SEARCH · Engineering Papers

Results for “Gaussian process model”

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 685 records · Page 38

All sources NASA NTRS DOE OSTI

NASA NTRS1992

Laboratory for Engineering Man/Machine Systems (LEMS): System identification, model reduction and deconvolution filtering using Fourier based modulating signals and high order statistics

Several important problems in the fields of signal processing and model identification, such as system structure identification, frequency response determination, high order model reduction, high resolution frequency analysis, deconvolution filtering, and etc. Each of these topics involves a wide range of applications and has received considerable attention. Using the Fourier based sinusoidal modulating signals, it is shown that a discrete autoregressive model can be constructed for the least squares identification of continuous systems. Some identification algorithms are presented for both SISO and MIMO systems frequency response determination using only transient data. Also, several new schemes for model reduction were developed. Based upon the complex sinusoidal modulating signals, a parametric least squares algorithm for high resolution frequency estimation is proposed. Numerical examples show that the proposed algorithm gives better performance than the usual. Also, the problem was studied of deconvolution and parameter identification of a general noncausal nonminimum phase ARMA system driven by non-Gaussian stationary random processes. Algorithms are introduced for inverse cumulant estimation, both in the frequency domain via the FFT algorithms and in the domain via the least squares algorithm.

Pan, Jianqiang↗

DOE OSTI2020

Data-driven predictive probability density function control of fiber length stochastic distribution shaping in refining process

Pulp is the most important raw material in paper in paper industries, whose Fiber Length Stochastic Distribution (FLSD) shaping directly determines the energy consumption and paper quality of the subsequent papermaking processes. However, the mean and variance are insufficient to describe the output FLSD shaping, which display non-Gaussian distribution properties. Therefore, the traditional control, method based on the mean and variance of fiber length is difficult to control the FLSD shaping effectively. In this paper, a novel data-driven predictive probability density function (PDF) control method is proposed for output FLSD shaping in refining process. Primarily, in order to improve the approximation accuracy of the PDF employing RBF neural network, the parameters (center value and width) of each RBF basis functions are tuned via utilizing iterative learning control (ILC) law, and the corresponding estimations of weights law can be obtained. Secondly, considering that the conventional linear model of weights vector has drawbacks of low accuracy and weak generalization ability, random vector functional link networks (RVFLNs) based data driven nonlinear modelling method is employed to characterize the prediction model between the input variables and weights vector. Finally, in order to reduce the randomness of the output FLSD, a minimum entropy control method under mean constraint is employed to design the predictive PDF controller for the output FLSD shaping with the help of the established nonlinear stochastic distribution model. Both simulations and experiments are given to illustrate the effectiveness and practicability of the proposed method.

Probability debsity function, Fiber length distrib↗

DOE OSTI2024

Combining High-Throughput Experiments and Active Learning to Characterize Deep Eutectic Solvents

The high tunability of deep eutectic solvents (DESs) stems from the ease of changing their precursors and relative compositions. However, measuring the physicochemical properties across large composition and temperature ranges, necessary to properly design target-specific DESs, is tedious and error-prone and represents a bottleneck in the advancement and scalability of DES-based applications. As such, active learning (AL) methodologies based on Gaussian processes (GPs) were developed in this work to minimize the experimental effort necessary to characterize DESs. Owing to its importance for large-scale applications, the reduction of DES viscosity through the addition of a low-molecular-weight solvent was explored as a case study. A high-throughput experimental screening was initially performed on nine different ternary DESs. Then, GPs were successfully trained to predict DES viscosity from its composition and temperature, showcasing the ability of these stochastic, nonparametric models to accurately describe the physicochemical properties of complex mixtures. Finally, the ability of GPs to provide estimates of their own uncertainty was leveraged through an AL framework to minimize the number of data points necessary to obtain accurate viscosity modes. This led to a significant reduction in data requirements, with many systems requiring only five independent viscosity data points to be properly described.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2019

Estimating redshift distributions using hierarchical logistic Gaussian processes

ABSTRACT This work uses hierarchical logistic Gaussian processes to infer true redshift distributions of samples of galaxies, through their cross-correlations with spatially overlapping spectroscopic samples. We demonstrate that this method can accurately estimate these redshift distributions in a fully Bayesian manner jointly with galaxy-dark matter bias models. We forecast how systematic biases in the redshift-dependent galaxy-dark matter bias model affect redshift inference. Using published galaxy-dark matter bias measurements from the Illustris simulation, we compare these systematic biases with the statistical error budget from a forecasted weak gravitational lensing measurement. If the redshift-dependent galaxy-dark matter bias model is mis-specified, redshift inference can be biased. This can propagate into relative biases in the weak lensing convergence power spectrum on the 10–30 per cent level. We, therefore, showcase a methodology to detect these sources of error using Bayesian model selection techniques. Furthermore, we discuss the improvements that can be gained from incorporating prior information from Bayesian template fitting into the model, both in redshift prediction accuracy and in the detection of systematic modelling biases.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2019

Residual Optically Stimulated Luminescent (OSL) Signals for Al 2 O 3 :C and a Readout System with Reproducible Partial Signal Clearance

Optically stimulated luminescent dosimeters (OSLDs) are devices that, when stimulated with light, emit light in proportion to the integrated ionizing radiation dose. The stimulation of OSL material results in the loss of a small fraction of signal stored within the dosimetric traps. Previous studies have investigated the signal loss due to readout stimulation, and the optical annealing of OSLDs. This study builds on former research by examining the behavior of OSL signal after annealing, exploring the functionality of a previously developed signal loss model, and comparing uncertainties for dosimeters reused with or without annealing. For a completely annealed dosimeter, the minimum signal level was 56 ± 8 counts, and readings followed a Gaussian distribution. For dosimeters above this signal level, the fractional signal loss due to the reading process has a linear relationship with the calculated signal. At low signal levels (below 20,000 counts), in this OSLD system,calculated signal percent errors increase significantly but otherwise are on average 0.72 ± 0.27%, 0.40 ± 0.19%, 0.33±0.12%, and 0.24±0.07% for 30, 75, 150, and 300 readings. Theoretical calculations of uncertainties showed that annealing before reusing dosimeters allows for dose errors below 1% with as few as 30 readings. Reusing dosimeters multiple times increases the dose errors especially with low numbers of readouts, so theoretically around 300 readings would be necessary to achieve errors around 1% or below in most scenarios. Note that these dose errors do not include the error associated with the signal-to-dose conversion factor.

61 RADIATION PROTECTION AND DOSIMETRY↗

NASA NTRS1976

Neural pulse frequency modulation of an exponentially correlated Gaussian process

The effect of NPFM (Neural Pulse Frequency Modulation) on a stationary Gaussian input, namely an exponentially correlated Gaussian input, is investigated with special emphasis on the determination of the average number of pulses in unit time, known also as the average frequency of pulse occurrence. For some classes of stationary input processes where the formulation of the appropriate multidimensional Markov diffusion model of the input-plus-NPFM system is possible, the average impulse frequency may be obtained by a generalization of the approach adopted. The results are approximate and numerical, but are in close agreement with Monte Carlo computer simulation results.

Hutchinson, C. E.↗

DOE OSTI2023

Neural network surrogate models for equations of state

Equation of state (EOS) data provide necessary information for accurate multiphysics modeling, which is necessary for fields such as inertial confinement fusion. Here, we suggest a neural network surrogate model of energy and entropy and use thermodynamic relationships to derive other necessary thermodynamic EOS quantities. We incorporate phase information into the model by training a phase classifier and using phase-specific regression models, which improves the modal prediction accuracy. Our model predicts energy values to 1% relative error and entropy to 3.5% relative error in a log-transformed space. Although sound speed predictions require further improvement, the derived pressure values are accurate within 10% relative error. Our results suggest that neural network models can effectively model EOS for inertial confinement fusion simulation applications.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2021

Robust Deep Gaussian Process-Based Probabilistic Electrical Load Forecasting Against Anomalous Events

The abnormal events, such as the unprecedented COVID-19 pandemic, can significantly change the load behaviors, leading to huge challenges for traditional short-term forecasting methods. This article proposes a robust deep Gaussian processes (DGP)-based probabilistic load forecasting method using a limited number of data. Since the proposed method only requires a limited number of training samples for load forecasting, it allows us to deal with extreme scenarios that cause short-term load behavior changes. In particular, the load forecasting at the beginning of abnormal event is cast as a regression problem with limited training samples and solved by double stochastic variational inference DGP. The mobility data are also utilized to deal with the uncertainties and pattern changes and enhance the flexibility of the forecasting model. The proposed method can quantify the uncertainties of load forecasting outcomes, which would be essential under uncertain inputs. Extensive comparison results with other state-of-the-art point and probabilistic forecasting methods show that our proposed approach can achieve high forecasting accuracies with only a limited number of data while maintaining the excellent performance of capturing the forecasting uncertainties.

anomalous events↗

DOE OSTI2025

Exploring the energy landscape of RBMs: reciprocal space insights into bosons, hierarchical learning and symmetry breaking

Deep generative models have become ubiquitous due to their ability to learn and sample from complex distributions. Despite the proliferation of various frameworks, the relationships among these models remain largely unexplored, a gap that hinders the development of a unified theory of AI learning. In this work, we address two central challenges: clarifying the connections between different deep generative models and deepening our understanding of their learning mechanisms. We focus on Restricted Boltzmann Machines (RBMs), a class of generative models known for their universal approximation capabilities for discrete distributions. By introducing a reciprocal space formulation for RBMs, we reveal a connection between these models, diffusion processes, and systems of coupled bosons. Our analysis shows that at initialization, the RBM operates at a saddle point, where the local curvature is determined by the singular values of the weight matrix, whose distribution follows the Marc̆enko-Pastur law and exhibits rotational symmetry. During training, this rotational symmetry is broken due to hierarchical learning, where different degrees of freedom progressively capture features at multiple levels of abstraction. This leads to a symmetry breaking in the energy landscape, reminiscent of Landau’s theory. This symmetry breaking in the energy landscape is characterized by the singular values and the weight matrix eigenvector matrix. We derive the corresponding free energy in a mean-field approximation. We show that in the limit of infinite size RBM, the reciprocal variables are Gaussian distributed. Our findings indicate that in this regime, there will be some modes for which the diffusion process will not converge to the Boltzmann distribution. To illustrate our results, we trained replicas of RBMs with different hidden layer sizes using the MNIST dataset. Our findings not only bridge the gap between disparate generative frameworks but also shed light on the fundamental processes underpinning learning in deep generative models.

97 MATHEMATICS AND COMPUTING↗

NASA NTRS1980

Propagation in a shearing plasma. II - Turbulence and the frequency range of pulsar microstructure

Numerical calculations are presented which explore the propagation of radio waves and pulses through a turbulent shearing plasma. A stochastic model is used to describe the turbulent velocity field, in which random fluctuations are superposed on a mean profile. Both Gaussian wave packets and shot-noise pulses polarized in the plane of shearing acquire quasi-periodic intensity modulations above the rest-frame plasma frequency. The frequency range over which these modulations appear is shown to depend on the spatial correlation length of the turbulent fluctuations and on their standard deviation from the mean velocity. The period of the modulations is variable from one realization of the random process to the next, and the average period is frequency dependent. The results of these calculations lend further support to a model for pulsar microstructure in which periodic micropulses are pure temporal modulations of the emitted radiation due to propagation effects in magnetospheric shearing regions. It is shown that turbulence on a scale of approximately 10 cm can produce these quasi-periodic modulations over a frequency range comparable to that of observed microstructure.

Harding, A. K.↗

DOE OSTI2024

Glass Design Using Machine Learning Property Models with Prediction Uncertainties: Nuclear Waste Glass Formulation

The United States Department of Energy is responsible for managing the legacy nuclear waste stored in underground tanks at the Hanford Site. The waste will be separately vitrified as low-activity waste and high-level waste fractions. Waste glass formulation algorithms have been traditionally developed using partial quadratic mixture property-composition models. Recently, machine learning (ML) techniques have been used to predict glass properties and discover new glass materials for nuclear waste vitrification, and these advancements can be utilized to improve waste glass composition design. In this proof-of-principle study, ML algorithms such as Gaussian process regression (GPR) were used to interpolate glass properties (e.g., viscosity, electrical conductivity, chemical durability). After selecting appropriate sets of GPR hyper-parameters for each property, an optimization program was developed to formulate glass compositions to maximize waste loading while simultaneously satisfying property within constraints. The results of the ML-based waste loadings and glass compositions were compared to those obtained using the traditional methods. Comparing to the previous glass design framework, the ML-based optimization methods offer improved glass designs and a streamlined approach to generation of optimally designed data and near real-time updates.

glass formulation, machine learning, constraints, ↗

NASA NTRS2005

Evaluation of a New Cloud Droplet Activation Parameterization wtih in Situ Data from CRYSTAL-FACE and CSTRIPE

The accuracy of the 2003 prognostic, physically based aerosol activation parameterization of A. Nenes and J. H. Seinfeld (NS) with modification introduced by C. Fountoukis and A. Nenes in 2005 (modified NS) is evaluated against extensive microphysical data sets collected on board the Center for Interdisciplinary Remotely Piloted Aircraft Studies (CIRPAS) Twin Otter aircraft for cumuliform and stratiform clouds of marine and continental origin. The cumuliform cloud data were collected during NASA's Cirrus Regional Study of Tropical Anvils and Cirrus Layers-Florida Area Cirrus Experiment (CRYSTAL-FACE, Key West, Florida, July 2002), while the stratiform cloud data were gathered during Coastal Stratocumulus Imposed Perturbation Experiment (CSTRIPE, Monterey, California, July 2003). In situ data sets of aerosol size distribution, chemical composition, and updraft velocities are used as input for the NS parameterization, and the evaluation is carried out by comparing predicted cloud droplet number concentrations (CDNC) with observations. This is the first known study in which a prognostic cloud droplet activation parameterization has been evaluated against a wide range of observations. On average, predicted droplet concentration in adiabatic regions is within -20% of observations at the base of cumuliform clouds and -30% of observations at different altitudes throughout the stratiform clouds, all within experimental uncertainty. Furthermore, CDNC is well parameterized using either a single mean updraft velocity w or by weighting droplet nucleation rates with a Gaussian probability density function of w. This study suggests that for nonprecipitating warm clouds of variable microphysics, aerosol composition, and size distribution the modified NS parameterization can accurately predict cloud droplet activation and can be successfully implemented for describing the aerosol activation process in global climate models.

Meskhidze, Nicholas↗

DOE OSTI2025

Real-time tracking and analysis of gas bubble dynamics in laser powder bed fusion using in-situ X-ray characterization and machine learning

Porosity defects remain a significant challenge in the laser powder bed fusion (LPBF) process, adversely affecting the mechanical properties and reliability of additively manufactured components. Here, this study investigates the real-time formation and trajectory of gas bubbles during LPBF of Al6061 alloy using advanced in-situ X-ray characterization and machine learning. The unsupervised Gaussian mixture model and particle tracking algorithm developed are able to precisely track and quantify the properties of gas bubbles and keyhole pores. Our analysis identified five distinct types of gas bubble formation and movement patterns, emphasizing the diverse origins and behaviors of these defects. It enables precise quantification of trajectories, velocities, and morphological changes of gas bubbles, offering a granular view of the subsurface dynamics within the melt pool. Additionally, we explored keyhole-induced pore dynamics, revealing the critical role of keyhole oscillation and collapse for the formation of both large and small gas pores. It defines four different regions of gas bubble movement within the melt pool, providing a clearer understanding of how local fluid dynamics affect pore behavior. The results underscore the importance of integrating in-situ experimental observation and automated machine learning to develop a more robust predictive model for defect formation in LPBF.

In-situ X-ray imaging↗

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 NTRS1980

Middle infrared multispectral aircraft scanner data - Analysis for geological applications

Multispectral middle IR (8-13 microns) data were acquired with an aircraft scanner over Utah. Because these digital image data were dominated by temperature, all six channels were highly correlated. Extensive processing was required to allow geologic photointerpretation based on subtle variations in spectral emittance between rock types. After preliminary processing, ratio images were produced and color ratio composites created from these. Sensor calibration and an atmospheric model allowed determination of surface brightness, temperature, emittance, and color composite emittance images. The best separation of major rock types was achieved with a principal component transformation, followed by a Gaussian stretch, followed by an inverse transformation to the original axes.

Kahle, A. B.↗

NASA NTRS1985

Three-dimensional calculation of shuttle charging in polar orbit

The charged particles environment in polar orbit can be of sufficient intensity to cause spacecraft charging. In order to gain a quantitative understanding of such effects, the Air Force is developing POLAR, a computer code which simulates in three dimensions the electrical interaction of large space vehicles with the polar ionospheric plasma. It models the physical processes of wake generation, ambient ion collection, precipitating auroral electron fluxes, and surface interactions, including secondary electron generation and backscattering, which lead to vehicle charging. These processes may be followed dynamically on a subsecond timescale so that the rapid passage through intense auroral arcs can be simulated. POLAR models the ambient plasma as isotropic Maxwellian electrons and ions (0+, H+), and allows for simultaneous precipitation of power-law, energetic Maxwellian, and accelerated Gaussian distributions of electrons. Magnetic field effects will be modeled in POLAR but are currently ignored.

Cooke, D. L.↗

DOE OSTI2025

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

Rajput, Kishansingh [Thomas Jefferson National Acc↗

DOE OSTI2020

Probabilistic Prediction of Geomagnetic Storms and the K p index

Geomagnetic activity is often described using summary indices to summarize the likelihood of space weather impacts, as well as when parameterizing space weather models. The geomagnetic index K p in particular, is widely used for these purposes. Current state-of-the-art forecast models provide deterministic K p predictions using a variety of methods – including empirically-derived functions, physics-based models, and neural networks – but do not provide uncertainty estimates associated with the forecast. This paper provides a sample methodology to generate a 3-hour-ahead K p prediction with uncertainty bounds and from this provide a probabilistic geomagnetic storm forecast. Specifically, we have used a two-layered architecture to separately predict storm (K p ≥ 5 – ) and non-storm cases. As solar wind-driven models are limited in their ability to predict the onset of transient-driven activity we also introduce a model variant using solar X-ray flux to assess whether simple models including proxies for solar activity can improve the predictions of geomagnetic storm activity with lead times longer than the L1-to-Earth propagation time. By comparing the performance of these models we show that including operationally-available information about solar irradiance enhances the ability of predictive models to capture the onset of geomagnetic storms and that this can be achieved while also enabling probabilistic forecasts.

79 ASTRONOMY AND ASTROPHYSICS↗