Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “regression models”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 883 records · Page 49
All sourcesNASA NTRSDOE OSTI
NASA NTRS2019

Machine Learning Application to Atmospheric Chemistry Modeling

Atmospheric chemistry models are a central tool to study the impact of chemical constituents on the environment, vegetation and human health. These models split the atmosphere in a large number of grid-boxes and consider the emission of compounds into these boxes and their subsequent transport, deposition, and chemical processing. The chemistry is represented through a series of simultaneous ordinary differential equations, one for each compound. Given the difference in life-times between the chemical compounds (milli-seconds for O (sup 1) D (Deuterium) to years for CH4) these equations are numerically stiff and solving them consists of a significant fraction of the computational burden of a chemistry model. We have investigated a machine learning approach to emulate the chemistry instead of solving the differential equations numerically. From a one-month simulation of the GEOS-Chem model we have produced a training dataset consisting of the concentration of compounds before and after the differential equations are solved, together with some key physical parameters for every grid-box and time-step. From this dataset we have trained a machine learning algorithm (regression forest) to be able to predict the concentration of the compounds after the integration step based on the concentrations and physical state at the beginning of the time step. We have then included this algorithm back into the GEOS-Chem model, bypassing the need to integrate the chemistry. This machine learning approach shows many of the characteristics of the full simulation and has the potential to be substantially faster. There are a wide range of application for such an approach - generating boundary conditions, for use in air quality forecasts, chemical data assimilation systems, etc. We discuss speed and accuracy of our approach, and highlight some potential future directions for improving it.

Keller, Christoph A.↗
NASA NTRS2019

Atmospheric Chemistry Modeling and Air Quality Forecasting Using Machine Learning

Atmospheric chemistry models are a central tool to study the impact of chemical constituents on the environment, vegetation and human health. These models split the atmosphere in a large number of grid-boxes and consider the emission of compounds into these boxes and their subsequent transport, deposition, and chemical processing. The chemistry is represented through a series of simultaneous ordinary differential equations, one for each compound. Given the difference in life-times between the chemical compounds (milli-seconds for O1D to years for CH4) these equations are numerically stiff and solving them consists of a significant fraction of the computational burden of a chemistry model.We have investigated a machine learning approach to emulate the chemistry instead of solving the differential equations numerically. From a one-month simulation of the GEOS-Chem model we have produced a training dataset consisting of the concentration of compounds before and after the differential equations are solved, together with some key physical parameters for every grid-box and time-step. From this dataset we have trained a machine learning algorithm (regression forest) to be able to predict the concentration of the compounds after the integration step based on the concentrations and physical state at the beginning of the time step. We have then included this algorithm back into the GEOS-Chem model, bypassing the need to integrate the chemistry.This machine learning approach shows many of the characteristics of the full simulation and has the potential to be substantially faster. There are a wide range of application for such an approach - generating boundary conditions, for use in air quality forecasts, chemical data assimilation systems, etc. We discuss speed and accuracy of our approach, and highlight some potential future directions for improving it.

Keller, Christoph A.↗
NASA NTRS1979

Reduction of particulate carryover from a pressurized fluidized bed

A bench-scale pressurized fluidized bed combustor (PFBC) constructed with a conical shape to reduce the particulate carryover is examined. The combustor was fed coal and limestone with the coal to air ratio varying from 0.033 to 0.098 (all lean) and the coal to limestone ratio varying from 0.06 to 0.36. Two cone angles were used and it is shown that the average particulate carryover of 2.5 grains/SCF is appreciably less than cylindrical fluidized bed combustors. In addition the carryover was correlated by multiple regression analysis to yield the dependence on bed depth and hence the collection efficiency, which was 20%. Finally, a comparison with a model indicated that the exhaust port may be below the transport disengaging height for most of the tests showing that further improvements could be achieved by increasing the freeboard height of the exhaust port above the bed.

Patch, R. W.↗
NASA NTRS1990

Threshold altitude resulting in decompression sickness

A review of case reports, hypobaric chamber training data, and experimental evidence indicated that the threshold for incidence of altitude decompression sickness (DCS) was influenced by various factors such as prior denitrogenation, exercise or rest, and period of exposure, in addition to individual susceptibility. Fitting these data with appropriate statistical models makes it possible to examine the influence of various factors on the threshold for DCS. This approach was illustrated by logistic regression analysis on the incidence of DCS below 9144 m. Estimations using these regressions showed that, under a noprebreathe, 6-h exposure, simulated EVA profile, the threshold for symptoms occurred at approximately 3353 m; while under a noprebreathe, 2-h exposure profile with knee-bends exercise, the threshold occurred at 7925 m.

Kumar, K. V.↗
NASA NTRS2001

A semi-automatic method for left ventricle volume estimate: an in vivo validation study

This study aims to the validation of the left ventricular (LV) volume estimates obtained by processing volumetric data utilizing a segmentation model based on level set technique. The validation has been performed by comparing real-time volumetric echo data (RT3DE) and magnetic resonance (MRI) data. A validation protocol has been defined. The validation protocol was applied to twenty-four estimates (range 61-467 ml) obtained from normal and pathologic subjects, which underwent both RT3DE and MRI. A statistical analysis was performed on each estimate and on clinical parameters as stroke volume (SV) and ejection fraction (EF). Assuming MRI estimates (x) as a reference, an excellent correlation was found with volume measured by utilizing the segmentation procedure (y) (y=0.89x + 13.78, r=0.98). The mean error on SV was 8 ml and the mean error on EF was 2%. This study demonstrated that the segmentation technique is reliably applicable on human hearts in clinical practice.

NASA Discipline Cardiopulmonary↗
NASA NTRS2004

Top-of-Atmosphere Direct Radiative Effect of Aerosols over Global Oceans from Merged CERES and MODIS Observations

The direct radiative effect of aerosols (DREA) is defined as the difference between radiative fluxes in the absence and presence of aerosols. In this study, the direct radiative effect of aerosols is estimated for 46 months (March, 2000 to December, 2003) of merged CERES and MODIS Terra global measurements over ocean. This analysis includes the contribution from clear regions in both clear and partly cloudy CERES footprints. MODIS-CERES narrow-to-broadband regressions are developed to convert clear-sky MODIS narrowband radiances to broadband SW radiances, and CERES clear-sky Angular Distribution Models (ADMs) are used to estimate the corresponding TOA radiative fluxes needed to determine the DREA. Clear-sky MODIS pixels are identified using two independent cloud masks: (i) the NOAA-NESDIS algorithm used for inferring aerosol properties from MODIS on the CERES Single Scanner Footprint TOA/Surface Fluxes and Clouds (SSF) product (NOAA-SSF); and (ii) the standard algorithm used by the MODIS aerosol group to produce the MODO4 product (MODO4). Over global oceans, direct radiative cooling by aerosols for clear scenes identified from MODO4 is estimated to be 5.5 W m-2, compared to 3.8 W m-2 for clear scenes from NOAA-SSF. Regionally, differences are largest in areas affected by dust aerosol, such as oceanic regions adjacent to the Saharan and Saudi Arabian deserts, and in northern Pacific Ocean regions influenced by dust transported from Asia. The net total-sky (clear and cloudy) DREA is negative (cooling) and is estimated to be -2.0 W m-2 from MOD04, and -1.6 W m-2 from NOAA-SSF. The DREA is shown to have pronounced seasonal cycles in the Northern Hemisphere and large year-to-year fluctuations near deserts. However, no systematic trend in deseasonalized anomalies of the DREA is observed over the 46-month time series considered.

Loeb, N. G.↗
NASA NTRS2005

Reference Models for Structural Technology Assessment and Weight Estimation

Previously the Exploration Concepts Branch of NASA Langley Research Center has developed techniques for automating the preliminary design level of launch vehicle airframe structural analysis for purposes of enhancing historical regression based mass estimating relationships. This past work was useful and greatly reduced design time, however its application area was very narrow in terms of being able to handle a large variety in structural and vehicle general arrangement alternatives. Implementation of the analysis approach presented herein also incorporates some newly developed computer programs. Loft is a program developed to create analysis meshes and simultaneously define structural element design regions. A simple component defining ASCII file is read by Loft to begin the design process. HSLoad is a Visual Basic implementation of the HyperSizer Application Programming Interface, which automates the structural element design process. Details of these two programs and their use are explained in this paper. A feature which falls naturally out of the above analysis paradigm is the concept of "reference models". The flexibility of the FEA based JAVA processing procedures and associated process control classes coupled with the general utility of Loft and HSLoad make it possible to create generic program template files for analysis of components ranging from something as simple as a stiffened flat panel, to curved panels, fuselage and cryogenic tank components, flight control surfaces, wings, through full air and space vehicle general arrangements.

Cerro, Jeff↗
NASA NTRS2021

Space-Borne Cloud-Native Satellite-Derived Bathymetry (SDB) Models Using ICESat-2 And Sentinel-2

Shallow nearshore coastal waters provide a wealth of societal, economic and ecosystem services, yet their topographic structure is poorly mapped due to a reliance upon expensive and time intensive methods. Space‐borne bathymetric mapping has helped address these issues, but has remained largely dependent upon in situ measurements. Here we fuse ICESat‐2 lidar data with Sentinel‐2 optical imagery, within the Google Earth Engine cloud platform, to create openly available spatially continuous high‐resolution bathymetric maps at regional‐to‐national scales in Florida, Crete and Bermuda. ICESat‐2 bathymetric classified photons are used to train three Satellite Derived Bathymetry (SDB) methods, including Lyzenga, Stumpf and Support Vector Regression algorithms. For each study site the Lyzenga algorithm yielded the lowest RMSE (approx. 10‐15%) when compared with validation data. We demonstrate a means of using ICESat‐2 for both model calibration and validation, thus cementing a pathway for fully space‐borne estimates of nearshore bathymetry in shallow, clear water environments.

N. Thomas↗
NASA NTRSDate not supplied

Sampling Functions from Gaussian Processes and Structured Covariance Gaussian Networks

When learning aerodynamic models from data, it is critical to incorporate estimates of model uncertainty. This motivates the design of probabilistic aerodynamic databases which can be sampled to generate physically and statistically plausible aerodynamic models. In this talk we discuss how to sample deterministic functions from two different kinds of probabilistic models and demonstrate their use. First, Gaussian Process Regressors (GPRs) are a widely used probabilistic kernel-based model which can be thought of as Gaussian distributions over functions. GPRs are generally trained by maximizing the marginal likelihood of seeing the training data over the kernel parameter space. Sample functions are easily generated by drawing points from the Gaussian distribution at desired input points. However, when the points are not known ahead of time, the classical sampling approach is not possible since successive function samples will generate different function realizations. We present an approach for sampling consistent function evaluations from a GPR over multiple samples. Second, we describe a neural network architecture which learns a conditional Gaussian distribution by maximizing the marginal likelihood at each point in the input space. We then discuss and compare several options for generating sample functions which match this distribution. Finally, we demonstrate the use of these probabilistic aerodynamic models in an atmospheric reentry simulation.

Gaussian process regression↗
NASA NTRSDate not supplied

Early Information Parameter-Set Analysis for Satellite Close Approaches using Machine Learning

Understanding orbital mechanics is essential in space flight and navigation applications, and leveraging modern force models for flight path projection remains an important aspect in space mission design and operation. However, force models do not capture all the dynamics or perturbations in the space environment and thus are subject to errors in predicting the state vectors. The further out the predicted miss distance between spacecraft is from the time of closest approach (TCA), the larger the propagated errors in the predicted miss distance at TCA is. The dependency on these force models for spacecraft flight state prediction calls for a more reliable method that can quantify, or even reduce, these propagated errors. With recent advances in the field artificial intelligence, specifically in machine and deep learning algorithms, a model that implements these approaches can improve on the modern force model approach. The goal for this work is to provide an early-information decision-making threshold, in order to prioritize risk assessment implementation, given the ongoing increase of space objects. In analyzing the relationship of several parameters from conjunction data messages(CDMs) and solar information, early information becomes viable in miss distance prediction with unsupervised learning techniques, which learn the parameters that are linked together with miss distance and probability of collision (Pc) variables. Another approach implemented for identifying relationships within CDMs is supervised learning, in which a shallow neural network binary classifier learns to distinguish events with Pc values¡108. These parameters detected in the unsupervised process are then applied to a regression neural network, which predicts the miss distance at TCA for a specific event within a given uncertainty bound. For the regression neural network, a Long Short Term Memory (LSTM) neural network is implemented, which yields memory about each time step in an event. Using an LSTM network, the model learns to predict miss distance within 0.2km of the value measured at TCA. Although there is a limited amount of "close miss" data to train a network, the network learns to associate parameters, like large energy dissipation rates with the secondary object, with an elevated Pc

Brianna I. Robertson↗
NASA NTRS1987

A model-free method for mass spectrometer response correction

A new method for correction of mass spectrometer output signals is described. Response-time distortion is reduced independently of any model of mass spectrometer behavior. The delay of the system is found first from the cross-correlation function of a step change and its response. A two-sided time-domain digital correction filter (deconvolution filter) is generated next from the same step response data using a regression procedure. Other data are corrected using the filter and delay. The mean squared error between a step response and a step is reduced considerably more after the use of a deconvolution filter than after the application of a second-order model correction. O2 consumption and CO2 production values calculated from data corrupted by a simulated dynamic process return to near the uncorrupted values after correction. Although a clean step response or the ensemble average of several responses contaminated with noise is needed for the generation of the filter, random noise of magnitude not above 0.5 percent added to the response to be corrected does not impair the correction severely.

Shykoff, Barbara E.↗
NASA NTRS1992

Development of an empirically based dynamic biomechanical strength model

The focus here is on the development of a dynamic strength model for humans. Our model is based on empirical data. The shoulder, elbow, and wrist joints are characterized in terms of maximum isolated torque, position, and velocity in all rotational planes. This information is reduced by a least squares regression technique into a table of single variable second degree polynomial equations determining the torque as a function of position and velocity. The isolated joint torque equations are then used to compute forces resulting from a composite motion, which in this case is a ratchet wrench push and pull operation. What is presented here is a comparison of the computed or predicted results of the model with the actual measured values for the composite motion.

Pandya, A.↗
NASA NTRS1984

Comments on 'Inference of cloud temperature and thickness by microwave radiometry from space'

The method proposed by Pandey et al. (1983) for estimating the temperature differential and thickness of clouds from microwave data obtained with the scanning multichannel microwave radiometers of the Seasat and Nimbus-7 satellites is examined critically. It is pointed out that both the thicknesses and the temperature differentials derived from them may not be meaningful unless accurate measurements of cloud-top height (from IR radiometry) and reliable data on liquid-water content are available. It is suggested that the good fits obtained in generating regression coefficients for the proposed method may be artifacts of the fixed or limited-range liquid-water densities of the cloud models used. With respect to cloud-top height, the need to quantify and account for differences in the fields of view and spatial resolutions of the IR and microwave radiometers, as undertaken for the case of precipitating clouds by Yeh and Liou (1983), is stressed.

Yeh, H.-Y. M.↗
NASA NTRS1986

Equations for Annular-Heat-Transfer Coefficients

Tables of coefficients converted to algebraic expressions. Plot of Equation for Nusselt number agrees closely with points from tabulated data. Equation for Nusselt number and those for coefficients A and B obtained by regression analysis of data. Other plots also show close agreement for radius of 0.1 and 0.2. In equation form, coefficients incorporated into mathematical models more readily than as tabular data. Equations simplify design and analysis of heat exchangers.

Yao, B.↗
NASA NTRS1991

The effect of clouds on the earth's radiation budget

The radiative fluxes from the Earth Radiation Budget Experiment (ERBE) and the cloud properties from the International Satellite Cloud Climatology Project (ISCCP) over Indonesia for the months of June and July of 1985 and 1986 were analyzed to determine the cloud sensitivity coefficients. The method involved a linear least squares regression between co-incident flux and cloud coverage measurements. The calculated slope is identified as the cloud sensitivity. It was found that the correlations between the total cloud fraction and radiation parameters were modest. However, correlations between cloud fraction and IR flux were improved by separating clouds by height. Likewise, correlations between the visible flux and cloud fractions were improved by distinguishing clouds based on optical depth. Calculating correlations between the net fluxes and either height or optical depth segregated cloud fractions were somewhat improved. When clouds were classified in terms of their height and optical depth, correlations among all the radiation components were improved. Mean cloud sensitivities based on the regression of radiative fluxes against height and optical depth separated cloud types are presented. Results are compared to a one-dimensional radiation model with a simple cloud parameterization scheme.

Ziskin, Daniel↗
NASA NTRS2004

CMC Property Variability and Life Prediction Methods for Turbine Engine Component Application

The ever increasing need for lower density and higher temperature-capable materials for aircraft engines has led to the development of Ceramic Matrix Composites (CMCs). Today's aircraft engines operate with >3000"F gas temperatures at the entrance to the turbine section, but unless heavily cooled, metallic components cannot operate above approx.2000 F. CMCs attempt to push component capability to nearly 2700 F with much less cooling, which can help improve engine efficiency and performance in terms of better fuel efficiency, higher thrust, and reduced emissions. The NASA Glenn Research Center has been researching the benefits of the SiC/SiC CMC for engine applications. A CMC is made up of a matrix material, fibers, and an interphase, which is a protective coating over the fibers. There are several methods or architectures in which the orientation of the fibers can be manipulated to achieve a particular material property objective as well as a particular component geometric shape and size. The required shape manipulation can be a limiting factor in the design and performance of the component if there is a lack of bending capability of the fiber as making the fiber more flexible typically sacrifices strength and other fiber properties. Various analysis codes are available (pcGINA, CEMCAN) that can predict the effective Young's Moduli, thermal conductivities, coefficients of thermal expansion (CTE), and various other properties of a CMC. There are also various analysis codes (NASAlife) that can be used to predict the life of CMCs under expected engine service conditions. The objective of this summer study is to utilize and optimize these codes for examining the tradeoffs between CMC properties and the complex fiber architectures that will be needed for several different component designs. For example, for the pcGINA code, there are six variations of architecture available. Depending on which architecture is analyzed, the user is able to specify the fiber tow size, tow spacing, weave parameter, and angle of orientation of fibers. By holding the volume fraction of the fibers constant, variations in tow spacing can be explored for different architectures. The CMC material properties are usually calculated assuming the component is manufactured perfectly. However, this is typically not the case so that a quantification of the material property variability is needed to account for processing and/or manufacturing imperfections. The overall inputs and outputs are presented using a regression software to rapidly investigate the tradeoffs associated with fiber architecture, material properties, and ultimately cost. This information is then propagated through lifing models and Larson-Miller data to assess timehemperature-dependent CMC strength. In addition, a first order cost estimation will be quantified from a current qualitative perspective. This cost estimation includes the manufacturing challenges, such as tooling, as well as the component cost for a particular application. Ultimately, a cost to performance ratio should be established that compares the effectiveness of CMCs to their current rival, nickel superalloys.

Cheplak, Matthew L.↗
NASA NTRS2000

Cascade Optimization Strategy with Neural Network and Regression Approximations Demonstrated on a Preliminary Aircraft Engine Design

A preliminary aircraft engine design methodology is being developed that utilizes a cascade optimization strategy together with neural network and regression approximation methods. The cascade strategy employs different optimization algorithms in a specified sequence. The neural network and regression methods are used to approximate solutions obtained from the NASA Engine Performance Program (NEPP), which implements engine thermodynamic cycle and performance analysis models. The new methodology is proving to be more robust and computationally efficient than the conventional optimization approach of using a single optimization algorithm with direct reanalysis. The methodology has been demonstrated on a preliminary design problem for a novel subsonic turbofan engine concept that incorporates a wave rotor as a cycle-topping device. Computations of maximum thrust were obtained for a specific design point in the engine mission profile. The results (depicted in the figure) show a significant improvement in the maximum thrust obtained using the new methodology in comparison to benchmark solutions obtained using NEPP in a manual design mode.

Hopkins, Dale A.↗
NASA NTRS2010

Decompression Sickness After Air Break in Prebreathe Described with a Survival Model

Data from Brooks City-Base show the decompression sickness (DCS) and venous gas emboli (VGE) consequences of air breaks in a resting 100% O2 prebreathe (PB) prior to a hypobaric exposure. METHODS: DCS and VGE survival times from 95 controls for a 60 min PB prior to 2-hr or 4-hr exposures to 4.37 psia are statistically compared to 3 break in PB conditions: a 10 min (n=40), 20 min (n=40), or 60 min break (n=32) 30 min into the PB followed by 30 min of PB. Ascent rate was 1,524 meters / min and all exposures included light exercise and 4 min of VGE monitoring of heart chambers at 16 min intervals. DCS survival time for combined control and air breaks were described with an accelerated log logistic model where exponential N2 washin during air break was described with a 10 min half-time and washout during PB with a 60 min half-time. RESULTS: There was no difference in VGE or DCS survival times among 3 different air breaks, or when air breaks were compared to control VGE times. However, 10, 20, and 60 min air breaks had significantly earlier survival times compared to control DCS times, certainly early in the exposures. CONCLUSION: Air breaks of 10, 20, and 60 min after 30 min of a 60 min PB reduced DCS survival time. The survival model combined discrete comparisons into a global description mechanistically linked to asymmetrical N2 washin and washout kinetics based on inspired pN2. Our unvalidated regression is used to compute additional PB time needed to compensate for an air break in PB within the range of tested conditions.

Conkin, J.↗