Search NASA⌕ Search

SEARCH · Search NASA

Results for “heteroskedastic”

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.

Comparison of Probabilistic Forecasts for Predictive Voltage Control

This paper explores predictive cooperative voltage control in distribution systems with highly variable sources such as photovoltaics (PV). The goal is to maintain the voltage profile within the limits despite the fluctuations due to sudden changes in solar power generation. The predictive voltage control method relies on probabilistic solar power and load forecasts to select the optimum Voltage Regulator (VR) taps appropriately. VR taps are selected to minimize the risk of voltage violation. A modified version of the IEEE 123 system is used as the case study. A 100% penetration of solar power is assumed for the distribution system with profiles for solar generation and loads added to the system. Three different probabilistic forecast models (Quantile Regression (QR), Gaussian distribution and volatility forecasting using Generalized Autoregressive Conditional Heteroskedasticity (GARCH)) are explored in this study. The results for the VR taps and Voltage Deviation Index (VDI) are compared to find the most effective forecast model.

Panamtash, Hossein↗

Sensitivity Analysis in the Presence of Intrinsic Stochasticity for Discrete Fracture Network Simulations

Abstract Large‐scale discrete fracture network (DFN) simulators are standard fare for studies involving the sub‐surface transport of particles since direct observation of real world underground fracture networks is generally infeasible. While these simulators have successfully been used in several engineering applications, estimates of output quantities of interest (QoI) — such as breakthrough time of particles reaching the edge of the system — suffer from two distinct types of uncertainty. A run of a DFN simulator requires several parameters to be set that dictate the placement and size of fractures, the density of fractures, and the overall permeability of the system; uncertainty on the proper parameters will lead to uncertainty in the QoI, called epistemic uncertainty. Furthermore, since these input settings to DFN simulators control the stochastic processes which place fractures and govern flow, understanding how this randomness affects the QoI requires several runs of the simulator at distinct random seeds. The uncertainty in the QoI attributed to different realizations (i.e., different seeds) of the same random process (i.e., identical input parameters) leads to a second type of uncertainty, called aleatoric uncertainty. In this paper, we perform a Sensitivity Analysis, which directly attributes the uncertainty observed in the QoI to the epistemic uncertainty from each input parameter and to the aleatoric uncertainty. Beyond the specific takeaways on which input variables influence uncertainty in the QoI the most, a major contribution of this paper is the introduction of a statistically rigorous workflow for characterizing the uncertainty in DFN flow simulations that exhibit heteroskedasticity.

58 GEOSCIENCES↗

Exact Bayesian computation using H-functions

In the Bayesian Statistics, the use of conjugate prior distributions allows to obtain exact posterior quantities, avoiding the use of approximate methods. However, there are only a few conjugate structures available, which limits the use in practical problems. In this work we propose a general procedure to obtain the posterior distribution in an exact form in non-conjugate location parameter models with known (possibly, heteroskedastic) variance. The theory is based on special functions, specifically the H-function, which embraces most of the probability distributions. We express explicitly in a computable form all the Bayesian tools for posterior inferences, that is the posterior and the posterior predictive distributions and their moments, as well as the cumulative posterior distribution. As an illustration we consider a problem in astronomy of estimating the distance to the Large Magellanic Cloud, the largest galaxy now orbiting the Milky Way Galaxy.

79 ASTRONOMY AND ASTROPHYSICS↗

Time-varying coefficient vector autoregressions model based on dynamic correlation with an application to crude oil and stock markets

This paper proposes a new time-varying coefficient vector autoregressions (VAR) model, in which the coefficient is a linear function of dynamic lagged correlation. The proposed model allows for flexibility in choices of dynamic correlation models (e.g. dynamic conditional correlation generalized autoregressive conditional heteroskedasticity (GARCH) models, Markov-switching GARCH models and multivariate stochastic volatility models), which indicates that it can describe many types of time-varying causal effects. Time-varying causal relations between West Texas Intermediate (WTI) crude oil and the US Standard and Poor’s 500 (S&P 500) stock markets are examined by the proposed model. The empirical results show that their causal relations evolve with time and display complex characters. Both positive and negative causal effects of the WTI on the S&P 500 in the subperiods have been found and confirmed by the traditional VAR models. Similar results have been obtained in the causal effects of S&P 500 on WTI. In addition, the proposed model outperforms the traditional VAR model.

54 ENVIRONMENTAL SCIENCES↗

THE EVOLUTION OF Ly{alpha}-EMITTING GALAXIES BETWEEN z = 2.1 AND z = 3.1

We describe the results of a new, wide-field survey for z = 3.1 Ly{alpha} emitters (LAEs) in the Extended Chandra Deep Field South (ECDF-S). By using a nearly top-hat 5010 A filter and complementary broadband photometry from the MUSYC survey, we identify a complete sample of 141 objects with monochromatic fluxes brighter than 2.4 Multiplication-Sign 10{sup -17} erg cm{sup -2} s{sup -1} and observers-frame equivalent widths (EWs) greater than {approx}80 A (i.e., 20 A in the rest frame of Ly{alpha}). The bright end of this data set is dominated by X-ray sources and foreground objects with Galaxy Evolution Explorer detections, but when these interlopers are removed, we are still left with a sample of 130 LAE candidates, 39 of which have spectroscopic confirmations. This sample overlaps the set of objects found in an earlier ECDF-S survey, but due to our filter's redder bandpass, it also includes 68 previously uncataloged sources. We confirm earlier measurements of the z = 3.1 LAE emission-line luminosity function and show that an apparent anticorrelation between EW and continuum brightness is likely due to the effect of correlated errors in our heteroskedastic data set. Finally, we compare the properties of z = 3.1 LAEs to LAEs found at z = 2.1. We show that in the {approx}1 Gyr after z {approx} 3, the LAE luminosity function evolved significantly, with L* fading by {approx}0.4 mag, the number density of sources with L > 1.5 Multiplication-Sign 10{sup 42} erg s{sup -1} declining by {approx}50%, and the EW scale length contracting from 70{sup +7}{sub -5} A to 50{sup +9}{sub -6} A. When combined with literature results, our observations demonstrate that over the redshift range z {approx} 0 to z {approx} 4, LAEs contain less than {approx}10% of the star formation rate density of the universe.

79 ASTRONOMY AND ASTROPHYSICS↗

Spatio-Temporal Deep Graph Network for Event Detection, Localization, and Classification in Cyber-Physical Electric Distribution System

This work proposes a deep graph learning framework to identify, locate, and classify power, cyber, and cyber power events at the distribution system level. The proposed algorithm jointly exploits spatial, temporal, and node-level cyber and physical data features. The developed graph neural network, together with a deep autoencoder, utilizes physical measurements from distribution level phasor measurement units and cyber data from communication network logs. The spatial structure of the synchrophasor measurements and network is incorporated through a weighted adjacency matrix. The temporal structure is incorporated by defining a spatial operation in the gated recurrent unit. This spatio-temporal learning element resides inside a power event detection, localization, and classification module that provides the degree of confidence for an event label. To accurately pinpoint the location of an event to the nearest bus equipped with a measurement unit, a combination of squared error and proximity score is utilized. Also included is a cyber event detection module that employs heteroskedasticity to analyze the significance of various cyber features during different types of attacks. Finally, a dual-bit cyber-power decision table determines the nature of the event. The proposed method is validated on two distribution systems modeled in OPAL-RT/Hypersim with limited phasor measurement units for different possible physical and cyber events. Further analyses include comparison with other state-of-the-art methods and validation in the presence of measurement noise. As a result, our method outperforms existing approaches and achieves an average detection accuracy of 97.97%, F1-score of 96.88%, precision of 96.53%, and recall of 98.57%.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Boron Coordination in Multicomponent Glasses: Analytical Models and Machine Learning With Uncertainty

Borosilicate glasses are extensively used in a variety of applications from kitchenware to nuclear waste immobilization due to the strong network formed by the Si-O-B bond that makes it resistant to chemical corrosion and gives it a low thermal expansion. Boron, however, exists in both trigonal BO3 and tetrahedral BO4 bonds in glass systems, which impacts the chemical durability and thermal resistance of the glass, amongst other properties. Boron coordination (N4), or the ratio of the amount of BO4 to BO3 within a glass, may aid in predicting these properties but is difficult to derive without experimental data due to the complexity of impacts from varied glass compositions and processing factors. For this reason, compositional models have been developed to predict boron coordination, but the models typically include a limited number of glass components. To help fill this gap in the models, in this work, a diverse multicomponent glass dataset of 809 glasses is compiled from a literature search, and then a number of analytical and machine learning (ML) models are trained on the dataset. Previously developed modified Bernstein and modified Du Stebbins analytical models were fitted to update parameters with the new dataset. Then, partially Bayesian neural networks, Gaussian process regressor, and heteroskedastic deterministic neural networks were evaluated. The ML models examined all have different strategies to overcome the potential for overfitting as a result of a limited training dataset, and return results that account for model uncertainty, which can be valuable for understanding model reliability. For the first time, cooling rate is introduced as an input parameter for ML models, showing consistent improvements in performance and solidifying the importance of including parameters outside of composition alone for N4 prediction. The machine learning models examined here show promise in accurate predictions of boron coordination in borosilicate glasses, all achieving R2 values of 0.91.

boron coordination↗

Quantile Kriging R Package

A re-implementation of Quantile Kriging. Quantile Kriging was described by Matthew Plumlee & Rui Tuo in their 2014 paper "Building Accurate Emulators for Stochastic Simulations via Quantile Kriging." With computational savings when dealing with replication from the recent paper "Practical heteroskedastic Gaussian process modeling for large simulation experiments”, it is now possible to apply Quantile Kriging to a wider class of problems. In addition to fitting the model, other useful tools are provided such as the ability to automatically perform leave-one-out cross validation.

Quinlan, KevinR.↗

¬¬Integration of Quantification of Margins and Uncertainties Methodology into Parallel Discrete Event Simulator Framework

Parallel Discrete Events Simulation (PDES) is becoming increasingly important to lab efforts in security and intelligence. It is used to model complex asynchronous systems such as computer networks, satellite systems, vehicular traffic, and human performance. Uncertainty Quantification (UQ) techniques have become a mainstay of Verification and Validation (V&V) efforts on physics simulations. However, PDES models are very different from traditional physics simulations, and research into UQ techniques for PDES is in its infancy. It is not clear which traditional UQ techniques can be applied to PDES, or what new techniques will need to be developed. The goal of this project was to identify existing UQ techniques that can be applied to PDES, and to develop new techniques as necessary. The project implemented or developed techniques to handle issues that do not appear in traditional physics simulations, but are common among PDES, including sampling techniques for high-dimensional homogenous inputs, response surfaces for high-variance heteroskedastic output, and characterization of skewed output distributions. These are foundational UQ techniques that must be used in any complete UQ analysis (Tong 2018).

97 MATHEMATICS AND COMPUTING↗

The Evolution of Ly-alpha Emitting Galaxies Between z = 2.1 and z = 3.l

We describe the results of a new, wide-field survey for z= 3.1 Ly-alpha emission-line galaxies (LAEs) in the Extended Chandra Deep Field South (ECDF-S). By using a nearly top-hat 5010 Angstrom filter and complementary broadband photometry from the MUSYC survey, we identify a complete sample of 141 objects with monochromatic fluxes brighter than 2.4E-17 ergs/cm^2/s and observers-frame equivalent widths greater than ~ 80 Angstroms (i.e., 20 Angstroms in the rest-frame of Ly-alpha). The bright-end of this dataset is dominated by x-ray sources and foreground objects with GALEX detections, but when these interlopers are removed, we are still left with a sample of 130 LAE candidates, 39 of which have spectroscopic confirmations. This sample overlaps the set of objects found in an earlier ECDF-S survey, but due to our filter's redder bandpass, it also includes 68 previously uncataloged sources. We confirm earlier measurements of the z=3.1 LAE emission-line luminosity function, and show that an apparent anti-correlation between equivalent width and continuum brightness is likely due to the effect of correlated errors in our heteroskedastic dataset. Finally, we compare the properties of z=3.1 LAEs to LAEs found at z=2.1. We show that in the approximately 1 Gyr after z approximately 3, the LAE luminosity function evolved significantly, with L * fading by approximately 0.4 mag, the number density of sources with L greater than 1.5E42 ergs/s declining by approximately 50%, and the equivalent width scalelength contracting from 70^{+7}_{-5} Angstroms to 50^{+9}_{-6} Angstroms. When combined with literature results, our observations demonstrate that over the redshift range z approximately 0 to z approximately 4, LAEs contain less than approximately 10% of the star-formation rate density of the universe.

Ciardullo, Robin↗