Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “model interpretability”

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 505 records · Page 28
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Electronic Structure Theory and Novel Materials

This grant supported research on electronic structure and materials theory, with focus on three main issues: (i) novel techniques to deal with correlation in the electronic ground-state, (ii) topological materials, (iii) the phase diagram of lattice spin models. Regarding (i), we applied to the homogeneous electron liquid an approach that we previously developed in the context of molecular systems. In this scheme the electronic occupation probabilities and the natural spin orbitals are used to construct an approximate two-body density matrix for the electronic ground-state. Regarding (ii) we used standard electronic structure methods based on density functional theory to model topological materials and interpret experimental observations. Finally, regarding (iii) we further developed a numerical approach to compute the renormalized couplings within real space renormalization group theory in the context of lattice spin models. The main findings were the following. (i) We found that with our approximate two-body density matrix, which works well for small molecules, is not sufficiently accurate for condensed phase systems. Missing a systematic way of improving on the adopted approximations, we decided not to pursue this approach. (ii) We performed two studies. In one, we investigated the influence of Te defects on the topological properties of a WTe2 monolayer, finding that while Te vacancies, even in modest concentration, destroy the topological character, Te adatoms do not, consistent with a recent experiment. In another study, we predicted Weyl semimetal character and strong anomalous Hall effect in the Heusler compensated ferrimagnet Ti2MnAl. (iii) We developed a new Monte Carlo method to do real space renormalization group calculations for lattice spin models. We subsequently extended the scheme to deal with lattice spin models in presence of quenched disorder, finding that the approach can distinguish systems with finite and strong disorder. In the finite disorder case, the method allows one to find with good approximation the critical coupling distribution and the critical exponents.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2020

Characterization of Spall in Hard Rock from Observations and Simulations of the Source Physics Experiment Phase I

In this work, spall signals from the Source Physics Experiments are presented, analyzed, and modeled for insight to the explosion source. The observed signal is similar in nature to nearby historical nuclear explosions, and the surface force-time history or velocity can be interpreted with the same model. We use the models for peak spall velocity, spalled mass, and spall depth and radius derived from historical nuclear explosions to parameterize the physical force-time history model from Stump (1985) and show that this parameterized model can be used for spall prediction. The spall signal is also investigated with a numerical continuum model that incorporates gravity. Peak velocity and dwell time are well predicted, and the multiple slap-down phases are captured if one includes a weak near-surface layer similar to the geologic observation.

58 GEOSCIENCES↗
DOE OSTI2020

Updates to Relevance Vector Machine: Multiclass Classification, Variable Selection, and Proof-of-Concept Application to Safeguards Fresh Fuel Verification using List-Mode Neutron Collar Data

To expand the capabilities of safeguards authorities to verify the integrity of fresh fuel assemblies, Oak Ridge National Laboratory has retrofit the existing electronics of the JCC-71 uranium neutron coincidence collar, which contains 18 3 He neutron detectors and an external 241 AmLi(α, n) neutron interrogation source arranged to surround a fresh nuclear fuel assembly. The new electronics system allows analysts to record list-mode neutron multiplicity data in addition to the singles and doubles rates that are currently measured. Based on previous proof-of-concept research, analysis of these new data will identify off-normal fuel configurations in an assembly and characterize or localize the specific partial fuel defects. The purpose of this report it to document the analysis algorithm development and then to demonstrate its capability for the safeguards verification of fresh fuel assemblies using list mode neutron collar data. To analyze the complex list-mode data collected with the upgraded uranium neutron collar, multivariate classification algorithms are being developed using a novel classification method, the relevance vector machine. This approach may be applied to multiclass problems to estimate the probability that test data belongs to one of many possible classes of data. In addition, our method identifies the most useful variables/channels for making predictions, which illuminates the basis for the model’s predictions, and this interpretability is largely unique among data analytics methods. Variable selection occurs during model training and parameter tuning and does not need any external hyperparameter tuning routines. Finally, we apply the modified relevance vector machine to a simulated dataset of list-mode neutron collar data generated with the radiation transport code MCNP. The method can correctly identify off-normal fuel configurations, categorize the data according to four fuel defect scenarios, and rank the channels in the data according to prediction utility. For nuclear safeguards applications, it is concluded that this method has the potential to increase the sensitivity and reliability to detect missing fuel rods from a standard 17 x 17 Pressurized Water Reactor (PWR) fresh fuel assembly. Within this analysis, “off-normal” (i.e., missing fuel rods) were correctly classified in 17 simulated test scenarios with one quarter (25%) of the fresh fuel rods missing using a training data set of 58 simulated measurements.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2022

Thermal model for time-domain thermoreflectance experiments in a laser-flash geometry

Time-domain thermoreflectance (TDTR) is a well-established pump–probe method for measuring thermal conductivity and interface conductance of multilayers. Interpreting signals in a TDTR experiment requires a thermal model. In standard front/front TDTR experiments, both pump and probe beams typically irradiate the surface of a multilayer. As a result, existing thermal models for interpreting thermoreflectance experiments assume that the pump and probe beams both interact with the surface layer. Here, we present a frequency-domain solution to the heat-diffusion equation of a multilayer in response to nonhomogeneous laser heating. This model allows analysis of experiments where the pump and probe beams irradiate opposite sides of a multilayer. We call such a geometry a front/back experiment to differentiate such experiments from standard TDTR experiments. As an example, we consider a 60nm amorphous Si film. We consider how signals differ in a front/front vs front/back geometry and compare thermal model predictions to experimental data.

Peng, Wanyue↗
DOE OSTI2020

Developing a common approach for classifying building stock energy models

Buildings contribute 40% of global greenhouse gas emissions; therefore, strategies that can substantially reduce emissions from the building stock are key components of broader efforts to mitigate climate change and achieve sustainable development goals. Models that represent the energy use of the building stock at scale under various scenarios of technology deployment have become essential tools for the development and assessment of such strategies. Within the past decade, the capabilities of building stock energy models have improved considerably, while model transferability and sharing has increased. Given these advancements, a new scheme for classifying building stock energy models is needed to facilitate communication of modeling approaches and the handling of important model dimensions. In this article, we present a new building stock energy model classification framework that leverages international modeling expertise from the participants of the International Energy Agency's Annex 70 on Building Energy Epidemiology. Drawing from existing classification studies, here we propose a multi-layer quadrant scheme that classifies modeling techniques by their design (top-down or bottom-up) and degree of transparency (black-box or white-box); hybrid techniques are also addressed. The quadrant scheme is unique from previous classification approaches in its non-hierarchical organization, coverage of and ability to incorporate emerging modeling techniques, and treatment of additional modeling dimensions. The new classification framework will be complemented by a reporting protocol and online registry of existing models as part of ongoing work in Annex 70 to increase the interpretability and utility of building stock energy models for energy policy making.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗
DOE OSTI2023

Complementing a continuum thermodynamic approach to constitutive modeling with symbolic regression

Here, an interpretable machine learning method, physics-informed genetic programming-based symbolic regression (P-GPSR), is integrated into a continuum thermodynamic approach to developing constitutive models. The proposed strategy for combining a thermodynamic analysis with P-GPSR is demonstrated by generating a yield function for an idealized material with voids, i.e., the Gurson yield function. First, a thermodynamic-based analysis is used to derive model requirements that are exploited in a custom P-GPSR implementation as fitness criteria or are strongly enforced in the solution. The P-GPSR implementation improved accuracy, generalizability, and training time compared to the same GPSR code without physics-informed fitness criteria. The yield function generated through the P-GPSR framework is in the form of a composite function that describes a class of materials and is characteristically more interpretable than GPSR-derived equations. The physical significance of the input functions learned by P-GPSR within the composite function is acquired from the thermodynamic analysis. Fundamental explanations of why the implemented P-GPSR capabilities improve results over a conventional GPSR algorithm are provided.

42 ENGINEERING↗
DOE OSTI2022

A Universal Machine Learning Model for Elemental Grain Boundary Energies

The grain boundary (GB) energy has a profound influence on the grain growth and properties of polycrystalline metals. Here, we show that the energy of a GB, normalized by the bulk cohesive energy, can be described purely by four geometric features. By machine learning on a large computed database of 361 small Σ (Σ<10) GBs of more than 50 metals, we develop a model that can predict the grain boundary energies to within a mean absolute error of 0.13 J m –2 . More importantly, this universal GB energy model can be extrapolated to the energies of high Σ GBs without loss in accuracy. These results highlight the importance of capturing fundamental scaling physics and domain knowledge in the design of interpretable, extrapolatable machine learning models for materials science.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Fingerprinting Interactions between Proteins and Ligands for Facilitating Machine Learning in Drug Discovery

Molecular recognition is fundamental in biology, underpinning intricate processes through specific protein–ligand interactions. This understanding is pivotal in drug discovery, yet traditional experimental methods face limitations in exploring the vast chemical space. Computational approaches, notably quantitative structure–activity/property relationship analysis, have gained prominence. Molecular fingerprints encode molecular structures and serve as property profiles, which are essential in drug discovery. While two-dimensional (2D) fingerprints are commonly used, three-dimensional (3D) structural interaction fingerprints offer enhanced structural features specific to target proteins. Machine learning models trained on interaction fingerprints enable precise binding prediction. Recent focus has shifted to structure-based predictive modeling, with machine-learning scoring functions excelling due to feature engineering guided by key interactions. Notably, 3D interaction fingerprints are gaining ground due to their robustness. Various structural interaction fingerprints have been developed and used in drug discovery, each with unique capabilities. This review recapitulates the developed structural interaction fingerprints and provides two case studies to illustrate the power of interaction fingerprint-driven machine learning. The first elucidates structure–activity relationships in β2 adrenoceptor ligands, demonstrating the ability to differentiate agonists and antagonists. The second employs a retrosynthesis-based pre-trained molecular representation to predict protein–ligand dissociation rates, offering insights into binding kinetics. Despite remarkable progress, challenges persist in interpreting complex machine learning models built on 3D fingerprints, emphasizing the need for strategies to make predictions interpretable. Binding site plasticity and induced fit effects pose additional complexities. Interaction fingerprints are promising but require continued research to harness their full potential.

3D structural interaction fingerprints↗
DOE OSTI2022

Co-optimized machine-learned manifold models for large eddy simulation of turbulent combustion

Many modeling approaches in large eddy simulation (LES) of turbulent combustion employ a projection of the thermochemical state onto a low-dimensional manifold within state space to reduce the number of transported variables and hence computational cost. Flamelet-generated manifolds (FGM) is an example of a well-established, physics-based approach, but increasingly, principal component analysis (PCA) is being used as a data-driven method for generating manifold models. For both approaches, the nonlinear relationship between the location on the predefined manifold and the outputs of interest, such as reaction rates, can be tabulated or encoded in a neural network. This work proposes a new approach for manifold modeling that extends these existing approaches. A modified neural network structure simultaneously encodes the definition of the manifold variables, the nonlinear mapping, and the subfilter closure for LES. This allows all three of these aspects of the model to be co-optimized, generating a model from any source of combustion thermochemical state data. The manifold parameterizing variables are constrained to be linear combinations of species, as in FGM and PCA-based models, to aid in interpretability and implementation. For LES, subfilter variances of the manifold variables are also included as inputs. Two types of a priori analysis are performed to evaluate the new approach. In the first, the model is trained on data from one-dimensional premixed flames. In this case, the approach recovers the behavior of flamelet-based manifold approaches, and in fact slightly improves performance by identifying an optimized progress variable. The approach is also applied to data from direct numerical simulations of spherical ignition kernels in isotropic turbulence. For any specified manifold dimensionality, the new approach provides substantially lower prediction errors than a PCA-based model developed from the same data set. Additionally, the LES formulation of the new approach can provide accurate predictions for filtered reaction rates across a variety of filter widths.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Search for resonant pair production of Higgs bosons in the $\textrm{b}\overline{\textrm{b}}\textrm{b}\overline{\textrm{b}}$ final state using large-area jets in proton-proton collisions at $\sqrt{s}$ = 13 TeV

A search is presented for the resonant production of a pair of standard model-like Higgs bosons using data from proton-proton collisions at a centre-of-mass energy of 13 TeV, collected by the CMS experiment at the CERN LHC in 2016–2018, corresponding to an integrated luminosity of 138 fb −1 . The final state consists of two b quark-antiquark pairs. The search is conducted in the region of phase space where at least one of the pairs is highly Lorentz-boosted and is reconstructed as a single large-area jet. The other pair may be either similarly merged or resolved, the latter reconstructed using two b-tagged jets. The data are found to be consistent with standard model processes and are interpreted as 95% confidence level upper limits on the product of the cross sections and the branching fractions of the spin-0 radion and the spin-2 bulk graviton that arise in warped extradimensional models. The limits set are in the range 9.74–0.29 fb and 4.94–0.19 fb for a narrow radion and a graviton, respectively, with masses between 1 and 3 TeV. For a radion and for a bulk graviton with widths 10% of their masses, the limits are in the range 12.5–0.35 fb and 8.23–0.23 fb, respectively, for the same masses. These limits result in the exclusion of a narrow-width graviton with a mass below 1.2 TeV, and of narrow and 10%-width radions with masses below 2.6, and 2.9 TeV, respectively.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2025

White Paper: Scalable Digital Twin Capabilities for Aging and Surveillance of Engineered Systems

This white paper presents a multi-year initiative to develop practical, secure, and scalable digital twin capabilities for engineered systems in aging and surveillance contexts—an approach pioneered at the National Nuclear Security Administration (NNSA) Lawrence Livermore National Laboratory (LLNL) that maps directly onto the needs and ambitions of the Navy for ship- and fleet-level digital twins. LLNL’s work in building part- and process-level digital twins for advanced manufacturing, with a vision to scale up to entire factory floors and, ultimately, enterprise-wide digital twins, offers an adaptable pathway for the Navy as it seeks to modernize lifecycle management, readiness, and predictive maintenance across ships and fleets. For our application, we integrate physics-based modeling with automated data ingestion, processing, and AI-driven calibration, creating hybrid models that are both interpretable and data responsive. We modernized legacy workflows, established centralized data infrastructure, automated experimental pipelines, and demonstrated end-to-end coupling of accelerated aging data with finite element simulations via optimization and surrogate modeling. The result is a generalizable framework that supports part-level digital twins today and lays the groundwork for future system-level twins suitable for Navy applications.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Electronic Spectroscopy and Dissociation Dynamics of Vinyl-Substituted Criegee Intermediates: 2-Butenal Oxide and Comparison with Methyl Vinyl Ketone Oxide and Methacrolein Oxide Isomers

The 2-butenal oxide Criegee intermediate [(CH 3 CH=CH)CHOO], an isomer of the four-carbon unsaturated Criegee intermediates derived from isoprene ozonolysis, is characterized on its first π * ←π electronic transition and by the resultant dissociation dynamics to O ( 1 D) + 2-butenal [(CH 3 CH=CH)CHO] products. The electronic spectrum of 2-butenal oxide under jet-cooled conditions is observed to be broad and unstructured with peak absorption at 373 nm, spanning to half maxima at 320 and 420 nm, and in good accord with the computed vertical excitation energies and absorption spectra obtained for its lowest energy conformers. The distribution of total kinetic energy released to products is ascertained through velocity map imaging of the O ( 1 D) products. About half of the available energy, deduced from the theoretically computed asymptotic energy, is accommodated as internal excitation of the 2-butenal fragment. A reduced impulsive model is introduced to interpret the photodissociation dynamics, which accounts for the geometric changes between 2-butenal oxide and the 2-butenal fragment, and vibrational activation of associated modes in the 2-butenal product. Application of the reduced impulsive model to the photodissociation of isomeric methyl vinyl ketone oxide reveals greater internal activation of the methyl vinyl ketone product arising from methyl internal rotation and rock, which is distinctly different than the dissociation dynamics of 2-butenal oxide or methacrolein oxide.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2020

Search for new phenomena in final states with large jet multiplicities and missing transverse momentum using $\sqrt{s}$ = 13 TeV proton-proton collisions recorded by ATLAS in Run 2 of the LHC

Results of a search for new particles decaying into eight or more jets and moderate missing transverse momentum are presented. The analysis uses 139 fb - 1 of proton-proton collision data at s = 13 TeV collected by the ATLAS experiment at the Large Hadron Collider between 2015 and 2018. The selection rejects events containing isolated electrons or muons, and makes requirements according to the number of b -tagged jets and the scalar sum of masses of large-radius jets. The search extends previous analyses both in using a larger dataset and by employing improved jet and missing transverse momentum reconstruction methods which more cleanly separate signal from background processes. No evidence for physics beyond the Standard Model is found. The results are interpreted in the context of supersymmetry-inspired simplified models, significantly extending the limits on the gluino mass in those models. In particular, limits on the gluino mass are set at 2 TeV when the lightest neutralino is nearly massless in a model assuming a two-step cascade decay via the lightest chargino and second-lightest neutralino.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2022

Two-dimensional parametric design of short pulse laser driven iron opacity experiments

High intensity short pulse lasers are commonly used to create high temperature and high density plasmas. These plasmas are used to study fundamental properties such as the equation of state and opacity. The targets often use small-diameter, thin buried layers of the materials of interest, surrounded by a tamper of low atomic number material such as plastic. Computational modeling is used to design and interpret experiments with short pulse lasers. Most of the modeling to date has assumed one-dimensional plane-parallel geometry. In this paper, the effects of radial gradients in the irradiation of thin planar targets are studied with one- and two-dimensional radiation/hydrodynamic simulations. It is found that the main effect of radial gradients is the averaging of plasma conditions and x-ray emission over the pattern of irradiation. Additionally, differences between one- and two-dimensional simulations arise because the plasma conditions and x-ray emission are, in general, nonlinear functions of the temperature of the plasma and thereby also nonlinear functions of the irradiation intensity. The differences increase with the ratio of the buried layer radius to the laser spot radius. The root mean square difference in the inferred iron L-shell opacity is less than 30% when the ratio is less than 1.0 and about 70% when the ratio is 1.5.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2020

On the Solution of ℓ 0 -Constrained Sparse Inverse Covariance Estimation Problems

The sparse inverse covariance matrix is used to model conditional dependencies between variables in a graphical model to fit a multivariate Gaussian distribution. Estimating the matrix from data are well known to be computationally expensive for large-scale problems. Sparsity is employed to handle noise in the data and to promote interpretability of a learning model. Although the use of a convex ℓ 1 regularizer to encourage sparsity is common practice, the combinatorial ℓ 0 penalty often has more favorable statistical properties. In this paper, we directly constrain sparsity by specifying a maximally allowable number of nonzeros, in other words, by imposing an ℓ 0 constraint. Here, we introduce an efficient approximate Newton algorithm using warm starts for solving the nonconvex ℓ 0 -constrained inverse covariance learning problem. Numerical experiments on standard data sets show that the performance of the proposed algorithm is competitive with state-of-the-art methods.

$\ell_0$-Constrained↗
DOE OSTI2022

Field Sensitivity Analysis of Turbulence Model Parameters for Flow Over a Wing

Reynolds-averaged-Navier-Stokes (RANS) turbulence models are a critical tool in computational-fluid-dynamics simulations of aerodynamic systems, but simulation results can be highly sensitive to RANS-model parameter choices. Sensitivity analysis can be used to quantify these impacts, and the objective of this study is to demonstrate field sensitivity analysis with respect to ten parameters in the 2003 Menter shear-stress-transport (SST) turbulence model. Here, the analysis is demonstrated for an application relevant to wind energy, namely, flow over a NACA 0015 wing at 12 degree angle of attack and a Reynolds number of 1.5 x 10 6 . We quantify sensitivity using Sobol indices and the mean-squared gradient, which are estimated using polynomial chaos and active subspace models, respectively. Our results indicate that there are substantial spatial variations in parameter sensitivities, with different sets of most-sensitive parameters near the wing, as well as in the downstream wake, consistent with the physical interpretations of the turbulence model inputs. We show that, for this particular turbulence model and flow, simultaneous dimension reduction is possible across all quantities of interest, enabling efficient exploration of model outcomes. Ultimately, this analysis provides new insights into turbulence model parameter sensitivities in incompressible flows, and also demonstrates the implementation of field sensitivity analysis for applications relevant to aerodynamics simulations.

17 WIND ENERGY↗