SEARCH · Engineering Papers

Results for “Adaptation models”

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 1,405 records · Page 78

All sources NASA NTRS DOE OSTI

DOE OSTI2022

DeepAdversaries: examining the robustness of deep learning models for galaxy morphology classification

With increased adoption of supervised deep learning methods for work with cosmological survey data, the assessment of data perturbation effects (that can naturally occur in the data processing and analysis pipelines) and the development of methods that increase model robustness are increasingly important. In the context of morphological classification of galaxies, we study the effects of perturbations in imaging data. In particular, we examine the consequences of using neural networks when training on baseline data and testing on perturbed data. We consider perturbations associated with two primary sources: (a) increased observational noise as represented by higher levels of Poisson noise and (b) data processing noise incurred by steps such as image compression or telescope errors as represented by one-pixel adversarial attacks. We also test the efficacy of domain adaptation techniques in mitigating the perturbation-driven errors. We use classification accuracy, latent space visualizations, and latent space distance to assess model robustness in the face of these perturbations. For deep learning models without domain adaptation, we find that processing pixel-level errors easily flip the classification into an incorrect class and that higher observational noise makes the model trained on low-noise data unable to classify galaxy morphologies. On the other hand, we show that training with domain adaptation improves model robustness and mitigates the effects of these perturbations, improving the classification accuracy up to 23% on data with higher observational noise. Domain adaptation also increases up to a factor of ${\approx}2.3$ the latent space distance between the baseline and the incorrectly classified one-pixel perturbed image, making the model more robust to inadvertent perturbations. Successful development and implementation of methods that increase model robustness in astronomical survey pipelines will help pave the way for many more uses of deep learning for astronomy.

79 ASTRONOMY AND ASTROPHYSICS↗

NASA NTRS1994

Near-surface ice on Mercury and the Moon: A topographic thermal model

A thermal model that can be easily adapted to craters of arbitrary shape is developed and applied to high-latitude impact craters on Mercury and the Moon, Chao Meng Fu crater at -87.5 deg L on Mercury, an unnamed bowl-shaped crater at 86.7 deg L on Mercury, and Peary crater at 88.6 deg L on the Moon. For an assumed input topography and grid of surface elements, the model computes for each element the irradiation from direct insolation and reflected and emitted radiation from other elements, taking into account shadowing by walls of the crater, partial obscuration of the solar disk near the poles and the diurnal, orbital, and seasonal cycles. Temperatures are computed over the surface grid as functions of depth and time from the surface to a specified depth and over the pertinent astronomical cycles, including the effects of direct and indirect surface irradiation, infrared radiation, heat conduction, and interior heating. Vapor fluxes and ice recession times are computed as functions of ice depth over the surface grid. Temperatures profiles, vapor fluxes, and ice recession times were computed for flat surfaces not associated with craters near the poles of Mercury and the Moon. It was found that water ice could have existed throughout geologic time within the maximum radar detection depth of recent observation of Mercury (J. K. Harmon and M. A. Slade, 1992, Science 258, 640-643) poleward of approximately 87 - 88 deg L on Mercury and poleward of approximately 73 deg L on the Moon. For Chao Meng Fu crater it was found that approximately 40% of the crater floor is permanently shadowed from direct solar insolation, while the remainder of the crater floor is periodically illuminated by a partially obscured Sun. Temperatures at the upper levels of the south wall can slightly exceed 550 K. Surface temperatures in the permanently shadowed region of the crater floor are under approximately 130 K, which could have allowed water ice to exist throughout geologic time within the radar detection depth of recent observation of Mercury. For small bowl-shaped crater on Mercury, it was found that most of the crater is permanently shadowed from direct solar radiation, except for a narrow semicircular band bordering the north rim. However, temperatures in the permanently shadowed region periodically reach a maximum near approximately 315 K due to efficient heating of the small crater by thermal emission and reflection from the small sunlit region, which periodically reaches temperatures exceeding 630 K. Water ice could not have existed throughout geologic time anywhere in this crater within the radar detection depth. For Peary crater on the Moon, the entire crater floor is permanently shadowed from direct solar insolation with maximum temperature under 120 K. The upper level of the north wall periodically reaches a maximum temperature near 310 K. The low temperatures on the crater floor would have allowed water ice to exist near the surface throughout geologic time, provided that the Moon's obliquity was always as low as it is at present.

Salvail, James R.↗

NASA NTRS1997

Analyzing Tabular and State-Transition Requirements Specifications in PVS

We describe PVS's capabilities for representing tabular specifications of the kind advocated by Parnas and others, and show how PVS's Type Correctness Conditions (TCCs) are used to ensure certain well-formedness properties. We then show how these and other capabilities of PVS can be used to represent the AND/OR tables of Leveson and the Decision Tables of Sherry, and we demonstrate how PVS's TCCs can expose and help isolate errors in the latter. We extend this approach to represent the mode transition tables of the Software Cost Reduction (SCR) method in an attractive manner. We show how PVS can check these tables for well-formedness, and how PVS's model checking capabilities can be used to verify invariants and reachability properties of SCR requirements specifications, and inclusion relations between the behaviors of different specifications. These examples demonstrate how several capabilities of the PVS language and verification system can be used in combination to provide customized support for specific methodologies for documenting and analyzing requirements. Because they use only the standard capabilities of PVS, users can adapt and extend these customizations to suit their own needs. Those developing dedicated tools for individual methodologies may find these constructions in PVS helpful for prototyping purposes, or as a useful adjunct to a dedicated tool when the capabilities of a full theorem prover are required. The examples also illustrate the power and utility of an integrated general-purpose system such as PVS. For example, there was no need to adapt or extend the PVS model checker to make it work with SCR specifications described using the PVS TABLE construct: the model checker is applicable to any transition relation, independently of the PVS language constructs used in its definition.

Owre, Sam↗

DOE OSTI2025

Multiclass Classification Using Bayesian Multivariate Adaptive Regression Splines

We present a new Bayesian model for the problem of multiclass classification. In this model, the probabilities of class membership of a given observation are determined by the mean of a latent Gaussian distribution. The mean functions of this latent distribution consist of combinations of highly flexible basis functions of the inputs: multivariate adaptive regression splines (MARS), first developed for multiple regression. We use reversible jump Markov chain Monte Carlo to make inference on the classification model, including the number of basis functions. We compare the probabilistic classification performance of our proposed approach to existing methods on simulated and benchmark data, and compare uncertainty estimates on simulated data. Our proposed method compares favorably with existing Bayesian and frequentist multiclass classification methods in out-of-sample probabilistic classification, and uncertainty estimation of these probabilistic classifications. We examine the fit of the proposed method to a data set of hurricane storm surge levels near Delaware Bay, US, and conclude that sea level rise is a key contributor to damage delivered by storm surge.

97 MATHEMATICS AND COMPUTING↗

NASA NTRS2019

Global Crop Production: Adaptation to Temperature Increase Is Possible by Maintaining the Current Growing Periods

Increasing temperature trends are expected to impact yields of major field crops by affecting various plant processes, such as phenology and growth. However, future projections, especially at the global scale, do not consider any agronomic adaptation in farming practices. We use an ensemble of seven Global Gridded Crop Models (GGCMs) contributing to the AgMIP-Gridded project to quantify the impacts and adaptation potential of field crops under increasing temperature. We study how uniform warming scenarios up to 6 K affect the productivity and growing period duration of five major crops. The design of the experiment aims at understanding adaptation measures targeted to temperature-driven impacts and therefore leaves out changes in any other climate variable. We find that, without adaptation the dominant effect of temperature increase is to shorten the growing period and to reduce yields and production, consistently across crops and regions. We then test the potential of two agronomic measures to contrast warming-induced yield reduction, assuming that (i) cultivars with adjusted phenology would be used to regain the reference growing period duration; (ii) rainfed systems would be converted to irrigated. Despite substantial uncertainties in model parametrization and little agreement in spatial patterns, at the global aggregation, model results are robust. We find that up to 3 K of temperature increase, production losses could be fully compensated by maintaining the original crop growing period. Irrigation would also compensate production losses, but would not reduce the temperature impacts. Across regions, the model ensemble estimates larger adaptation potentials from unchanged growing periods in continental and temperate regions, than in tropical and arid regions, where also irrigation has respectively little effects and availability.

Irrigation↗

NASA NTRS2004

Dynamical Modeling of Mars' Paleoclimate

This report summarizes work undertaken under a one-year grant from the NASA Mars Fundamental Research Program. The goal of the project was to initiate studies of the response of the Martian climate to changes in planetary obliquity and orbital elements. This work was undertaken with a three-dimensional numerical climate model based on the Geophysical Fluid Dynamics Laboratory (GFDL) Skyhi General Circulation Model (GCM). The Mars GCM code was adapted to simulate various obliquity and orbital parameter states. Using a version of the model with a basic water cycle (ice caps, vapor, and clouds), we examined changes in atmospheric water abundances and in the distribution of water ice sheets on the surface. This work resulted in a paper published in the Journal of Geophysical Research - Planets. In addition, the project saw the initial incorporation of a regolith water transport and storage scheme into the model. This scheme allows for interaction between water in the pores of the near subsurface (<3m) and the atmosphere. This work was not complete by the end of the one-year grant, but is now continuing within the auspices of a three-year grant of the same title awarded by the Mars Fundamental Research Program in late 2003.

Richardson, Mark I.↗

NASA NTRS2010

Classification of Clouds and Deep Convection from GEOS-5 Using Satellite Observations

With the increased resolution of global atmospheric models and the push toward global cloud resolving models, the resemblance of model output to satellite observations has become strikingly similar. As we progress with our adaptation of the Goddard Earth Observing System Model, Version 5 (GEOS-5) as a high resolution cloud system resolving model, evaluation of cloud properties and deep convection require in-depth analysis beyond a visual comparison. Outgoing long-wave radiation (OLR) provides a sufficient comparison with infrared (IR) satellite imagery to isolate areas of deep convection. We have adopted a binning technique to generate a series of histograms for OLR which classify the presence and fraction of clear sky versus deep convection in the tropics that can be compared with a similar analyses of IR imagery from composite Geostationary Operational Environmental Satellite (GOES) observations. We will present initial results that have been used to evaluate the amount of deep convective parameterization required within the model as we move toward cloud system resolving resolutions of 10- to 1-km globally.

Putman, William↗

NASA NTRSDate not supplied

How Eye Drift Adapts to the History of Eye Positions

Oculomotor behavior is a powerful tool to probe brain function and disease. Previous studies have shown how fixational drift velocities depend on eccentricity (Bertolini et al., 2013; Romano et al., 2017) and how fixational drift can adapt after prolonged eccentric fixation (Otero-Millan et al., 2019). Here, eighteen subjects completed two experiments where we measured drift velocity while fixating a briefly flashed target, to remove retinal slip cues. In the first experiment, we measured their baseline relationship between fixational drift velocity and eye position. In the second experiment, we investigated the adaptation of that relationship after prolonged fixation at a very eccentric position. This adaptation usually manifests by the presence of rebound nystagmus (RN) upon returning to central gaze. Our results replicate the finding that drift increases at central fixation in the direction consistent with rebound nystagmus, that is, towards the previously held eccentric position, highlighting that drift velocity depends on the history of eye positions. When examining the entire relationship between drift velocity and eye position we find a mostly linear relationship but with increased slope relative to the baseline condition. However, the adaptation of the integrator cannot be modeled by a simple change in leakiness, as that would not result in increased drift at central gaze. Other mechanisms of adaptation such as shifts in the position or velocity set-points for gaze holding could be present but are indistinguishable with the current data.

eye movement↗

DOE OSTI2021

Grid-Interactive Multi-Zone Building Control Using Reinforcement Learning with Global-Local Policy Search: Preprint

In this paper, we develop a grid-interactive multi-zone building controller based on a deep reinforcement learning (RL) approach. The controller is designed to facilitate building operation during normal conditions and demand response events, while ensuring occupants comfort and energy efficiency. We leverage a continuous action space RL formulation, and devise a two-stage global-local RL training framework. In the first stage, a global fast policy search is performed using a gradient-free RL algorithm. In the second stage, a local fine-tuning is conducted using a policy gradient method. In contrast to the state-of-the-art model predictive control (MPC) approach, the proposed RL controller does not require complex computation during real-time operation and can adapt to non-linear building models. We illustrate the controller performance numerically using a five-zone commercial building.

30 DIRECT ENERGY CONVERSION↗

DOE OSTI2026

Unsupervised domain adaptation for radioisotope identification in gamma spectroscopy

Training machine learning models for radioisotope identification using gamma spectroscopy remains an elusive challenge for many practical applications, largely stemming from the difficulty of acquiring and labeling large, diverse experimental datasets. Simulations can mitigate this challenge, but the accuracy of models trained on simulated data can deteriorate substantially when deployed to an out-of-distribution operational environment. In this study, we demonstrate that unsupervised domain adaptation (UDA) can improve the ability of a model trained on synthetic data to generalize to a new testing domain, provided unlabeled data from the target domain are available. Conventional supervised techniques are unable to utilize this data because the absence of isotope labels precludes defining a supervised classification loss. Instead, we first pretrain a spectral classifier using labeled synthetic data and subsequently leverage unlabeled target data to align the learned feature representations between the source and target domains. We compare a range of different UDA techniques, finding that minimizing the maximum mean discrepancy (MMD) between source and target feature vectors yields the most consistent improvement to testing scores. For instance, using a custom transformer-based neural network, we achieved a testing accuracy of $0.904 \pm 0.022$ on an experimental LaBr test set after performing unsupervised feature alignment via MMD minimization, compared to $0.754 \pm 0.014$ before alignment. Overall, our results highlight the potential of using UDA to adapt a radioisotope classifier trained on synthetic data for real-world deployment.

Lalor, Peter W.↗

DOE OSTI2023

Adaptation Strategies Strongly Reduce the Future Impacts of Climate Change on Simulated Crop Yields

Abstract Simulations of crop yield due to climate change vary widely between models, locations, species, management strategies, and Representative Concentration Pathways (RCPs). To understand how climate and adaptation affects yield change, we developed a meta‐model based on 8703 site‐level process‐model simulations of yield with different future adaptation strategies and climate scenarios for maize, rice, wheat and soybean. We tested 10 statistical models, including some machine learning models, to predict the percentage change in projected future yield relative to the baseline period (2000–2010) as a function of explanatory variables related to adaptation strategy and climate change. We used the best model to produce global maps of yield change for the RCP4.5 scenario and identify the most influential variables affecting yield change using Shapley additive explanations. For most locations, adaptation was the most influential factor determining the projected yield change for maize, rice and wheat. Without adaptation under RCP4.5, all crops are expected to experience average global yield losses of 6%–21%. Adaptation alleviates this average projected loss by 1–13 percentage points. Maize was most responsive to adaptive practices with a projected mean yield loss of −21% [range across locations: −63%, +3.7%] without adaptation and −7.5% [range: −46%, +13%] with adaptation. For maize and rice, irrigation method and cultivar choice were the adaptation types predicted to most prevent large yield losses, respectively. When adaptation practices are applied, some areas are predicted to experience yield gains, especially at northern high latitudes. These results reveal the critical importance of implementing adequate adaptation strategies to mitigate the impact of climate change on crop yields.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2023

Off-policy deep reinforcement learning with automatic entropy adjustment for adaptive online grid emergency control

Electric overloading conditions and contingencies put modern power systems at risk of voltage collapse and blackouts. Load shedding is crucial to maintain voltage stability for grid emergency control. However, the rule- or model-based schemes rely on accurate dynamic system models and face considerable challenges in adapting to various operating conditions and uncertain event occurrences. Here, to address these issues, this paper proposes a novel deep reinforcement learning (DRL)-based voltage stability control algorithm with automatic entropy adjustment (AEA) for grid emergency control. Various dynamic network components for complex system operations are modeled to construct the DRL environment. An off-policy soft actor-critic architecture is developed to maximize the expected reward and policy entropy simultaneously. The AEA mechanism is proposed to facilitate the policy maximum entropy procedure, and the proposed method can automatically provide effective discrete and continuous actions against various fault scenarios. Our approach accomplishes high sampling efficiency, scalability, and auto-adaptivity of the control policies under high uncertainties. Comparative studies with the existing DRL-based control methods in IEEE benchmarks indicate salient performance improvement of the proposed method for dynamic system emergency control.

24 POWER TRANSMISSION AND DISTRIBUTION↗

NASA NTRS2017

OVERFLOW Contribution to HiLiftPW-3

We plan to perform the following sets of computations: For all our contributions (except where stated) Code: OVERFLOW, Turbulence model: SAnegRCQCR2000. - 1. Results will be submitted for both the full chord flap gap (Case 1a) and partially-sealed Chord Flap gap (Case 1c): 1. Grid Refinement Study; 2. Grids: structured overset grids supplied by HiLiftPW committee; 3. Connectivity: Domain Connectivity Framework, DCF. - 2. Results will be submitted for JAXA Standard Model and Nacelle/Pylon Off (Case 2a), Nacelle/Pylon On (Case 2c): 1. Alpha Study; 2. Grids: structured overset grids supplied by HiLiftPW committee; 3. Connectivity: Pegasus 5 (Peg5). - 3. A study of the effects of different connectivity paradigms: 1. DCF vs Peg5 for HLCRM cases; 2. DCF vs. C3P (NASA Ames) vs. Peg5 for JSM cases; 3. JSM grids will be the focus where we will hopefully see some type of trends with reference to wind tunnel data. - 4. Adaption cases will be attempted for (and submitted where appropriate): 1. Cases 1c,1d: HLCRM; 2. Cases 2c and 2d: JSM; 3. Grid: Near Body grids provided by committee, OffBody grids Cartesian; 4. AMR NearBody and OffBody Adaption. - 5. Case 3 Turbulence model verification study: 1. Grid: Series of 3 finest grids as defined on http://turbmodels.larc.nasa.gov/airfoilwakeverif.html; 2. Turbulence models: SAneg and SAneg— RCQCR2000. OVERFLOW 2.2 is a Reynolds-averaged Navier-Stokes (RANS) code developed by NASA...

OVERFLOW↗

DOE OSTI2025

Scalability analysis of heavy-duty gas turbines using data-driven machine learning

With the increasing integration of variable renewable energy sources into power systems, the role of flexible power generation technologies like gas turbines (GT) in rapid grid balancing remains crucial. This sustained importance underscores the need for scaled and precise modeling of GT to ensure effective integration within evolving energy frameworks. While physics-driven GT models integrate thermodynamics, fluid dynamics, and combustion principles, they often rely on approximate mathematical representations to accommodate scaling that may not capture the actual complex dynamics for GTs and inertial effects associated to GTs with different ratings. In this study, a data-driven model is proposed using machine learning (ML) techniques to conduct GT scalability analysis and performance evaluation with high accuracy. The ML model, trained on data from various operating conditions and performance parameters, aims to uncover intricate relationships and patterns, resembling GT characteristics at different scales (ratings). The model is developed to capture complex system interaction and to adapt to changing operational scenarios at different capacities, providing valuable insights of power system dynamics. In this study, the real-time digital simulator platform was employed to generate training data for the ML model and assess its dynamic characteristics. The ultimate objective was to develop a detailed modeling framework based on governing equations and data-driven ML capable of predicting key performance indicators, in thermal systems such as GTs, including power output, speed, fuel consumption, and exhaust temperature under diverse operating conditions at different scales. The developed ML framework demonstrated high accuracy, with mean relative errors for GT power prediction, reference speed, exhaust temperature, and compressor pressure ratio (CPR) parameters consistently below 0.1% across typical load fluctuation scenarios. Maximum deviations were limited to approximately 0.5 K for exhaust temperature and 0.009 for CPR, underscoring the model’s ability to replicating dynamic GT behavior with high precision. The adaptability of the ML model enables its application across diverse operational conditions and its extension to other thermal systems. By leveraging advanced ML techniques, this study presents a robust and scalable modeling framework that enhances GT simulation precision, facilitating improved integration into evolving power systems.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2023

The importance of capturing power system operational details in resource adequacy assessments

Traditional methods for assessing the resource adequacy (RA) of a power system are becoming obsolete due to emerging trends such as the increasing deployment of variable renewable energy and storage. Consequently, analysts are recommending that RA be assessed using a Monte Carlo simulation approach that models chronological power system operations over many instances of possible operating conditions. However, this approach is necessarily more complex and computationally demanding, which is an obstacle to real-world implementation. Here, in this study, we investigate which operational details of power systems are important to capture in order to accurately evaluate a system's RA, versus details that add complexity but do not meaningfully affect RA results. To do so, we develop a probabilistic RA assessment framework by adapting an existing production cost model and apply it to a case study based on the IEEE Reliability Test System. Our results indicate that multi-year data, storage dispatch, and transmission limits are key details to incorporate. Accurate RA results can be obtained using non-economic dispatch strategies as long as they are coordinated with detailed operational strategies. We also demonstrate how popular expectation-based RA metrics can mask important differences in the characteristics of loss of load events.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

DOE OSTI2020

Predictive Skill of Deep Learning Models Trained on Limited Sequence Data

In this report we investigate the utility of one-dimensional convolutional neural network (CNN) models in epidemiological forecasting. Deep learning models, especially variants of recurrent neural networks (RNNs) have been studied for influenza forecasting, and have achieved higher forecasting skill compared to conventional models such as ARIMA models. In this study, we adapt two neural networks that employ one-dimensional temporal convolutional layers as a primary building block temporal convolutional networks and simple neural attentive meta-learner for epidemiological forecasting and test them with influenza data from the US collected over 2010-2019. We find that epidemiological forecasting with CNNs is feasible, and their forecasting skill is comparable to, and at times, superior to, RNNs. Thus CNNs and RNNs bring the power of nonlinear transformations to purely data-driven epidemiological models, a capability that heretofore has been limited to more elaborate mechanistic/compartmental disease models.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

NASA NTRS2023

An Interface Specification for Urban Air Mobility Performance Models to Support Air Traffic Management Research

Performance modeling of Urban Air Mobility vehicles in support of Air Traffic Management research poses new challenges. These aircraft often rely on a combination of fixed-wing and rotorcraft performance capabilities to enable a new and still-evolving concept of high-volume operations in densely populated areas. These characteristics lead to variability in performance model format and functionality, complicating model development and leading to difficulties for users integrating the models into their applications. This paper describes an interface specification for performance models that is intended to help address this issue. The interface aims to support the core functionality of performance models while also providing the necessary flexibility to both model developers and users. The specification describes the required model documentation, required inputs to the model, and required outputs from the model in general terms that are adaptable to most programming languages. The utilization of an appropriate interface specification will support the development of an Urban Air Mobility performance model database and help to improve interoperability of these models for a broad array of user applications.

Trajectory Prediction↗

NASA NTRS2023

An Interface Specification for Urban Air Mobility Performance Models to Support Air Traffic Management Research

Trajectory Prediction↗