During X4, NASA led discussions and collaborated with seven industry airspace partners to develop initial airspace management concepts for UAM that drove what would be tested and evaluated during the simulations. The initial capabilities defined for PSU leveraged UTM UAS Service Supplier (USS) as a starting point and evolved to meet UAM requirements. These discussions were also an opportunity for NASA to collaborate with industry to develop a set of initial Community-Based Rules (CBRs). These CBRs enabled how UAM traffic would be cooperatively managed among UAM operators. The collaborative, iterative process of developing CBRs with industry during X4 provided insight into the challenges involved and identified the need for a suitable and effective forum for future CBR development. In parallel with these discussions, NASA conducted a series of seven software sprints and two collaborative simulations with the airspace partners that built up in complexity. Over the course of the year, all seven partners successfully completed all the sprints by demonstrating the required capabilities. They also participated in the two collaborative simulations to demonstrate how multiple PSUs could work together in a collaborative, more complex environment with higher traffic density. The X4 simulation accomplished NASA's objectives and helped advance the development of the seven participating PSUs. The lessons learned provided insight into key elements of the UAM Notional Architecture from the FAA ConOps v1 [1] and UTM technologies when applied to UAM: - Having a Concept of Use (ConUse) defined prior to the activity would accelerate the time and effort from concept development to testing. - While the UAM Notional Architecture provided a starting point for a federated architecture that support services provided by third-party providers, the USS and PSU differed in their capability definitions for operational intent submission and sharing, conformance monitoring, airspace authorization, strategic conflict management, airspace constraints and dynamic replanning. - While existing DSS developed for UTM provided a way for PSU and other UAM services (such as DCB) to discover relevant operations from each other, additional complexity and challenges were found during X4 testing and illuminated the need for a more suitable solution for UAM. Industry can leverage the results of this demonstration to accelerate UAM requirements, CBRs, and standards development. The FAA and other government municipalities and agencies will be able to leverage results to inform future policies and identify additional gaps that require further analysis, moving toward operationalization of UAM.

Urban Air Mobility↗
DOE OSTI2023

Plant science corpus

The plant science corpus consists of the titles and abstracts of plant science articles in PubMed published prior to 2021 with a small number of 2021 records due to modification of records.

Botany/trend/methods↗
DOE OSTI2023

Multilevel-in-width training for deep neural network regression

A common challenge in regression is that for many problems, the degrees of freedom required for a high-quality solution also allows for overfitting. Regularization is a class of strategies that seek to restrict the range of possible solutions so as to discourage overfitting while still enabling good solutions, and different regularization strategies impose different types of restrictions. In this paper, we present a multilevel regularization strategy that constructs and trains a hierarchy of neural networks, each of which has layers that are wider versions of the previous network's layers. Here, we draw intuition and techniques from the field of Algebraic Multigrid (AMG), traditionally used for solving linear and nonlinear systems of equations, and specifically adapt the Full Approximation Scheme (FAS) for nonlinear systems of equations to the problem of deep learning. Training through V-cycles then encourage the neural networks to build a hierarchical understanding of the problem. We refer to this approach as multilevel-in-width to distinguish from prior multilevel works which hierarchically alter the depth of neural networks. The resulting approach is a highly flexible framework that can be applied to a variety of layer types, which we demonstrate with both fully connected and convolutional layers. We experimentally show with PDE regression problems that our multilevel training approach is an effective regularizer, improving the generalize performance of the neural networks studied.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

Input specific neural networks

Neural networks have emerged as powerful tools for mapping between inputs and outputs. However, their black-box nature limits the ability to encode or impose specific structural relationships between inputs and outputs. Many scientific and engineering problems, such as constitutive modeling in solid mechanics, require networks that can enforce convexity, monotonicity, or other structural constraints to ensure physical consistency. Here, we introduce the Input Specific Neural Network (ISNN), a new architecture that enables multiple, distinct constraints to be imposed on different input subsets for scalar-valued outputs. This framework unifies convex, monotone–convex, monotone, and arbitrary mappings within a single network for the first time. Two ISNN architectures with analytical first- and second-order derivatives are developed. We demonstrate the performance on synthetic toy problems, inverse problems in isotropic hyperelasticity, and finite element simulations. ISNNs achieve improved extrapolation behavior, require fewer invariant inputs than standard input convex networks for polyconvex potentials, and enable significant computational savings via manual differentiation. We also show how ISNNs can be used to learn structural relationships between inputs and outputs via a binary gating mechanism. Particularly, ISNNs are employed to model a homogenized anisotropic free energy potential in a decoupled multiscale setting. The network learns whether or not the potential should be modeled as polyconvex and retains only the relevant layers while using the minimum number of inputs. ISNNs provide a flexible foundation for embedding structural priors into neural networks, enhancing both interpretability and stability. They are broadly applicable across computational mechanics and other scientific domains requiring constrained functional relationships.

Jadoon, Asghar A. [Univ. of Texas, Austin, TX (Uni↗
DOE OSTI2020

Basic Design Layout of a Modular Pumped Hydro Teaching Unit

This work was originally completed in July 2015 as a proposal to Santa Fe Community College (SFCC) for their engineering program and a follow up to a feasibility study completed for the Community College the year earlier as indicated below. The work was ahead of its time and funding was hard to obtain to implement a project. This work is now being formally released for distribution to a wider audience based on the evolving circumstances in greater society that now favor this approach with funding more readily available. Due to the distribution outside of SFCC, specific architectural drawing sets of the pertinent buildings called out in the design have been removed. No changes to the design have been made. The design layout utilizing a building is not meant to limit the implementation of the design. The elevation difference required may be obtained with a building, a small hill, jacking up the upper tank, or similar means. The design presented is a basic design, identifying major components and how they are connected together. The user may change components under their own liability. All local applicable codes and standards must be followed when building an actual unit, under the authority of a responsible engineer or owner. Appropriate measures must be followed to protect people from hazards such as electricity, water pressure, and general construction. A leak test to verify water tightness of the system should be conducted prior to any operations. The vision of this design is to allow a class of 10 to 15 students to construct the assembly within a month or so time period, and then learn to operate the assembly at their school location under the guidance of their instructor. This will allow the students to learn about the practical concept of storing electricity with water, and gain a hands on experience of the real world challenges faced with full Modular Pumped Hydro Projects.

13 HYDRO ENERGY↗
NASA NTRS1990

Maps or analogies? A comparison of instructional aids for menu navigation

The relative efficacy of a variety of spatial maps and analogies as instructional aids for the learning and navigation of hierachical data bases is studied. Subjects performed single- and multinode retrieval tasks with novel data bases. The results of one experiment indicated that subjects who studied an analogy prior to performing data-retrieval tasks were less affected by increasing search distance than were subjects who studied a spatial map. The second experiment suggested that the benefits of analogy relative to other instructional aids increased as time passed between initial instruction and interaction with the data base.

Webb, Jayson M.↗