Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “multivariate data”

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 415 records · Page 23
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Multivariable degradation modeling and life prediction using multivariate fractional Brownian motion

In system prognostics and health management, multivariable degradation models have been widely developed to predict the life of complex systems using degradation data of multiple Performance Characteristics (PCs). Recent studies have detected a Long-Term Memory (LTM) effect among the degradation process of various PCs, implying a strong coupling phenomenon between the future degradation behavior and historical degradation trajectory. Although the LTM has been widely integrated into single-PC-based degradation modeling, it has not been considered in multi-PC-based scenarios. To capture LTM among multiple PCs, this article proposes a novel LTM-integrated Multivariate Degradation Model (MDM) for system life prediction based on multivariate fractional Brownian motion, which simultaneously incorporates the cross-correlation among different PCs. To estimate parameters of the LTM-integrated MDM, a maximum likelihood method is developed. Here, two likelihood-ratio hypothesis tests are developed to test the existence of the overall and individual LTM effect among multiple PCs. Both simulation studies and physical experiments on the performance degradation of solar energy conversion and storage devices are conducted to validate the proposed model. Results reveal that the proposed LTM-integrated MDM significantly outperforms existing MDMs in life prediction, while the lifetime uncertainty is heavily underestimated by those traditional approaches that neglect the LTM.

42 ENGINEERING↗
DOE OSTI2021

Scalable statistical inference of photometric redshift via data subsampling

Handling big data has largely been a major bottleneck in traditional statistical models. Consequently, when accurate point prediction is the primary target, machine learning models are often preferred over their statistical counterparts for bigger problems. But full probabilistic statistical models often outperform other models in quantifying uncertainties associated with model predictions. We develop a data-driven statistical modeling framework that combines the uncertainties from an ensemble of statistical models learned on smaller subsets of data carefully chosen to account for imbalances in the input space. We demonstrate this method on a photometric redshift estimation problem in cosmology, which seeks to infer a distribution of the redshift—the stretching effect in observing the light of far-away galaxies—given multivariate color information observed for an object in the sky. Our proposed method performs balanced partitioning, graph-based data subsampling across the partitions, and training of an ensemble of Gaussian process models.

data subsampling↗
NASA NTRS1989

K-distribution and polarimetric terrain radar clutter

A multivariate K-distribution is proposed to model the statistics of fully polarimetric radar data from earth terrain with polarizations HH, HV, VH, and VV. In this approach, correlated polarizations of radar signals, as characterized by a covariance matrix, are treated as the sum of N n-dimensional random vectors; N obeys the negative binomial distribution with a parameter alpha and mean N-bar. Subsequently, an n-dimensional K-distribution, with either zero or nonzero mean, is developed in the limit of infinite N-bar or illuminated area. The probability density function (PDF) of the K-distributed vector normalized by its Euclidean norm is independent of the parameter alpha and is the same as that derived from a zero-mean Gaussian-distributed random vector.

Yueh, S. H.↗
NASA NTRS2009

Parametric Analysis of a Hover Test Vehicle using Advanced Test Generation and Data Analysis

Large complex aerospace systems are generally validated in regions local to anticipated operating points rather than through characterization of the entire feasible operational envelope of the system. This is due to the large parameter space, and complex, highly coupled nonlinear nature of the different systems that contribute to the performance of the aerospace system. We have addressed the factors deterring such an analysis by applying a combination of technologies to the area of flight envelop assessment. We utilize n-factor (2,3) combinatorial parameter variations to limit the number of cases, but still explore important interactions in the parameter space in a systematic fashion. The data generated is automatically analyzed through a combination of unsupervised learning using a Bayesian multivariate clustering technique (AutoBayes) and supervised learning of critical parameter ranges using the machine-learning tool TAR3, a treatment learner. Covariance analysis with scatter plots and likelihood contours are used to visualize correlations between simulation parameters and simulation results, a task that requires tool support, especially for large and complex models. We present results of simulation experiments for a cold-gas-powered hover test vehicle.

Gundy-Burlet, Karen↗
NASA NTRS2010

Fast Multivariate Search on Large Aviation Datasets

Multivariate Time-Series (MTS) are ubiquitous, and are generated in areas as disparate as sensor recordings in aerospace systems, music and video streams, medical monitoring, and financial systems. Domain experts are often interested in searching for interesting multivariate patterns from these MTS databases which can contain up to several gigabytes of data. Surprisingly, research on MTS search is very limited. Most existing work only supports queries with the same length of data, or queries on a fixed set of variables. In this paper, we propose an efficient and flexible subsequence search framework for massive MTS databases, that, for the first time, enables querying on any subset of variables with arbitrary time delays between them. We propose two provably correct algorithms to solve this problem (1) an R-tree Based Search (RBS) which uses Minimum Bounding Rectangles (MBR) to organize the subsequences, and (2) a List Based Search (LBS) algorithm which uses sorted lists for indexing. We demonstrate the performance of these algorithms using two large MTS databases from the aviation domain, each containing several millions of observations Both these tests show that our algorithms have very high prune rates (>95%) thus needing actual

Bhaduri, Kanishka↗
NASA NTRS2022

New Constraints on Titan's Stratospheric n-Butane Abundance

Curiously, n-butane has yet to be detected at Titan, though it is predicted to be present in a wide range of abundances that span over 2.5 orders of magnitude. We have searched infrared spectroscopic observations of Titan for signals from n-butane (n-C4H10) in Titan's stratosphere. Three sets of Cassini Composite Infrared Spectrometer Focal Plane 4 (1050–1500 cm−1) observations were selected for modeling, having been collected from different flybys and pointing latitudes. We modeled the observations with the Nonlinear Optimal Estimator for MultivariatE Spectral AnalySIS radiative transfer tool. Temperature profiles were retrieved for each of the data sets by modeling the ν4 emission from methane near 1305 cm−1. Then, incorporating the temperature profiles, we retrieved abundances of all of Titan's known trace gases that are active in this spectral region, reliably reproducing the observations. We then systematically tested a set of models with varying abundances of n-butane, investigating how the addition of this gas affected the fits. We did this for several different photochemically predicted abundance profiles from the literature, as well as for a constant-with-altitude profile. Ultimately, though we did not produce any firm detection of n-butane, we derived new upper limits on its abundance specific to the use of each profile and to multiple different ranges of stratospheric altitudes. These results will tightly constrain the C4 chemistry of future photochemical modeling of Titan's atmosphere and also motivate the continued search for n-butane and its isomer, isobutane.

Brendan L. Steffens↗
NASA NTRS1975

The LANDSAT-1 multispectral scanner as a tool in the classification of inland lakes

Relationships between LANDSAT-1 multispectral scanner (MSS) data and the trophic status of a group of lakes in the north-northeastern part of the United States were studied by predicting the magnitudes of two trophic state indicators, estimating lake position on a multivariate trophic scale, and automatically classifying lakes according to their trophic state. Initially, the principal component ordination was employed with 100 lakes. MSS data for some 20 lakes was then extracted from computer-compatible tapes (CCT) using a binary marking technique. The output was in the form of descriptive statistics and photographic concatenations. Color ratios were incorporated into regression models for the prediction of Secchi disc transparency, chlorophyll a, and lake position on the tropic scale. Results indicate that the LANDSAT-1 system, although handicapped by low spectral and spatial resolutions as well as excessive cloud cover, can be used as a supplemental data source in lake survey programs.

Boland, D. H. P.↗
DOE OSTI2022

Data for The Leaf Economics Spectrum of Triploid and Tetraploid C4 Grass Miscanthus x giganteus

The leaf economics spectrum (LES) describes multivariate correlations in leaf structural, physiological and chemical traits, originally based on diverse C3 species grown under natural ecosystems. However, the specific contribution of C4 species to the global LES is studied less widely. C4 species have a CO2 concentrating mechanism which drives high rates of photosynthesis and improves resource use efficiency, thus potentially pushing them towards the edge of the LES. Here, we measured foliage morphology, structure, photosynthesis, and nutrient content for hundreds of genotypes of the C4 grass Miscanthus × giganteus grown in two common gardens over two seasons. We show substantial trait variations across M. × giganteus genotypes and robust genotypic trait relationships. Compared to the global LES, M. × giganteus genotypes had higher photosynthetic rates, lower stomatal conductance, and less nitrogen content, indicating greater water and photosynthetic nitrogen use efficiency in the C4 species. Additionally, tetraploid genotypes produced thicker leaves with greater leaf mass per area and lower leaf density than triploid genotypes. By expanding the LES relationships across C3 species to include C4 crops, these findings highlight that M. × giganteus occupies the boundary of the global LES and suggest the potential for ploidy to alter LES traits.

Biomass Analytics↗
DOE OSTI2024

Data-Driven Computation of Probabilistic Marching Cubes for Efficient Visualization of Level-Set Uncertainty

Uncertainty visualization is an important emerging research area. Being able to visualize data uncertainty can help scientists improve trust in analysis and decision-making. However, visualizing uncertainty can add computational overhead, which can hinder the efficiency of analysis. In this paper, we propose novel data-driven techniques to reduce the computational requirements of the probabilistic marching cubes (PMC) algorithm. PMC is an uncertainty visualization technique that studies how uncertainty in data affects level-set positions. However, the algorithm relies on expensive Monte Carlo (MC) sampling for the multivariate Gaussian uncertainty model because no closed-form solution exists for the integration of multivariate Gaussian. In this work, we propose the eigenvalue decomposition and adaptive probability model techniques that reduce the amount of MC sampling in the original PMC algorithm and hence speed up the computations. Our proposed methods produce results that show negligible differences compared with the original PMC algorithm demonstrated through metrics, including root mean squared error, maximum error, and difference images. We demonstrate the performance and accuracy evaluations of our data-driven methods through experiments on synthetic and real datasets.

Athawale, Tushar↗
NASA NTRS1993

Multivariable State-Space Identification in the Delta and Shift Operators

This paper develops algorithms for multivariable state-space identification which can be used to estimate models in any operator of interest i.e., delta-rule, shift, Laplace s, etc. The approach is based on the State-Space from Frequency Data (SSFD) algorithm which was designed specifically to eliminate distortions from windowing effects.

multivariable state-space identification SSFD wind↗
DOE OSTI2020

On the quantification and efficient propagation of imprecise probabilities with copula dependence

This paper addresses the problem of quantification and propagation of uncertainties associated with dependence modeling when data for characterizing probability models are limited. Practically, the system inputs are often assumed to be mutually independent or correlated by a multivariate Gaussian distribution. However, this subjective assumption may introduce bias in the response estimate if the real dependence structure deviates from this assumption. In this work, we overcome this limitation by introducing a flexible copula dependence model to capture complex dependencies. Here, a hierarchical Bayesian multimodel approach is proposed to quantify uncertainty in dependence model-form and model parameters that result from small data sets. This approach begins by identifying, through Bayesian multimodel inference, a set of candidate marginal models and their corresponding model probabilities, and then estimating the uncertainty in the copula-based dependence structure, which is conditional on the marginals and their parameters. The overall uncertainties integrating marginals and copulas are probabilistically represented by an ensemble of multivariate candidate densities. A novel importance sampling reweighting approach is proposed to efficiently propagate the overall uncertainties through a computational model. Through an example studying the influence of constituent properties on the out-of-plane properties of transversely isotropic E-glass fiber composites, we show that the composite property with copula-based dependence model converges to the true estimate as data set size increases, while an independence or arbitrary Gaussian correlation assumption leads to a biased estimate.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
NASA NTRS2019

Deep Learning Method for Detecting Precursors to Adverse Events

With the recent advancements in Deep Learning methods, the ability to model large complex heterogeneous data sets are fundamentally changing industry and research. Coupled with hardware improvements, and ease of implementation, a wide variety of deep neural network architectures can quickly be developed to solve a sweeping range of problems such as: object detection in images, automatic healthcare diagnosis using heterogenous data sources, real time language translating and sentence prediction, upscaling low resolution images, and forecasting of multivariate timeseries. Generally, many of these architectures outperform classical machine learning approaches in their respective tasks, however, this typically comes at a cost of interpretability. These black box algorithms generally suffer from lack of transparency in both model complexity as well as the rationale behind the prediction. This lack of comprehension, is driving an emerging area of interest in “Explainable AI”. An algorithm called: “Deep Temporal Multiple Instance Learning”1 was a recently developed to identify precursors to adverse events and has been applied in the aviation domain. The deep learning architecture is designed to capture the evolution of the probability of the outcome over the time preceding the adverse event using a multiple instance learning approach as illustrated in Figure 1. Precursors are defined when the probability of the event has exceeded a threshold at some point in the timeseries, at which point, a sensitivity analysis is performed to determine contributing factors. The contributing factors are used to explain and define the precursor during the periods where the probability score is high. The identified contributing factors are then presented to subject matter experts to provide objective insights into the leading factors associated with the particular adverse event. The algorithm has been tested on flight data from a commercial airline and has the ability to discover precursors to known adverse events that take the form of safety critical operations, such as unstable approach events on final approach. Apart from detecting precursors to adverse events, the converse can also be leveraged to discover corrective actions. These positive actions manifest themselves as periods in the timeseries when the precursor score has been lowered from an elevated state; meaning that if the system had been left uncorrected, it would have eventually reached the adverse event state. Characterizing these state changes can help identify successful interventions that may not have been known before. Policy makers and procedure designers can use this additional knowledge to craft more safety and efficient resilient procedures for future operations and therefore improve the overall performance of the National Airspace.

Matthews, Bryan L.↗
DOE OSTI2023

New methods for trace analysis of gamma-irradiated pentaerythritol tetranitrate

High explosives (HEs) are used in a diverse range of applications in which they could be exposed to various radiation levels that may cause potential chemical changes. This study further evaluated pentaerythritol tetranitrate (PETN) that was previously aged with a low-level 2 kGy dose of gamma irradiation in order to understand chemical changes caused by irradiation. Both unirradiated PETN and gamma-irradiated PETN were analyzed using ultra high-pressure liquid chromatography coupled to quadrupole time of flight mass spectrometry (UHPLC-QTOF). The resulting data were processed in a non-targeted manner using Fisher's ratio analysis and multivariate curve resolution-alternating least squares (MCR-ALS) to aid in discovery and identification of the chemical changes brought about by irradiation without a priori knowledge. The application of using UHPLC-QTOF in combination with chemometric techniques for the analysis of irradiated samples has not previously been performed. Here, in this work, we show how to use this method to provide chemical information that would otherwise not be discernible, such as the discovery of degradation of the various homologues of PETN. Major differences identified with radiolytic aging of the PETN sample included decomposition products that resulted from the degradation of the trigger linkage – the O–NO 2 bonds – resulting in the formation of alcohol and aldehyde groups. Similar degradation was also observed in the PETN homologues as well as interconversion from one homologue to another.

38 RADIATION CHEMISTRY, RADIOCHEMISTRY, AND NUCLEA↗
NASA NTRS2001

Surface Elevation Changes in West Antarctica from Satellite Altimetry: Mass Balance Implications

Time-series of surface elevation change, which are constructed from 7-years (1992-1999) of ERS-1 and 2 satellite radar altimeter data of Antarctica, show significant seasonal, inter-annual, and long-term changes. Elevation time-series are created from altimeter crossovers among 90-day data periods on a 50 km grid to 81.5 degrees S and fit with a multivariate linear/sinusoidal function to give the average rate of elevation change (dH/dt) and account for seasonal changes. On the major Ronne, Filchner, and Ronne ice shelves, the dH/dt are small or near zero. In contrast, the ice shelves of the Antarctic Peninsula and along the West Antarctic coast appear to be thinning significantly, with a 23 +/- 3 cm a(exp -1) surface elevation decrease on the Larsen ice shelf and a 65 +/- 4 cm a(exp -1) decrease on the Dotson ice shelf. Significant elevation decreases are obtained over most of the drainage basins of the Pine Island and Thwaites glaciers. Significant increases are obtained over most of the other grounded ice in Marie Byrd Land, the Antarctic Peninsula, and Coates Land. Over the sector from 85 degrees W to 115 degrees W, which includes the Pine Island and Thwaites basins, the average elevation is significantly decreasing by 8.1 cm a(exp -1). The corresponding ice thickness change is about -11 cm a(exp -1), with a corresponding mass loss of 82 Gt a(exp -1), and a 0.22 mm a(exp -1) contribution to global sea level rise. In terms of elevation change, the decrease in the Pine Island-Thwaites sector is largely balanced by the increase in the Marie Byrd Land, but only balanced by about 1/4 in terms of ice thickness change and contribution to sea level rise. The overall average elevation change for the grounded ice is + 1.2 cm a(exp -1). Using an average bedrock uplift of 2.5 cm a(exp -1), implies an average ice thickness decrease of 1.3 cm a(exp -1), a mass loss of 22 Gt a(exp -1), and a 0.06 mm a(exp -1) contribution to global sea level rise.

Zwally, H. Jay↗
NASA NTRS2006

Diurnal Variability of Vertical Structure from a TRMM Passive Microwave "Virtual Radar" Retrieval

Robust description of the diurnal cycle from TRMM observations is complicated by the limitations of Low Earth Orbit (LEO) sampling; from a 'climatological' perspective, sufficient sampling must exist to control for both spatial and seasonal variability, before tackling an additional diurnal component (e.g., with 8 additional 3-hourly or 24 1-hourly bins). For documentation of vertical structure, the narrow sample swath of the TRMM Precipitation Radar limits the resolution of any of these components. A neural-network based 'virtual radar" retrieval has been trained and internally validated, using multifrequency / multipolarization passive microwave(TM1) brightness temperatures and textures parameters and lightning (LIS) observations, as inputs, and PR volumetric reflectivity as targets (outputs). By training the algorithms (essentially highly multivariate, nonlinear regressions) on a very large sample of high-quality co-located data from the center of the TRMM swath, 3D radar reflectivity and derived parameters (VIL, IWC, Echo Tops, etc.) can be retrieved across the entire TMI swath, good to 8-9% over the dynamic range of parameters. As a step in the retrieval (and as an output of the process), each TMI multifrequency pixel (at 85 GHz resolution) is classified into one of the 25 archetypal radar profile vertical structure "types", previously identified using cluster analysis. The dynamic range of retrieved vertical structure appears to have higher fidelity than the current (Version 6) experimental GPROF hydrometeor vertical structure retrievals. This is attributable to correct representation of the prior probabilities of vertical structure variability in the neural network training data, unlike the GPROF cloud-resolving model training dataset used in the V6 algorithms. The LIS lightning inputs are supplementary inputs, and a separate offline neural network has been trained to impute (predict) LIS lightning from passive-microwave-only data. The virtual radar retrieval is thus, in principle, extensible to Aqua/AMSR-E and NPOESS/CMIS passive microwave instruments. The virtual radar approach yields a threefold increase in effective sampling from the mission, albeit of lower-quality "retrieved" data, reducing the variance of local estimates by one third (or the standard deviation by-0.57). In this talk, the variance reduction is leveraged to more finely resolve global diurnal variability in both space and time (local hour).

Boccippio, Dennis J.↗
NASA NTRS2008

Generation of Boundary Manikin Anthropometry

The purpose of this study was to develop 3D digital boundary manikins that are representative of the anthropometry of a unique population. These digital manikins can be used by designers to verify and validate that the components of the spacesuit design satisfy the requirements specified in the Human Systems Integration Requirements (HSIR) document. Currently, the HSIR requires the suit to accommodate the 1st percentile American female to the 99th percentile American male. The manikin anthropometry was derived using two methods: Principal Component Analysis (PCA) and Whole Body Posture Based Analysis (WBPBA). PCA is a statistical method for reducing a multidimensional data set by using eigenvectors and eigenvalues. The goal is to create a reduced data set that encapsulates the majority of the variation in the population. WBPBA is a multivariate analytical approach that was developed by the Anthropometry and Biomechanics Facility (ABF) to identify the extremes of the population for a given body posture. WBPBA is a simulation-based method that finds extremes in a population based on anthropometry and posture whereas PCA is based solely on anthropometry. Both methods yield a list of subjects and their anthropometry from the target population; PCA resulted in 20 female and 22 male subjects anthropometry and WBPBA resulted in 7 subjects' anthropometry representing the extreme subjects in the target population. The subjects anthropometry is then used to 'morph' a baseline digital scan of a person with the same body type to create a 3D digital model that can be used as a tool for designers, the details of which will be discussed in subsequent papers.

Young, Karen S.↗
NASA NTRSDate not supplied

Structured Covariance Gaussian Networks for Orion Crew Module Aerodynamic Uncertainty Quantification

In this paper we propose a new approach for nonlinear regression and uncertainty quantification. The method is based on a pair of neural networks which parameterize mean and dense covariance functions of a multivariate Gaussian process, trained together to maximize the log-likelihood of observing the given data. The covariance matrix is made positive definite at every input by construction. We also propose a sampling approach that produces viable surrogate function realizations from the Gaussian process. We call the proposed model a Structured Covariance Gaussian Network (SCGN). We illustrate the use of SCGNs for learning an aerodynamic response surface with built-in uncertainty for the Orion crew module. We find that SCGN provides an efficient and systematic way to learn nonlinear functional relationships and dense covariances. We compare results to a baseline Gaussian process regressor and observe that the SCGN provides comparable uncertainty descriptions with improved scalability to dataset size. The sample functions generated by SCGN are fast to evaluate online and are therefore convenient for use in trajectory simulations. These results suggest that SCGN may be a viable computational method for aerodynamic uncertainty quantification.

machine learning↗
NASA NTRSDate not supplied

Structured Covariance Gaussian Networks for Orion Crew Module Aerodynamic Uncertainty Quantification

In this paper we propose a new approach for nonlinear regression and uncertainty quantification. The method is based on a pair of neural networks which parameterize mean and dense covariance functions of a multivariate Gaussian process, trained together to maximize the log-likelihood of observing the given data. The covariance matrix is made positive definite at every input by construction. We also propose a sampling approach that produces viable surrogate function realizations from the Gaussian process. We call the proposed model a Structured Covariance Gaussian Network (SCGN). We illustrate the use of SCGNs for learning an aerodynamic response surface with built-in uncertainty for the Orion crew module. We find that SCGN provides an efficient and systematic way to learn nonlinear functional relationships and dense covariances. We compare results to a baseline Gaussian process regressor and observe that the SCGN provides comparable uncertainty descriptions with improved scalability to dataset size. The sample functions generated by SCGN are fast to evaluate online and are therefore convenient for use in trajectory simulations. These results suggest that SCGN may be a viable computational method for aerodynamic uncertainty quantification.

machine learning↗