Search NASA⌕ Search

SEARCH · Search NASA

Results for “Deep uncertainty”

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 145 records · Page 8
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

Uncertainty based Online Ensemble on Non-Stationary Data for Fusion Science

Machine Learning (ML) is poised to play a pivotal role in the development and operation of next-generation fusion devices. Fusion data shows non-stationary behavior due to drifts in the data. The drifts can arise from both experimental evolution and machine wear-and-tear. ML models assume stationary distribution and fail to maintain performance when encountered with non-stationary data streams.Online learning can be used to continuously adapt the models with new data as it is acquired. However, traditional online learning can suffer from short-term performance degradation, as ground truth are not available before making the prediction. To address this challenge, we propose uncertainty aware ensemble approach for online learning. We use Deep Gaussian Process Approximation (DGPA) technique for calibrated uncertainty estimation and use the uncertainty values to guide a meta-algorithm that produces predictions based on ensemble of learners. Moreover, DGPA also provides uncertainty estimation along with the predictions for decision makers. This paper demonstrates that the proposed method outperforms traditional online learning approach, and a naive ensemble without uncertainty guidance by about 7% and 6%, respectively, on B-coil deflection prediction at DIII-D Fusion Facility.

Rajput, Kishansingh [Thomas Jefferson National Acc↗
DOE OSTI2025

Uncertainty based Online Ensemble on Non-Stationary Data for Fusion Science

Machine Learning (ML) is poised to play a pivotal role in the development and operation of next-generation fusion devices. Fusion data shows non-stationary behavior due to drifts in the data. The drifts can arise from both experimental evolution and machine wear-and-tear. ML models assume stationary distribution and fail to maintain performance when encountered with non-stationary data streams.Online learning can be used to continuously adapt the models with new data as it is acquired. However, traditional online learning can suffer from short-term performance degradation, as ground truth are not available before making the prediction. To address this challenge, we propose uncertainty aware ensemble approach for online learning. We use Deep Gaussian Process Approximation (DGPA) technique for calibrated uncertainty estimation and use the uncertainty values to guide a meta-algorithm that produces predictions based on ensemble of learners. Moreover, DGPA also provides uncertainty estimation along with the predictions for decision makers. This paper demonstrates that the proposed method outperforms traditional online learning approach, and a naive ensemble without uncertainty guidance by about 7% and 6%, respectively, on B-coil deflection prediction at DIII-D Fusion Facility.

Rajput, Kishansingh [Thomas Jefferson National Acc↗
NASA NTRS2023

A Pilot Project for Quantifying the Effect of Medical Provider Knowledge, Skills, and Abilities on Outcomes for Spaceflight Using a Probabilistic Risk Assessment Tool

In order to enable the future of long-duration deep space exploration we must confront the uncertainty in medical risk. Limitations of communication, resupply, and evacuation in deep space will require a high degree of crew autonomy and accurate risk assessment will be critical to ensure adequate crew training and medical system design. To address this, NASA’s Human Research Program Exploration Medical Capability Element has developed the Informing Mission Planning via Analysis of Complex Tradespaces Tool (IMPACT). IMPACT is a suite of tools that can provide evidence-based, data-driven trade space assessments between available medical resources in the mass- and volume-constrained environment of a deep space exploration vehicle. In the current model, medical conditions either can or cannot be treated based on the availability of medical system resources and equipment. However, medical outcomes often depend just as much on the knowledge, skills, and abilities (KSA) of the provider operating the system. This paper presents a method for modeling and quantifying the effect of medical officer KSA on medically relevant mission risk outcomes during spaceflight.

capabilities↗
DOE OSTI2021

Deep in the Sierra Nevada critical zone: saprock represents a large terrestrial organic carbon stock

Large uncertainty remains in the spatial distribution of deep soil organic carbon (OC) storage and how climate controls belowground OC. This research aims to quantify OC stocks, characterize soil OC age and chemical composition, and evaluate climatic impacts on OC storage from the soil surface through the deep critical zone to bedrock. These objectives were carried out at four sites along a bio-climosequence in the Sierra Nevada, California. On average, 74% of OC was stored below the A horizon, and up to 30% of OC was stored in saprock (friable weakly weathered bedrock). Radiocarbon, spectroscopic, and isotopic analyses revealed the coexistence of very old organic matter (OM) (mean radiocarbon age = 20 300 years) with relatively recent OM (mean radiocarbon age = 4800 years) and highly decomposed organic compounds with relatively less decomposed material in deep soil and saprock. This co-mingling of OM suggests OC is prone to both active cycling and long-term protection from degradation. In addition to having direct effects on OC cycling, climate indirectly controls deep OC storage through its impact on the degree of regolith weathering (e.g. thickening). Although deep OC concentrations are low relative to soil, thick saprock represents a large, previously unrealized OC pool.

58 GEOSCIENCES↗
NASA NTRS2015

A Reference Field for GCR Simulation and an LET-Based Implementation at NSRL

Exposure to galactic cosmic rays (GCR) on long duration deep space missions presents a serious health risk to astronauts, with large uncertainties connected to the biological response. In order to reduce the uncertainties and gain understanding about the basic mechanisms through which space radiation initiates cancer and other endpoints, radiobiology experiments are performed. Some of the accelerator facilities supporting such experiments have matured to a point where simulating the broad range of particles and energies characteristic of the GCR environment in a single experiment is feasible from a technology, usage, and cost perspective. In this work, several aspects of simulating the GCR environment in the laboratory are discussed. First, comparisons are made between direct simulation of the external, free space GCR field and simulation of the induced tissue field behind shielding. It is found that upper energy constraints at the NASA Space Radiation Laboratory (NSRL) limit the ability to simulate the external, free space field directly (i.e. shielding placed in the beam line in front of a biological target and exposed to a free space spectrum). Second, variation in the induced tissue field associated with shielding configuration and solar activity is addressed. It is found that the observed variation is within physical uncertainties, allowing a single reference field for deep space missions to be defined. Third, an approach for simulating the reference field at NSRL is presented. The approach allows for the linear energy transfer (LET) spectrum of the reference field to be approximately represented with discrete ion and energy beams and implicitly maintains a reasonably accurate charge spectrum (or, average quality factor). Drawbacks of the proposed methodology are discussed and weighed against alternative simulation strategies. The neutron component and track structure characteristics of the proposed strategy are discussed in this context.

Slaba, Tony C.↗
DOE OSTI2025

Comprehensive assessment of deep reinforcement learning approaches for economic dispatch in nuclear-driven microgrids

As the electrical grid integrates more variable renewable energy sources such as wind and solar, the demand for distributed and flexible systems to address this increased variability becomes critical. Nuclear-driven microgrids provide a promising solution by offering stable generation to complement intermittent renewables, ensuring grid reliability and operating efficiency. This paper proposes a recurrent deep reinforcement learning framework for optimal economic dispatch in a nuclear-powered microgrid integrating renewable energy sources, small modular reactors, battery storage systems, and balance-of-plant dynamics. A three-agent control architecture is developed, where demand and renewable energy agents act as forecasters, and a reinforcement learning-based dispatch agent performs real-time energy allocation. A nonlinear programming formulation is first used to generate an optimal baseline for benchmarking. The proposed dispatch controller, based on Proximal Policy Optimization enhanced with Long Short-Term Memory networks, exploits temporal correlations in system dynamics by taking advantage of the time series used as inputs to improve policy robustness under uncertainty. Comparative analysis against established deep reinforcement learning methods, including Proximal Policy Optimization with a feedforward architecture, Soft Actor-Critic, and Twin Delayed Deep Deterministic Policy Gradient, demonstrates superior performance. Numerical results indicate that the proposed controller achieves a 0.39% cost reduction relative to the nonlinear programming benchmark and outperforms other learning-based methods by generating additional revenue of up to 0.35%. All reinforcement learning controllers compute dispatch actions in less than 0.3 s, resulting in a computational speedup of more than three orders of magnitude over the nonlinear programming baseline. The findings of this paper highlight their applicability for real-time operation and control in nuclear-integrated microgrids under volatile operating conditions.

24 POWER TRANSMISSION AND DISTRIBUTION↗
NASA NTRS2014

NASA Team 2 Sea Ice Concentration Algorithm Retrieval Uncertainty

Satellite microwave radiometers are widely used to estimate sea ice cover properties (concentration, extent, and area) through the use of sea ice concentration (IC) algorithms. Rare are the algorithms providing associated IC uncertainty estimates. Algorithm uncertainty estimates are needed to assess accurately global and regional trends in IC (and thus extent and area), and to improve sea ice predictions on seasonal to interannual timescales using data assimilation approaches. This paper presents a method to provide relative IC uncertainty estimates using the enhanced NASA Team (NT2) IC algorithm. The proposed approach takes advantage of the NT2 calculations and solely relies on the brightness temperatures (TBs) used as input. NT2 IC and its associated relative uncertainty are obtained for both the Northern and Southern Hemispheres using the Advanced Microwave Scanning Radiometer for the Earth Observing System (AMSR-E) TB. NT2 IC relative uncertainties estimated on a footprint-by-footprint swath-by-swath basis were averaged daily over each 12.5-km grid cell of the polar stereographic grid. For both hemispheres and throughout the year, the NT2 relative uncertainty is less than 5%. In the Southern Hemisphere, it is low in the interior ice pack, and it increases in the marginal ice zone up to 5%. In the Northern Hemisphere, areas with high uncertainties are also found in the high IC area of the Central Arctic. Retrieval uncertainties are greater in areas corresponding to NT2 ice types associated with deep snow and new ice. Seasonal variations in uncertainty show larger values in summer as a result of melt conditions and greater atmospheric contributions. Our analysis also includes an evaluation of the NT2 algorithm sensitivity to AMSR-E sensor noise. There is a 60% probability that the IC does not change (to within the computed retrieval precision of 1%) due to sensor noise, and the cumulated probability shows that there is a 90% chance that the IC varies by less than +/-3%. We also examined the daily IC variability, which is dominated by sea ice drift and ice formation/melt. Daily IC variability is the highest, year round, in the MIZ (often up to 20%, locally 30%). The temporal and spatial distributions of the retrieval uncertainties and the daily IC variability is expected to be useful for algorithm intercomparisons, climate trend assessments, and possibly IC assimilation in models.

Sea Ice↗
DOE OSTI2022

Accelerated Probabilistic Marching Cubes by Deep Learning for Time-Varying Scalar Ensembles

Visualizing the uncertainty of ensemble simulations is challenging due to the large size and multivariate and temporal features of en-semble data sets. One popular approach to studying the uncertainty of ensembles is analyzing the positional uncertainty of the level sets. Probabilistic marching cubes is a technique that performs Monte Carlo sampling of multivariate Gaussian noise distributions for positional uncertainty visualization of level sets. However, the technique suffers from high computational time, making interactive visualization and analysis impossible to achieve. This paper introduces a deep-learning-based approach to learning the level-set uncertainty for two-dimensional ensemble data with a multivariate Gaussian noise assumption. We train the model using the first few time steps from time-varying ensemble data in our workflow. We demonstrate that our trained model accurately infers uncertainty in level sets for new time steps and is up to 170X faster than that of the original probabilistic model with serial computation and 10X faster than that of the original parallel computation.

Han, Mengjiao↗
DOE OSTI2025

Learning and discovering multiple solutions using physics-informed neural networks with random initialization and deep ensemble

In this work we explore the capability of physics-informed neural networks (PINNs) to discover multiple solutions. Many real-world phenomena governed by nonlinear differential equations (DEs), such as fluid flow, exhibit multiple solutions under the same conditions, yet capturing this solution multiplicity remains a significant challenge. A key difficulty lies in providing appropriate initial conditions or guesses, as widely used time-marching schemes and Newton’s method are highly sensitive to these choices when solving complex computational problems. While machine learning models, particularly PINNs, have shown promise in solving DEs, their ability to capture multiple solutions remains underexplored. In this work, we propose a simple and practical approach using PINNs to learn and discover multiple solutions. We first demonstrate that PINNs, when combined with random initialization and deep ensemble method—originally developed for uncertainty quantification—can effectively uncover multiple solutions to nonlinear ordinary and partial DEs. Although training large ensembles of PINNs may appear computationally demanding, this can be done efficiently using vectorization techniques supported by modern deep learning frameworks, allowing many networks to be trained simultaneously. Our approach highlights the critical role of initialization in shaping solution diversity, addressing an often-overlooked aspect of machine learning for scientific computing. Furthermore, we propose utilizing PINN-generated solutions as initial conditions or initial guesses for conventional numerical solvers to enhance accuracy and efficiency in capturing multiple solutions. Extensive numerical experiments, including the Allen–Cahn equation and cavity flow, where our approach successfully identifies both stable and unstable solutions, validate the effectiveness of our method. These findings establish a general and efficient framework for addressing solution multiplicity in nonlinear DEs.

97 MATHEMATICS AND COMPUTING↗
NASA NTRSDate not supplied

Near Rectilinear Halo Orbit Determination with Simulated DSN Observations

This paper presents the results of a high-fidelity simulation of spacecraft orbit determination in a near rectilinear halo orbit (NRHO). Others in the literature have examined this problem with linear covariance analysis, but the highly-nonlinear dynamics of this orbit challenge the assumptions underlying such analyses. The present work builds on similar analysis performed by other authors to contribute a fuller understanding of the operational requirements for NRHO navigation. The present work serves as a check to the assumptions of previous studies and an independent verification of those results. The results from the literature are extended by quantifying the space of orbital states from which a spacecraft with given control authority can safely return to the nominal path. Spacecraft state uncertainty estimates are evaluated as a function of time. Simulated range and range-rate measurements with the Deep Space Network (DSN) ground stations are used to model orbit determination accuracy. Orbit maintenance maneuvers are performed using both short-horizon and long-horizon stationkeeping targeting. Monte Carlo analysis of orbit determination and stationkeeping is performed. This paper quantifies the achievable state uncertainty with deep space network (DSN)-only range and range-rate observations. This paper also addresses requirements on the frequency of DSN observation periods and correlates ground contact frequency with navigation accuracy. The results of several related studies are presented and discussed: the effect of missing ground station passes, the effect of missing stationkeeping maneuvers, the sensitivity of the spacecraft state estimate to realistic error sources, and stationkeeping propellant budget.

Nathan L Parrish↗
DOE OSTI2022

Enhanced deep neural networks with transfer learning for distribution LMP considering load and PV uncertainties

As the flexibility of generation and demand increases in distribution systems, the residential loads are emerging as a promising means to participate in demand response and the transactive energy market. Market pricing is an instrumental mechanism for the distribution system operator to exploit the full potential of the flexible resources. The distribution locational marginal price (DLMP) can be used to guide the residential load consumption. This type of market signal helps the distribution system operator to optimize the scheduling of all resources while satisfying related network constraints through a day-ahead market. However, solving the optimization problem for large-scale systems can be computationally expensive. To address the scalability and practicability limitations of the DLMP framework, a learning-based approach is proposed in this paper to complement the day-ahead distribution market framework. Here, the proposed approach combines long short-term memory and transfer learning to develop deep neural network that can capture the spatial–temporal correlation of the input data. The model can determine the optimal DLMP for each node in a distribution system without the system parameters required to formulate the optimization problem. Testing results on IEEE 33-bus and 123-bus systems show that the proposed approach can generate a comparable DLMP against the optimization solutions.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2023

Scenario Discovery Analysis of Drivers of Solar and Wind Energy Transitions Through 2050

Deep human-Earth system uncertainties and strong multi-sector dynamics make it difficult to anticipate which conditions are most likely to lead to higher or lower adoption of renewable energy, and models project a broad range of future solar and wind energy shares across future scenarios. To elucidate these dynamics, we explore a large data set of scenarios simulated from the Global Change Analysis Model (GCAM) and use scenario discovery to identify the most significant factors affecting solar and wind adoption by mid-century. We generated a data set of over 4,000 scenarios from GCAM by varying 12 different socioeconomic factors at high and low levels, including assumptions about future energy demand, resource costs, and fossil fuel emissions paths, as well as specific technology assumptions including wind and solar backup requirements and storage costs. Using scenario discovery, we assess the most important factors globally and regionally in creating high fractions of solar and wind energy and explore interconnected effects on other systems including water and non-CO 2 emissions. Globally and regionally, we found that solar and wind-related technology costs were the primary drivers of high wind and solar energy adoption, though a few regions depend heavily on other parameters like carbon capture and storage costs, population and gross domestic product trajectories, and fossil fuel costs. We also identify four key paths to high solar and wind energy by mid-century and discuss their tradeoffs in terms of other outcomes.

14 SOLAR ENERGY↗
DOE OSTI2026

Measurement of Inclusive Charged-Current $\barν_μ$ Scattering on C, CH, Fe, and Pb at $\langle E_{\barν}\rangle \sim$ 6 GeV with MINERvA

We report MINERvA's first measurement of inclusive charged-current $\barν_μ$ cross sections on carbon, hydrocarbon, iron, and lead, and their ratios to the cross section on hydrocarbon, as functions of the antimuon transverse momentum, $p_{\mathrm{T}}$. Using a wide-band $\barν_μ$ beam with mean energy $\sim 6~\text{GeV}$, these measurements probe all interaction modes, including the transition from resonance production to deep-inelastic scattering. The total uncertainties are typically $5-10\%$ for the absolute cross sections and $2-5\%$ for the ratios. Comparisons with multiple neutrino interaction models reveal significant discrepancies in the $p_{\mathrm{T}}$ dependence, particularly for heavier nuclei. The disagreements are most pronounced at low $p_{\mathrm{T}}$ but extend across the full $p_{\mathrm{T}}$ range, indicating missing or mis-modelled nuclear effects.

Klustová, A. [Imperial Coll., London] (ORCID:00000↗
DOE OSTI2021

COVID-19 dynamics across the US: A deep learning study of human mobility and social behavior

This paper presents a deep learning framework for epidemiology system identification from noisy and sparse observations with quantified uncertainty. The proposed approach employs an ensemble of deep neural networks to infer the time-dependent reproduction number of an infectious disease by formulating a tensor-based multi-step loss function that allows us to efficiently calibrate the model on multiple observed trajectories. The method is applied to a mobility and social behavior-based SEIR model of COVID-19 spread. The model is trained on Google and Unacast mobility data spanning a period of 66 days, and is able to yield accurate future forecasts of COVID-19 spread in 203 US counties within a time-window of 15 days. Interestingly, a sensitivity analysis that assesses the importance of different mobility and social behavior parameters reveals that attendance of close places, including workplaces, residential, and retail and recreational locations, has the largest impact on the effective reproduction number. Furthermore, the model enables us to rapidly probe and quantify the effects of government interventions, such as lock-down and re-opening strategies. Taken together, the proposed framework provides a robust workflow for data-driven epidemiology model discovery under uncertainty and produces probabilistic forecasts for the evolution of a pandemic that can judiciously provide information for policy and decision making. All codes and data accompanying this manuscript are available at https://github.com/PredictiveIntelligenceLab/DeepCOVID19.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2025

Evaluation of Seismic Artificial Intelligence with Uncertainty

Artificial intelligence has transformed the seismic community with deep learning models (DLMs) that are trained to complete specific tasks within workflows. However, there is still a lack of robust evaluation frameworks for evaluating and comparing DLMs. Here, we address this gap by designing an evaluation framework that jointly incorporates two crucial aspects: performance uncertainty and learning efficiency. To target these aspects, we meticulously construct the training, validation, and test splits using a clustering method tailored to seismic data and enact an expansive training design to segregate performance uncertainty arising from stochastic training processes and random data sampling. The framework’s ability to guard against misleading declarations of model superiority is demonstrated through the evaluation of PhaseNet (Zhu and Beroza, 2018), a popular seismic phase picking DLM, under three training approaches. Our framework helps practitioners choose the best model for their problem and set performance expectations by explicitly analyzing model performance with uncertainty at varying budgets of training data.

58 GEOSCIENCES↗
DOE OSTI2025

Accurate and uncertainty-aware multi-task prediction of HEA properties using prior-guided deep Gaussian processes

Surrogate modeling techniques have become indispensable in accelerating the discovery and optimization of high-entropy alloys (HEAs), especially when integrating computational predictions with sparse experimental observations. This study systematically evaluates the training and testing performance of four prominent surrogate models—conventional Gaussian processes (cGP), Deep Gaussian processes (DGP), encoder-decoder neural networks for multi-output regression and eXtreme Gradient Boosting (XGBoost)—applied to a hybrid dataset of experimental and computational properties of the 8-component HEA system Al-Co-Cr-Cu-Fe-Mn-Ni-V. We specifically assess their capabilities in predicting correlated material properties, including yield strength, hardness, modulus, ultimate tensile strength, elongation, and average hardness under dynamic/quasi-static conditions, alongside auxiliary computational properties. The comparison highlights the strengths of hierarchical deep modeling approaches in handling heteroscedastic, heterotopic, and incomplete data commonly encountered in materials science. Our findings illustrate that combined surrogate models such as DGPs infused with machine-learned priors outperform other surrogates by effectively capturing inter-property correlations and by assimilating prior knowledge. This enhanced predictive accuracy positions the combined surrogate models as powerful tools for robust and data-efficient materials design.

36 MATERIALS SCIENCE↗