Search NASA⌕ Search

SEARCH · Search NASA

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 217 records · Page 12

The magnetic anomaly of the Ivreazone

A magnetic field survey was made in the Ivreazone in 1969/70. The results were: significant anomaly of the vertical intensity is found. It follows the basic main part of the Ivrea-Verbano zone and continues to the south. The width of the anomaly is about 10 km, the maximum measures about +800 gamma. The model interpretation shows that possibly the anomaly belongs to an amphibolitic body, which in connection with the Ivrea-body was found by deep seismic sounding. Therefore, the magnetic anomaly provides further evidence for the conception that the Ivrea-body has to be regarded as a chip of earthmantle material pushed upward by tectonic processes.

Vertical Intensity of Magnetic Field↗

Integration of Full Tensor Gravity and Z-Axis Tipper Electromagnetic Passive Low Frequency EM Instruments for Simultaneous Data Acquisition - Final Technical Report

Ground gravity is a common and useful tool for geothermal exploration. Gravity surveys map density changes in the subsurface that may be caused by tectonic deformation such as faulting, fracturing, plutonism, volcanism, hydrothermal alteration, etc. Full Tensor Gravity Gradient (FTG) data has been used for over a decade in both petroleum and mining exploration to map changes in density associated with geologic structure. Measuring the gravity gradient, rather than the gravity field, provides significantly higher resolution data. Modeling studies have shown FTG data to be a viable tool for geothermal exploration, but no FTG data had been acquired for geothermal applications to date. Electromagnetic methods have been used for geothermal exploration for some time. The Z-Axis Tipper Electromagnetic (ZTEM) was a newer technology that had found success in mapping deep conductivity changes for mining applications. ZTEM had also been used in limited tests for geothermal exploration. This newer technology provided the ability to cost effectively map large areas whilst detailing the electrical properties of the geological structures at depths. The ZTEM is passive and it uses naturally occurring audio frequency magnetic (AFMAG) signals as the electromagnetic triggering source. These geophysical methods were to be tested over a known geothermal site to determine whether or not the data provided the information required for accurately interpreting the subsurface geologic structure associated with a geothermal deposit. After successful acquisition and analysis of the known source area, an additional survey of a “greenfield” area was to be completed. The final step was to develop a combined interpretation model and determine if the combination produced a higher confident geophysical model compared to models developed using each of the technologies individually.

47 OTHER INSTRUMENTATION↗

Adjustments to the law of the wall above an Amazon forest explained by a spectral link

Modification to the law of the wall represented by a dimensionless correction function ϕ RSL (z/h) is derived using atmospheric turbulence measurements collected at two sites in the Amazon in near-neutral stratification, where z is the distance from the forest floor and h is the mean canopy height. The sites are the Amazon Tall Tower Observatory for z/h∈ [1,2.3] and the Green Ocean Amazon (GoAmazon) site for z/h∈ [1,1.4]. Here, a link between the vertical velocity spectrum E ww (k) (k is the longitudinal wavenumber) and ϕ RSL is then established using a co-spectral budget (CSB) model interpreted by the moving-equilibrium hypothesis. The key finding is that ϕ RSL is determined by the ratio of two turbulent viscosities and is given as ν t,BL /ν t,RSL, where ν t,RSL = (1/A)∫$^{∞}_{0}$ τ(k)E ww (k)dk, ν t,BL = k v (z−d)u * , τ(k) is a scale-dependent decorrelation time scale between velocity components, A = C R /(1−C I ) = 4.5 is predicted from the Rotta constant C R = 1.8, and the isotropization of production constant C I = 3/5 given by rapid distortion theory, k v is the von Kármán constant, u * is the friction velocity at the canopy top, and d is the zero-plane displacement. Because the transfer of energy across scales is conserved in E ww (k) and is determined by the turbulent kinetic energy dissipation rate (ε), the CSB model also predicts that ϕ RSL scales with L BL /L d , where L BL is the length scale of attached eddies to z = d, and L d = u$_{*}^{3}$/ε is a macro-scale dissipation length.

54 ENVIRONMENTAL SCIENCES↗

Surrogate model evaluation and building energy benchmarking for commercial buildings

Building energy consumption benchmarking involves challenges associated with various energy patterns for different building types; heating, ventilating, and air-conditioning (HVAC) system types; and climates. Given significant variation in energy use patterns, accurate prediction of long-term energy use using surrogate models remains challenging. Multiple linear regression (MLR) is commonly used for building energy benchmarking because of its simple structure; however, it lacks accuracy compared to other black-box models. Although many studies have compared surrogate models and offer guidance on model selection based on metrics, they do not provide detailed analysis on improving the surrogate model accuracy. In this paper, we implement a surrogate model using polynomial ridge regression (i.e., MLR with interaction terms combined with ridge regularization) for small office and retail strip mall buildings across six HVAC system types and all climate zones, for electricity and natural gas in baseline and proposed scenarios. A simulation workflow is developed using OpenStudio TM /EnergyPlus TM to generate simulation data using measures over a wide range of efficiency inputs. Enhancements based on statistical insights are used for improving the model accuracy using filters, input transformations, and change points. Surrogate models achieved average coefficient of variation of the root mean squared error (CVRMSE) values of 2.17, 1.06, 2.05, and 3.26 for proposed electricity, proposed natural gas, baseline electricity, and baseline natural gas, respectively, with enhancements reducing CVRMSE by an average of 14.9% across all combinations. We provide model interpretation via Shapley additive explanations to determine which input variables most influence energy consumption and provide supportive arguments for enhancements.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Level Density Parameter: A Tool to Study the Particle Spectra

The compound nucleus {sup 76}Kr* is formed in the heavy-ion fusion reactions by an asymmetric entrance channel {sup 12}C+{sup 64}Zn and the symmetric entrance channel {sup 31}P+{sup 45}Sc at the excitation energy of 75 MeV and angular momentum of 39 eta. Neutron energy spectra of the asymmetric system ({sup 12}C+{sup 64}Zn) at different angles are well described by the statistical model predictions using the normal value of the level density parameter a = A/8 MeV{sup -1}. However, in the case of the symmetric system ({sup 31}P+{sup 45}Sc), the statistical model interpretation of the data requires the change in the value of a = A/10 MeV{sup -1}. The delayed evolution of the compound system in case of the symmetric {sup 31}P+{sup 45}Sc system may lead to the formation of a temperature equilibrated dinuclear complex, which may be responsible for the neutron emission at higher temperature, while the protons and alpha particles are evap orated after neutron emission when the system is sufficiently cooled down and the higher lambda-values do not contribute in the formation of the compound nucleus for the symmetric entrance channel in case of charged particle emission.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Special Issue: Geostatistics and Machine Learning

Abstract Recent years have seen a steady growth in the number of papers that apply machine learning methods to problems in the earth sciences. Although they have different origins, machine learning and geostatistics share concepts and methods. For example, the kriging formalism can be cast in the machine learning framework of Gaussian process regression. Machine learning, with its focus on algorithms and ability to seek, identify, and exploit hidden structures in big data sets, is providing new tools for exploration and prediction in the earth sciences. Geostatistics, on the other hand, offers interpretable models of spatial (and spatiotemporal) dependence. This special issue on Geostatistics and Machine Learning aims to investigate applications of machine learning methods as well as hybrid approaches combining machine learning and geostatistics which advance our understanding and predictive ability of spatial processes.

58 GEOSCIENCES↗

Shock dynamics of phase diagrams

A thermodynamic phase transition denotes a drastic change of state of a physical system due to a continuous change of thermodynamic variables, as for instance pressure and temperature. The classical van der Waals equation of state is the simplest model that predicts the occurrence of a critical point associated with the gas–liquid phase transition. Nevertheless, below the critical temperature theoretical predictions of the van der Waals theory significantly depart from the observed physical behaviour. We develop a novel approach to classical thermodynamics based on the solution of Maxwell relations for a generalised family of nonlocal entropy functions. This theory provides an exact mathematical description of discontinuities of the order parameter within the phase transition region, it explains the universal form of the equations of state and the occurrence of triple points in terms of the dynamics of nonlinear shock wave fronts. -- Highlights: •A new generalisation of van der Waals equation of state. •Description of phase transitions in terms of shock dynamics of state curves. •Proof of the universality of equations of state for a general class of models. •Interpretation of triple points as confluence of classical shock waves. •Correspondence table between thermodynamics and nonlinear conservation laws.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Nano-structured platinum group metal-free catalysts and their integration in fuel cell electrode architectures

The novel platinum group metal-free (PGM-free) catalyst for the oxygen reduction reaction (ORR) is synthesized by a modified sacrificial support method (SSM). The catalyst chemical/surface composition is studied by X-ray photoelectron spectroscopy, and the morphology of the material is observed using both HR-SEM and HR-TEM, demonstratingthe open-frame, self-supported catalysts. This new catalyst’s electrochemical performance is evaluated by polarization curves and has behaviour comparable to the state-of-the-art PGM-free catalysts. Meso-structure imaging shows pores on the order of 100 nm, the mean size of the individual silica particles in the sacrificial support. For the first time, PGM-free catalyst layer (CL) morphology in a membrane electrode assembly (MEA) is studied in detail by combined nano- and micro X-ray computed tomography (CT) and interpretational modelling. The highly inhomogeneous, high-tortuosity, through-thickness structure of the CL is observed with micro-CT. The nano-CT method for these thick PGM-free electrodes is not sufficient to capture the full through-thickness morphology of these electrodes. Water retention curves suggest water pooling at the MEA components‘ interfaces and significant dependence of capilary pressure and saturation on through-thickness location. Lastly, this study is the first of its kind to identify morphology-dependent transport losses in the thick PGM-free electrodes using scale-bridging between meso -, micro-, and macro.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Effect of progressively increasing lithium conditioning on edge transport and stability in high triangularity NSTX H-modes

A sequence of H-mode discharges with increasing levels of pre-discharge lithium evaporation (‘dose’) was conducted in high triangularity and elongation boundary shape in NSTX. Energy confinement increased, and recycling decreased with increasing lithium dose, similar to a previous lithium dose scan in medium triangularity and elongation plasmas. Data-constrained SOLPS interpretive modeling quantified the edge transport change: the electron particle diffusivity decreased by 10-30x. The electron thermal diffusivity decreased by 4x just inside the top of the pedestal, but increased by up to 5x very near the separatrix. These results provide a baseline expectation for lithium benefits in NSTX-U, which is optimized for a boundary shape similar to the one in this experiment.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

A Transdisciplinary Review of Deep Learning Research and Its Relevance for Water Resources Scientists

Deep learning (DL), a new generation of artificial neural network research, has transformed industries, daily lives, and various scientific disciplines in recent years. DL represents significant progress in the ability of neural networks to automatically engineer problem-relevant features and capture highly complex data distributions. I argue that DL can help address several major new and old challenges facing research in water sciences such as interdisciplinarity, data discoverability, hydrologic scaling, equifinality, and needs for parameter regionalization. This review paper is intended to provide water resources scientists and hydrologists in particular with a simple technical overview, transdisciplinary progress update, and a source of inspiration about the relevance of DL to water. The review reveals that various physical and geoscientific disciplines have utilized DL to address data challenges, improve efficiency, and gain scientific insights. DL is especially suited for information extraction from image-like data and sequential data. Techniques and experiences presented in other disciplines are of high relevance to water research. Meanwhile, less noticed is that DL may also serve as a scientific exploratory tool. A new area termed AI neuroscience , where scientists interpret the decision process of deep networks and derive insights, has been born. This budding subdiscipline has demonstrated methods including correlation-based analysis, inversion of network-extracted features, reduced-order approximations by interpretable models, and attribution of network decisions to inputs. Moreover, DL can also use data to condition neurons that mimic problem-specific fundamental organizing units, thus revealing emergent behaviors of these units. Vast opportunities exist for DL to propel advances in water sciences.

54 ENVIRONMENTAL SCIENCES↗

Rapid design of top-performing metal-organic frameworks with qualitative representations of building blocks

Abstract Data-driven materials design often encounters challenges where systems possess qualitative (categorical) information. Specifically, representing Metal-organic frameworks (MOFs) through different building blocks poses a challenge for designers to incorporate qualitative information into design optimization, and leads to a combinatorial challenge, with large number of MOFs that could be explored. In this work, we integrated Latent Variable Gaussian Process (LVGP) and Multi-Objective Batch-Bayesian Optimization (MOBBO) to identify top-performing MOFs adaptively, autonomously, and efficiently. We showcased that our method (i) requires no specific physical descriptors and only uses building blocks that construct the MOFs for global optimization through qualitative representations, (ii) is application and property independent, and (iii) provides an interpretable model of building blocks with physical justification. By searching only ~1% of the design space, LVGP-MOBBO identified all MOFs on the Pareto front and 97% of the 50 top-performing designs for the CO 2 working capacity and CO 2 /N 2 selectivity properties.

36 MATERIALS SCIENCE↗

Inverse design of photonic surfaces via multi fidelity ensemble framework and femtosecond laser processing

We demonstrate a multi-fidelity (MF) machine learning ensemble framework for the inverse design of photonic surfaces, trained on a dataset of 11,759 samples that we fabricate using high throughput femtosecond laser processing. The MF ensemble combines an initial low fidelity model for generating design solutions, with a high fidelity model that refines these solutions through local optimization. The combined MF ensemble can generate multiple disparate sets of laser-processing parameters that can each produce the same target input spectral emissivity with high accuracy (root mean squared errors < 2%). SHapley Additive exPlanations analysis shows transparent model interpretability of the complex relationship between laser parameters and spectral emissivity. Finally, the MF ensemble is experimentally validated by fabricating and evaluating photonic surface designs that it generates for improved efficiency energy harvesting devices. Our approach provides a powerful tool for advancing the inverse design of photonic surfaces in energy harvesting applications.

97 MATHEMATICS AND COMPUTING↗

Design and testing of ultrasound probe adapters for a robotic imaging platform

Medical imaging-based triage is a critical tool for emergency medicine in both civilian and military settings. Ultrasound imaging can be used to rapidly identify free fluid in abdominal and thoracic cavities which could necessitate immediate surgical intervention. However, proper ultrasound image capture requires a skilled ultrasonography technician who is likely unavailable at the point of injury where resources are limited. Instead, robotics and computer vision technology can simplify image acquisition. As a first step towards this larger goal, here, we focus on the development of prototypes for ultrasound probe securement using a robotics platform. The ability of four probe adapter technologies to precisely capture images at anatomical locations, repeatedly, and with different ultrasound transducer types were evaluated across more than five scoring criteria. Testing demonstrated two of the adapters outperformed the traditional robot gripper and manual image capture, with a compact, rotating design compatible with wireless imaging technology being most suitable for use at the point of injury. Next steps will integrate the robotic platform with computer vision and deep learning image interpretation models to automate image capture and diagnosis. This will lower the skill threshold needed for medical imaging-based triage, enabling this procedure to be available at or near the point of injury.

47 OTHER INSTRUMENTATION↗

Observation of D 2 molecule line emission after massive D 2 injection into runaway electron plateaus in DIII-D

Molecular deuterium line emission is observed in both the visible and ultraviolet (UV) wavelength ranges after massive (> 100 Torr-L) injection of D 2 gas into post-disruption runaway electron (RE) dominated plasmas in the DIII-D tokamak. D 2 UV line emission is found to be the dominant source of radiated power, surpassing D Lyα. Interpretive modeling with a collisional-radiative model (CRM) indicates that D 2 radiation surpasses D radiation because Lyα is strongly trapped, while D 2 UV lines are mostly untrapped. The CRM also indicates that the D 2 line emission is completely dominated by RE impact (rather than thermal electron impact), so the D 2 line emission can serve as a good diagnostic for the spatial localization of REs. Analysis of D 2 visible lines indicates that the D 2 molecules in the plasma are thermally equilibrated with the background plasma, with vibrational, rotational, and kinetic temperatures all near 0.3 eV. D 2 spectroscopy therefore serves as a convenient diagnostic of background plasma temperature. As a result, measurement of D 2 radiated power also appears to serve as a useful diagnostic for constraining neutral transport modeling.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

VARIATIONS IN THE MASS FUNCTIONS OF CLUSTERED AND ISOLATED YOUNG STELLAR OBJECTS

We analyze high-quality, complete stellar catalogs for four young (roughly 1 Myr) and nearby (within {approx}300 pc) star-forming regions: Taurus, Lupus3, ChaI, and IC348, which have been previously shown to have stellar groups whose properties are similar to those of larger clusters such as the Orion Nebula Cluster (ONC). We find that stars at higher stellar surface densities within a region or belonging to groups tend to have a relative excess of more massive stars, over a wide range of masses. We find statistically significant evidence for this result in Taurus and IC348 as well as in the ONC. These differences correspond to having typically a {approx}10%-20% higher mean mass in the more clustered environment. Stars in ChaI show no evidence for a trend with either surface density or grouped status, and there are too few stars in Lupus3 to make any definitive interpretation. Models of clustered star formation do not typically extend to sufficiently low masses or small group sizes in order for their predictions to be tested, but our results suggest that this regime is important to consider.

79 ASTRONOMY AND ASTROPHYSICS↗

Determining pseudoscalar meson photoproduction amplitudes from complete experiments

In preparation for a new generation of complete experiments with the goal of performing a high precision extraction of pseudoscalar meson photo-production amplitudes, we present expressions that allow the direct numerical calculation of matrix elements with arbitrary spin projections from Chew-Goldberger-Low-Nambu (CGLN) amplitudes. We use this numerical tool to verify the most general analytic form of the cross section, dependent upon the three polarization vectors of the beam, target and recoil baryon, including all single, double and triple-polarization terms involving 16 spin-dependent observables. Analytic expressions that determine the recoil baryon polarization are presented, together with examples of their potential use with quasi-4? detectors to deduce observables. We assemble the analytic equations relating the 16 experimental observables and the CGLN amplitudes and use our independent method of numerical evaluation to resolve sign differences that exist in the literature.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Characterizing W sources in the all-W wall, all-RF WEST tokamak environment

In this work, experimental data, together with interpretive modeling tools, are examined to study trends in the tungsten (W) source in the all-W environment of the WEST tokamak, both from the divertor and from the main chamber. In particular, a poloidal limiter protecting an ion cyclotron resonance heating (ICRH) antenna is used as proxy for main chamber sourcing. The key study is carried out by stepping up lower hybrid current drive (LHCD) power, as the only auxiliary power source. Limiter and divertor W sources exhibit a qualitatively similar proportionality to the total power crossing the separatrix, P SEP , although the main chamber source remains substantially lower than the divertor source, for the range of P SEP accessible in the experiments. Intepretive modeling of the limiter source is carried out with a particle-in-cell (PIC) sheath model coupled to a surface sputtering model. Oxygen is used as a proxy for all light impurity species allowing for characterization of the critical W erosion regions. To get a good quantitative match to the data, it is necessary to assume that the oxygen arrives at the surface mostly at high ionization stages (4+ and above). A separate simulation with SOLEDGE-EIRENE, constrained to measured upstream scrape-off-layer plasma profiles, gives oxygen fractional abundances that are compatible with the PIC simulation result. This is understood to arise from transport processes that dominate over recombination. Substituting the LHCD by ICRH, in an equivalent experiment, the local W source exhibits a 3× enhancement. This can be matched by the simulation, by assuming local RF electric field rectification, based on ~100 eV peak-to-peak, near-antennna electric field. This work has highlighted the particular importance of understanding the ion charge state balance of light impurities as these are most likely the dominant sputtering species in fusion devices with high-Z walls.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Shedding Light on the Dark Sector with Direct WIMP Production

A Weakly Interacting Massive Particle (WIMP) provides an attractive dark matter candidate, and should be within reach of the next generation of high-energy colliders. We consider the process of direct WIMP pair-production, accompanied by an initial-state radiation photon, in electron-positron collisions at the proposed International Linear Collider (ILC). We present a parametrization of the differential cross section for this process which conveniently separates the model-independent information provided by cosmology from the model-dependent inputs from particle physics. As an application, we consider two simple models, one supersymmetric, and another of the 'universal extra dimensions'' (UED) type. The discovery reach of the ILC and the expected precision of parameter measurements are studied in each model. For each of the two examples, we perform a shape-discrimination analysis of the photon energy spectrum and show that in a large part of the relevant parameter space it allows to rule out the alternative model interpretation.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗