Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “nonlinear regression”

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 55 records · Page 3
All sourcesNASA NTRSDOE OSTI
DOE OSTI2017

Growth and nonlinear saturation of electromagnetic ion cyclotron waves in multi-ion species magnetospheric plasma

Here, the growth and saturation of electromagnetic ion cyclotron (EMIC) waves is essential to the magnetospheric dynamics. Determining and isolating the effects of multiple ion parameters such as temperatures, anisotropies, and relative abundances is important for quantifying these processes in the magnetospheric plasma. In order to study these process, we utilize a 2.5–D hybrid model (where ions are modeled with the particle–in–cell (PIC) method, and electrons are modeled as background neutralizing fluid) to study the nonlinear electromagnetic wave–particle interactions of hot H + , cold H + , cold He + , and cold or hot O + ions for a broad range of typical magnetospheric parameters. The excitation of EMIC waves is driven by the temperature anisotropy of hot H + in our model. As a result, we quantify the parametric dependence of the linear growth, the nonlinear saturation level of perpendicular magnetic fluctuations, and the temporal evolution of the ion temperature anisotropies. We establish the relation between key plasma parameters and the saturated EMIC wave power, using either power law fits or a nonlinear regression method. We construct the dispersion relation of the waves using the results of the model and investigate the energy content in the various branches of the dispersion ( k ∥ –ω space), showing that the different modes can generate wave power in different regions of k space. We find that large O + concentration reduces the growth and saturated amplitude of the waves; but the waves are less sensitive to the temperature of the O + in the temperature range relevant to the magnetosphere.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2021

Deep Learning for Rapid Analysis of Spectroscopic Ellipsometry Data

High‐throughput experimental approaches to rapidly develop new materials require high‐throughput data analysis methods to match. Spectroscopic ellipsometry is a powerful method of optical properties characterization, but for unknown materials and/or layer structures the data analysis using traditional methods of nonlinear regression is too slow for autonomous, closed‐loop, high‐throughput experimentation. Herein, three methods (termed spectral, piecewise, and pointwise) of spectroscopic ellipsometry data analysis based on deep learning are introduced and studied. After initial training, the incremental time for inferring optical properties can be a thousand times faster than traditional methods. Results for multilayer sample structures with optically isotropic materials are presented, appropriate for high‐throughput studies of thin films of phase‐change materials such as GeSbTe (GST) alloys. Results for studies on highly birefringent layered materials are also presented, exemplified by the transition metal dichalcogenide MoS 2 . How the materials under test and the experimental objectives may guide the choice of analysis methods are discussed. The utility of our approach is demonstrated by analyzing data measured on a composition spread of GeSbTe phase‐change alloys containing 177 distinct compositions, and identifying the composition with optimal phase‐change figure of merit in only 1.4 s of analysis time.

Li, Yifei↗
DOE OSTI2018

Energy modeling and efficiency analysis of aluminum die-casting processes

Energy modeling and efficiency analysis are considered the foundation of manufacturing process optimization to improve quality and efficiency and reduce energy consumption and carbon emissions during aluminum die-casting processes. Our report introduced an energy modeling method to connect gas and electric energy consumption with production rate of aluminum die-casting processes based on data collected at workshops with various combination of machines and products. The detailed modeling process involved the development of a data-acquiring system and the comparison of various kinds of nonlinear regression methods. The resulting models were confirmed with actual production data and were also used to improve production scheduling. Furthermore, it was found that if the modeling results are reasonably used and production is accordingly well-scheduled, 10 to 15% of energy savings could be realized without sacrificing profits.

42 ENGINEERING↗
DOE OSTI2018

Transient analysis of advancing contact angle measurements on polished rock surfaces

We report that contact angle measurements for gas-liquid-rock systems are important for modeling multi-phase flow and transport in the subsurface. These data are needed in applications such as the extraction of oil and gas resources, geologic sequestration of carbon dioxide, contaminant fate and transport, and aquifer recharge through the vadose zone. Contact angles are frequently measured with the sessile drop method. Previous research has largely ignored the dynamic behavior of sessile droplets on geologic materials. This study investigates the dynamic behavior of sessile water droplets on prepared rock surfaces in the presence of air. Droplet diameter and advancing contact angle were determined at 0.5 s intervals for ~90 s on flat polished disks of Burlington limestone, Crossville sandstone, Mancos shale, Sierra White granite, Vermilion Bay granite, and Westerly granite using a Krüss DSA 30 Drop Shape Analyzer. The droplet diameter and advancing contact angle data sets were nonlinearly regressed against time using two different two-parameter models. The median coefficients of determination for the fits were 0.85 and 0.96, respectively. The resulting parameter estimates were used to compute the apparent equilibrium contact angle, θ e , for each disk following droplet diameter stabilization. Estimates of θ e ranged from 37.2° for Mancos shale to 75.6° for Burlington limestone. Analysis of variance indicated statistically significant differences in θ e between the rock types at the 95% confidence level. The variability of θe on the polished rock surfaces, as quantified by the coefficient of variation (CV) for θ e , varied between ~3 and ~9%; there were no significant differences in CV between the rock types. Neutron radiography indicated changes droplet morphology over time were due to the spontaneous imbibition of water into the rock matrix. Finally, the transient analysis employed in this study permits a more meaningful estimate of the equilibrium contact angle for rocks than taking the initial value or averaging over time as is frequently done.

58 GEOSCIENCES↗
DOE OSTI2018

State space parameterization of explosive eigenvalues during autoignition

Explosive modes such as ignition and extinction are characterized by an eigenvalue of the chemical Jacobian matrix with positive real part, representing the transient instability of chain-branching chemistry and thermal feedback. Formation and eigen-decomposition of the Jacobian matrix are expensive operations whose cost increases cubically with chemical mechanism size. As an alternative to directly computing the eigenvalues of the Jacobian, we explore principal component analysis (PCA) along with nonlinear regression as a methodology to parameterize the eigenvalues by state variables (or linear combinations thereof). We evaluate this modeling strategy using homogeneous autoignition data on two different applications: pseudotransient continuation (Ψtc)-based ODE solvers and chemical explosive mode analysis (CEMA). Results indicate that the PCA-based parameterization of the eigenvalues appears feasible for Ψtc solvers in autoignition calculations over a range of temperatures and pressures. Our results also show that eigenvalue models are capable of tracking sharp discontinuities (such as ignition or extinction) in the eigenvalue for computational flame diagnostics such as CEMA.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2016

Capillary pressure – saturation relationships for gas shales measured using a water activity meter

Hydraulic fracturing of gas shale formations involves pumping a large volume of fracking fluid into a hydrocarbon reservoir to fracture the rock and thus increase its permeability. The majority of the fracking fluid introduced is never recovered and the fate of this lost fluid, often called “leak off,” has become the source of much debate. Information on the capillary pressure – saturation relationship for each wetting phase is needed to simulate leak off using numerical reservoir models. The petroleum industry commonly employs air – water capillary pressure – saturation curves to predict these relationships for mixed wet reservoirs. Traditional methods of measuring this curve are unsuitable for gas shales due to high capillary pressures associated with the small pores present. Still, a possible alternative method is the water activity meter which is used widely in the soil sciences for such measurements. However, its application to lithified material has been limited. Here, this study utilized a water activity meter to measure air – water capillary pressures (ranging from 1.3 to 219.6 MPa) at several water saturation levels in both the wetting and drying directions. Water contents were measured gravimetrically. Seven types of gas producing shale with different porosities (2.5–13.6%) and total organic carbon contents (0.4–13.5%) were investigated. Nonlinear regression was used to fit the resulting capillary pressure – water saturation data pairs for each shale type to the Brooks and Corey equation. Data for six of the seven shale types investigated were successfully fitted (median R 2 = 0.93), indicating this may be a viable method for parameterizing capillary pressure – saturation relationships for inclusion in numerical reservoir models. As expected, the different shale types had statistically different Brooks and Corey parameters. However, there were no significant differences between the Brooks and Corey parameters for the wetting and drying measurements, suggesting that hysteresis may not need to be taken into account in leak off simulations.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2017

Single toxin dose-response models revisited

The goal of this paper is to offer a rigorous analysis of the sigmoid shape single toxin dose-response relationship. The toxin efficacy function is introduced and four special points, including maximum toxin efficacy and inflection points, on the dose-response curve are defined. The special points define three phases of the toxin effect on mortality: (1) toxin concentrations smaller than the first inflection point or (2) larger then the second inflection point imply low mortality rate, and (3) concentrations between the first and the second inflection points imply high mortality rate. Probabilistic interpretation and mathematical analysis for each of the four models, Hill, logit, probit, and Weibull is provided. Two general model extensions are introduced: (1) the multi-target hit model that accounts for the existence of several vital receptors affected by the toxin, and (2) model with a nonzero mortality at zero concentration to account for natural mortality. Special attention is given to statistical estimation in the framework of the generalized linear model with the binomial dependent variable as the mortality count in each experiment, contrary to the widespread nonlinear regression treating the mortality rate as continuous variable. The models are illustrated using standard EPA Daphnia acute (48 h) toxicity tests with mortality as a function of NiCl or CuSO{sub 4} toxin. - Highlights: • The paper offers a rigorous study of a sigmoid dose-response relationship. • The concentration with highest mortality rate is rigorously defined. • A table with four special points for five morality curves is presented. • Two new sigmoid dose-response models have been introduced. • The generalized linear model is advocated for estimation of sigmoid dose-response relationship.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2015

Machine Learning for Quantum Mechanical Properties of Atoms in Molecules

Here we introduce machine learning models of quantum mechanical observables of atoms in molecules. Instant out-of-sample predictions for proton and carbon nuclear chemical shifts, atomic core level excitations, and forces on atoms reach accuracies on par with density functional theory reference. Locality is exploited within nonlinear regression via local atom-centered coordinate systems. The approach is validated on a diverse set of 9 k small organic molecules. Linear scaling of computational cost in system size is demonstrated for saturated polymers with up to submesoscale lengths.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

Data-Efficient Methods for Determining Flory–Huggins χ Parameters in Multicomponent Polymer Formulations

Polymer formulations are essential in diverse applications including personal care products, coatings, paints, adhesives, and plastic materials. Designing these formulations requires navigating large, complex design spaces, where phase and self-assembly behavior critically impact performance. The Flory–Huggins χ parameter, which quantifies segmental miscibility, is widely used to parametrize the excess free energy of mixing in formulation models. In this work, we introduce two data-efficient, top-down methods for estimating χ parameters using the Random Phase Approximation (RPA): (i) Boundary Nonlinear Regression (Boundary-NLR), which fits theoretical spinodal boundaries to experimental phase boundaries, and (ii) Surrogate Model Inverse Parameter Estimation (SMIPE), which uses a Gaussian Process Classifier to fit sparse phase maps via a surrogate model. Both methods allow rapid parametrization of polymer field-theoretic models without the need for additional experiments. We evaluate these approaches on data sets involving polymer–solvent–nonsolvent ternary mixtures and block copolymer–solvent systems, demonstrating their robustness to experimental noise and their relevance for real-world formulation design.

copolymers↗
DOE OSTI2013

Synchrotron radiation-based far-infrared spectroscopic ellipsometer with full Mueller-matrix capability

We developed far-IR spectroscopic ellipsometer at the U4IR beamline of the National Synchrotron Light Source in Brookhaven National Laboratory. This ellipsometer is able to measure both, rotating analyzer and full-Mueller matrix spectra using rotating retarders, and wire-grid linear polarizers. We utilize exceptional brightness of synchrotron radiation in the broad spectral range between about 20 and 4000 cm{sup -1}. Fourier-transform infrared (FT-IR) spectrometer is used for multi-wavelength data acquisition. The sample stage has temperature variation between 4.2 and 450 K, wide range of {theta}-2{theta} angular rotation, {chi} tilt angle adjustment, and X-Y-Z translation. A LabVIEW-based software controls the motors, sample temperature, and FT-IR spectrometer and also allows to run fully automated experiments with pre-programmed measurement schedules. Data analysis is based on Berreman's 4 Multiplication-Sign 4 propagation matrix formalism to calculate the Mueller matrix parameters of anisotropic samples with magnetic permeability {mu}{ne} 1. A nonlinear regression of the rotating analyzer ellipsometry and/or Mueller matrix (MM) spectra, which are usually acquired at variable angles of incidence and sample crystallographic orientations, allows extraction of dielectric constant and magnetic permeability tensors for bulk and thin-film samples. Applications of this ellipsometer setup for multiferroic and ferrimagnetic materials with {mu}{ne} 1 are illustrated with experimental results and simulations for TbMnO{sub 3} and Dy{sub 3}Fe{sub 5}O{sub 12} single crystals. We demonstrate how magnetic and electric dipoles, such as magnons and phonons, can be distinguished from a single MM measurement without adducing any modeling arguments. The parameters of magnetoelectric components of electromagnon excitations are determined using MM spectra of TbMnO{sub 3}.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2022

Multi-species temperature and number density analysis of a laser-produced plasma using dual-comb spectroscopy

Dual-comb spectroscopy (DCS) represents a novel method of using absorption spectroscopy as a diagnostic tool for multispecies analysis of excitation temperatures and column densities in laser-produced plasmas (LPPs). DCS was performed on a LPP generated by ablating a multielement alloy containing Nd, Gd, and Fe. Transitions from all three elements were observed in absorption spectra measured from 530.08 to 535.19 nm at seven time-delays from 31 to 250 μs after ablation. The spectra were fit using a nonlinear regression algorithm to determine peak areas, and excitation temperatures and column densities were determined for the three atomic species separately using Boltzmann plots. The measured excitation temperatures of Nd I and Gd I showed good agreement at all time-delays, whereas the Fe I temperature was found to be higher, and the ratios between the column densities varied with delay. The observations are understood via effects of LPP spatial averaging, elemental fractionation, and molecular formation and are compared and contextualized with previous work studying LPPs using other spectroscopic techniques. A brief discussion of the precision and accuracy of the determined excitation temperatures and column densities is also presented.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2024

Machine learning without a processor: Emergent learning in a nonlinear analog network

Standard deep learning algorithms require differentiating large nonlinear networks, a process that is slow and power-hungry. Electronic contrastive local learning networks (CLLNs) offer potentially fast, efficient, and fault-tolerant hardware for analog machine learning, but existing implementations are linear, severely limiting their capabilities. These systems differ significantly from artificial neural networks as well as the brain, so the feasibility and utility of incorporating nonlinear elements have not been explored. Here, we introduce a nonlinear CLLN—an analog electronic network made of self-adjusting nonlinear resistive elements based on transistors. We demonstrate that the system learns tasks unachievable in linear systems, including XOR (exclusive or) and nonlinear regression, without a computer. We find our decentralized system reduces modes of training error in order (mean, slope, curvature), similar to spectral bias in artificial neural networks. The circuitry is robust to damage, retrainable in seconds, and performs learned tasks in microseconds while dissipating only picojoules of energy across each transistor. This suggests enormous potential for fast, low-power computing in edge systems like sensors, robotic controllers, and medical devices, as well as manufacturability at scale for performing and studying emergent learning.

Science & Technology - Other Topics↗
DOE OSTI2024

Real-time capable modeling of ICRF heating on NSTX and WEST via machine learning approaches

Abstract A real-time capable core Ion Cyclotron Range of Frequencies (ICRF) heating model on NSTX and WEST is developed. The model is based on two nonlinear regression algorithms, the random forest ensemble of decision trees and the multilayer perceptron neural network. The algorithms are trained on TORIC ICRF spectrum solver simulations of the expected flat-top operation scenarios in NSTX and WEST assuming Maxwellian plasmas. The surrogate models are shown to successfully capture the multi-species core ICRF power absorption predicted by the original model for the high harmonic fast wave and the ion cyclotron minority heating schemes while reducing the computational time by six orders of magnitude. Although these models can be expanded, the achieved regression scoring, computational efficiency and increased model robustness suggest these strategies can be implemented into integrated modeling frameworks for real-time control applications.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2018

Improving BA Control Performance Through Advanced Regulation Requirements Prediction

The paper presents a comprehensive approach to predict balancing authority (BA) regulation requirements in order to improve BAs control performance. The proposed probabilistic approach takes into account multiple uncertain and time dependent factors affecting the system balance (e.g., wind and solar generation, electrical loads). A set of methods to predict the regulation has been developed: (1) probabilistic analysis of regulation requirements with respect to time of the day, (2) linear or nonlinear regression models linking regulation requirements to different influencing factors, (3) adopting time series forecasting techniques (e.g., autoregressive integrated moving average (ARIMA)) to account for temporal continuity and auto-correlation in the regulation requirements data. Proposed methodology has been tested and validated using actual California Independent System Operator (ISO) data.

Solar Generation, Wind Generation, Area Control Er↗
DOE OSTI2026

Flash Heating of a Hollow Rod to Determine Thermal Properties

In a nuclear reactor, prolonged exposure to high neutron flux levels can change material properties of reactor components. Due to the danger posed to personnel by highly radioactive components such as spent fuel, noncontact methods must be used to determine these properties. The objective of the current research is to test the viability of using a flash-heating method to determine the thermal diffusivity of a silicon carbide (SiC) tube, which could be used as cladding for a nuclear fuel rod. The flash-heating experiment was conducted at Oak Ridge National Laboratory and temperature results were analyzed as a two-Dimensional cross section which assumed axial symmetry. The model was fitted iteratively to the temperature measurements using nonlinear regression which required values for thermal diffusivity, Biot number and a term to determine the magnitude of the heat absorbed by the flash. The method could be expanded in the future to estimate the thermal diffusivity of spent nuclear fuel rods in an effort to increase the certainty of new reactor designs using precise thermal properties throughout the core life. A statistical analysis of the results of this work is provided as part of the analysis.

McMasters, Robert L [Virginia Military Institute, ↗
DOE OSTI2022

AN ESTIMATE OF SPENT NUCLEAR FUEL MECHANICAL LOADS IN THE GENERAL 30 CM PACKAGE DROP SCENARIO

The US Department of Energy Spent Fuel and Waste Science and Technology (SFWST) program is performing research to determine the mechanical loading conditions applied to spent nuclear fuel (SNF) during normal conditions of transport to inform mechanical tests of SNF and close an important knowledge gap related to the practical disposition of SNF in the US. Researchers at Pacific Northwest National Laboratory (PNNL) have completed an extensive finite element study to characterize and estimate the potential mechanical loads on SNF during a hypothetical 30 cm drop of an SNF transportation package. This modeling study is validated with test data collected by the SFWST program during a physical test campaign that included one-third scale package drop tests and full-scale single fuel assembly drop tests. The test campaign was led by Sandia National Laboratories (SNL) and included international collaboration with Equipos Nucleares S.A, S.M.E (ENSA) and Bundesanstalt für Materialforschung und -prüfung (BAM). The key contribution of the modeling study is to go beyond the limitations of the limited number of physical tests to estimate the impact response to variations in impact angle, initial gap conditions, fuel assembly design, burnup and other parameters that affect the mechanical loads. The methodology of this study included a classic parametric study to calculate the impact response of highly detailed fuel assemblies over many combinations of parameters. Models of a 17x17 pressurized water reactor fuel assembly and a generic 10x10 boiling water reactor fuel assembly were both used in this study to cover the major fuel assembly types in the US inventory. Over 2,000 impact responses were calculated. The results of the parametric study were evaluated using traditional methods and basic statistics. The results were also used to construct a damage model using multiple nonlinear regression techniques to predict the mechanical loads over the full range of all input parameters. The damage model was found to work very well for all impact angle cases where the cask came to rest on its side. It was concluded that end drop cases where the cask remained vertical (instead of tipping over onto its side) were not sufficiently characterized by the current set of parametric study cases to include in the damage model, but it was not a priority to fully investigate that range because the highest mechanical loads were observed in the broader range of side impact cases. This modeling work provides sufficient insight into the mechanical loads on SNF during a hypothetical 30 cm package drop that, when considered along with the physical test data collected by the SNL-led team, the SFSWT program can consider the knowledge gap closed.

Klymyshyn, Nicholas A.↗
DOE OSTI2015

SU-F-BRD-13: A Phenomenological Relative Biological Effectiveness (RBE) Model for Proton Therapy Based On All Published in Vitro Cell Survival Data

Purpose: Proton therapy treatments are currently planned and delivered using the assumption that the proton relative biological effectiveness (RBE) relative to photons is 1.1. This assumption ignores strong experimental evidence that suggests the RBE varies. A recent review study (Paganetti 2014), collected over 72 experimental reports on proton RBE, providing a comprehensive dataset for predicting proton RBE. Using this dataset we develop a model to predict RBE based on dose, linear energy transfer (LET) and the tissue specific parameter (α/β)x. Methods: The relationship of the RBE on dose, dose average LET (LETd) and (α/β)x was explored using 290 experimental data points. The RBE was calculated for each experimental point for doses ranging from 1 to 10 Gy. An RBE model, based on the linear-quadratic model, was derived using a nonlinear regression fit to the experimental data. Results: The proposed RBE model predicts that the RBE increases with LET and decreases with (α/β)x. The model additionally predicts a decrease in RBE with increasing dose for low (α/β)x values (< 3 Gy). However the model predicts an increase in RBE with increasing dose for large (α/β)x (> 4 Gy). Previous phenomenological models were based on a small subset of experimental data and deviate significantly from our findings. Conclusion: The proposed RBE model is derived using the most comprehensive collection of proton RBE experimental data to date. The model agrees with previous theoretical predictions on the relationship between RBE, LETd and (α/β)x and also makes predictions on the relationship between RBE and dose. The proposed model shows a relationship between α and LETd but no significant relationship between β and LETd. Previously published phenomenological models based on a limited data set might have to be revised.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2021

Progressive transfer learning for low-frequency data prediction in full waveform inversion

To effectively overcome the cycle-skipping issue in full waveform inversion (FWI), we developed a deep neural network (DNN) approach to predict the absent low-frequency components by exploiting the hidden physical relation connecting the low- and the high-frequency data. To efficiently solve this challenging nonlinear regression problem, two novel strategies were proposed to design the DNN architecture and to optimize the learning process: (1) dual data feed structure; (2) progressive transfer learning. With the dual data feed structure, not only the high-frequency data, but also the corresponding beat tone data are fed into the DNN to relieve the burden of feature extraction. The second strategy, progressive transfer learning, enables us to train the DNN using a single evolving training dataset. Within the framework of the progressive transfer learning, the training dataset continuously evolves in an iterative manner by gradually retrieving the subsurface information through the physics-based inversion module, progressively enhancing the prediction accuracy of the DNN and propelling the inversion process out of the local minima. Here, the synthetic numerical experiments suggest that, without any a priori geological information, the low-frequency data predicted by the progressive transfer learning are sufficiently accurate for an FWI engine to produce reliable subsurface velocity models free of cycle-skipping artifacts.

02 PETROLEUM↗