Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Real Data”

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

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

At least 163 records · Page 9

A Method to Test Model Calibration Techniques: Preprint

This paper describes a method for testing model calibration techniques. Calibration is commonly used in conjunction with energy retrofit audit models. An audit is conducted to gather information about the building needed to assemble an input file for a building energy modeling tool. A calibration technique is used to reconcile model predictions with utility data, and then the 'calibrated model' is used to predict energy savings from a variety of retrofit measures and combinations thereof. Current standards and guidelines such as BPI-2400 and ASHRAE-14 set criteria for 'goodness of fit' and assume that if the criteria are met, then the calibration technique is acceptable. While it is logical to use the actual performance data of the building to tune the model, it is not certain that a good fit will result in a model that better predicts post-retrofit energy savings. Therefore, the basic idea here is that the simulation program (intended for use with the calibration technique) is used to generate surrogate utility bill data and retrofit energy savings data against which the calibration technique can be tested. This provides three figures of merit for testing a calibration technique, 1) accuracy of the post-retrofit energy savings prediction, 2) closure on the 'true' input parameter values, and 3) goodness of fit to the utility bill data. The paper will also discuss the pros and cons of using this synthetic surrogate data approach versus trying to use real data sets of actual buildings.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

A Method to Test Model Calibration Techniques

This paper describes a method for testing model calibration techniques. Calibration is commonly used in conjunction with energy retrofit audit models. An audit is conducted to gather information about the building needed to assemble an input file for a building energy modeling tool. A calibration technique is used to reconcile model predictions with utility data, and then the 'calibrated model' is used to predict energy savings from a variety of retrofit measures and combinations thereof. Current standards and guidelines such as BPI-2400 and ASHRAE-14 set criteria for 'goodness of fit' and assume that if the criteria are met, then the calibration technique is acceptable. While it is logical to use the actual performance data of the building to tune the model, it is not certain that a good fit will result in a model that better predicts post-retrofit energy savings. Therefore, the basic idea here is that the simulation program (intended for use with the calibration technique) is used to generate surrogate utility bill data and retrofit energy savings data against which the calibration technique can be tested. This provides three figures of merit for testing a calibration technique, 1) accuracy of the post-retrofit energy savings prediction, 2) closure on the 'true' input parameter values, and 3) goodness of fit to the utility bill data. The paper will also discuss the pros and cons of using this synthetic surrogate data approach versus trying to use real data sets of actual buildings.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Analysis of Real-World Preignition Data Using Neural Networks

Increasing adoption of downsized, boosted, spark-ignition engines has improved vehicle fuel economy, and continued improvement is desirable to reduce carbon emissions in the near-term. However, this strategy is limited by damaging preignition events which can cause hardware failure. Research to date has shed light on various contributing factors related to fuel and lubricant properties as well as calibration strategies, but the causal factors behind an individual preignition cycle remain elusive. If actionable precursors could be identified, mitigation through active control strategies would be possible. This paper uses artificial neural networks to search for identifiable precursors in the cylinder pressure data from a large real-world data set containing many preignition cycles. It is found that while follow-up preignition cycles in clusters can be readily predicted, the initial preignition cycle is not predictable based on features of the cylinder pressure. Further, this indicates that the alternating pattern of preignition cycles within clusters is influenced by the thermodynamic state as reflected in the pressure, but that the trigger for the initial preignition cycle is not thermodynamic in nature, but more likely tied to a critical threshold in the chemistry of the fuel/lubricant mixture in the upper crevice or other factors related to the presence of an ignition source.

33 ADVANCED PROPULSION SYSTEMS↗

Spherical Harmonic-based Random Fields Based on Real Particle 3D Data: Improved Numerical Algorithm and Quantitative Comparison to Real Particles

Many parameters affect the cyclone efficiency, and these parameters can have different effects in different flow regimes. Therefore the maximum-efficiency cyclone length is a function of the specific geometry and operating conditions in use. In this study, we obtained a relationship describing the minimum particle diameter or maximum cyclone efficiency by using a theoretical approach based on cyclone geometry and fluid properties. We have compared the empirical predictions with corresponding literature data and observed good agreement. The results address the importance of fluid properties. Inlet and vortex finder cross-sections, cone-apex diameter, inlet Reynolds number and surface roughness are found to be the other important parameters affecting cyclone height. The surface friction coefficient, on the other hand, is difficult to employ in the calculations.We developed a theoretical approach to find the maximum-efficiency heights for cyclones with tangential inlet and we suggested a relation for this height as a function of cyclone geometry and operating parameters. In order to generalize use of the relation, two dimensionless parameters, namely for geometric and operational variables, we defined and results were presented in graphical form such that one can calculate and enter the values of these dimensionless parameters and then can find the maximum efficiency height of his own specific cyclone.

99 GENERAL AND MISCELLANEOUS↗

Locating Surface Explosions by Combining Seismic and Infrasound Data

An outline of a Bayesian source location framework for using seismic and acoustic observations is developed and tested on synthetic and real data. Seismic and acoustic phenomena are both commonly used in detection and location of a variety of natural or man–made events, such as volcanic eruptions, quarry blasts, and military exercises. Typically, seismic and acoustic observations have been utilized independently of each other. Here, we outline a Bayesian formulation for combining the two observations in a single estimate of the location and origin time. Using realistic estimates of uncertainty, we subsequently explore how combining the different observation types can benefit event location at local to near–regional distances. We apply the method to synthetic data and to real observations from a mining blast in Bingham Mine in Utah. Our findings suggest that, for relatively sparse or azimuthally limited observations, the relative strengths of the two different phenomenologies enable more precise joint–event localization than either seismic or infrasonic measurements alone.

58 GEOSCIENCES↗

A NEW METHOD FOR DERIVING THE STELLAR BIRTH FUNCTION OF RESOLVED STELLAR POPULATIONS

We present a new method for deriving the stellar birth function (SBF) of resolved stellar populations. The SBF (stars born per unit mass, time, and metallicity) is the combination of the initial mass function (IMF), the star formation history (SFH), and the metallicity distribution function (MDF). The framework of our analysis is that of Poisson Point Processes (PPPs), a class of statistical models suitable when dealing with points (stars) in a multidimensional space (the measurement space of multiple photometric bands). The theory of PPPs easily accommodates the modeling of measurement errors as well as that of incompleteness. Our method avoids binning stars in the color–magnitude diagram and uses the whole likelihood function for each data point; combining the individual likelihoods allows the computation of the posterior probability for the population's SBF. Within the proposed framework it is possible to include nuisance parameters, such as distance and extinction, by specifying their prior distributions and marginalizing over them. The aim of this paper is to assess the validity of this new approach under a range of assumptions, using only simulated data. Forthcoming work will show applications to real data. Although it has a broad scope of possible applications, we have developed this method to study multi-band Hubble Space Telescope observations of the Milky Way Bulge. Therefore we will focus on simulations with characteristics similar to those of the Galactic Bulge.

79 ASTRONOMY AND ASTROPHYSICS↗

Studies of the muon momentum calibration and performance of the ATLAS detector with $pp$ collisions at $\sqrt{s}=13$ TeV

This paper presents the muon momentum calibration and performance studies for the ATLAS detector based on the $pp$ collisions data sample produced at $\sqrt{s}$ = 13 TeV at the LHC during Run 2 and corresponding to an integrated luminosity of 139 fb –1 . An innovative approach is used to correct for potential charge-dependent momentum biases related to the knowledge of the detector geometry, using the $Z$ → $μ^+$ $μ^–$ resonance. The muon momentum scale and resolution are measured using samples of $J/ψ$ → $μ^+$ $μ^–$ and $Z$ → $μ^+$ $μ^–$ events. A calibration procedure is defined and applied to simulated data to match the performance measured in real data. The calibration is validated using an independent sample of $Υ$ → $μ^+$ $μ^–$ events. At the Z ($J/ψ$) peak, the momentum scale is measured with an uncertainty at the 0.05% (0.1%) level, and the resolution is measured with an uncertainty at the 1.5% (2%) level. The charge-dependent bias is removed with a dedicated in situ correction for momenta up to 450 GeV with a precision better than 0.03 TeV –1 .

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

WE-AB-207A-06: Progressive Dose Control for Cone Beam CT with Deformation Assisted Temporal Nonlocal Means Method

Purpose: To reduce cone beam CT (CBCT) imaging dose, we previously proposed a progressive dose control (PDC) scheme to employ temporal correlation between CBCT images at different fractions for image quality enhancement. A temporal non-local means (TNLM) method was developed to enhance quality of a new low-dose CBCT using existing high-quality CBCT. To enhance a voxel value, the TNLM method searches for similar voxels in a window. Due to patient deformation among the two CBCTs, a large searching window was required, reducing image quality and computational efficiency. This abstract proposes a deformation-assisted TNLM (DA-TNLM) method to solve this problem. Methods: For a low-dose CBCT to be enhanced using a high-quality CBCT, we first performed deformable image registration between the low-dose CBCT and the high-quality CBCT to approximately establish voxel correspondence between the two. A searching window for a voxel was then set based on the deformation vector field. Specifically, the search window for each voxel was shifted by the deformation vector. A TNLM step was then applied using only voxels within this determined window to correct image intensity at the low-dose CBCT. Results: We have tested the proposed scheme on simulated CIRS phantom data and real patient data. The CITS phantom was scanned on Varian onboard imaging CBCT system with coach shifting and dose reducing for each time. The real patient data was acquired in four fractions with dose reduced from standard CBCT dose to 12.5% of standard dose. It was found that the DA-TNLM method can reduce total dose by over 75% on average in the first four fractions. Conclusion: We have developed a PDC scheme which can enhance the quality of image scanned at low dose using a DA-TNLM method. Tests in phantom and patient studies demonstrated promising results.

60 APPLIED LIFE SCIENCES↗

Robust Event Classification Using Imperfect Real-world PMU Data

Here, this paper studies robust event classification using imperfect real-world phasor measurement unit (PMU) data. By analyzing the real-world PMU data, we find it is challenging to directly use this dataset for event classifiers due to the low data quality observed in PMU measurements and event logs. To address these challenges, we develop a novel machine learning framework for training robust event classifiers, which consists of three main steps: data preprocessing, fine-grained event data extraction, and feature engineering. Specifically, the data preprocessing step addresses the data quality issues of PMU measurements (e.g., bad data and missing data); in the fine-grained event data extraction step, a model-free event detection method is developed to accurately localize the events from the inaccurate event timestamps in the event logs; and the feature engineering step constructs the event features based on the patterns of different event types, in order to improve the performance and the interpretability of the event classifiers. Based on the proposed framework, we develop a workflow for event classification using the real-world PMU data streaming into the system in real time. Using the proposed framework, robust event classifiers can be efficiently trained based on many off-the-shelf lightweight machine learning models. Numerical experiments using the real-world dataset from the Western Interconnection of the U.S power transmission grid show that the event classifiers trained under the proposed framework can achieve high classification accuracy while being robust against low-quality data.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

An investigation on machine learning predictive accuracy improvement and uncertainty reduction using VAE-based data augmentation

The confluence of ultrafast computers with large memory, rapid progress in Machine Learning (ML) algorithms, and the availability of large datasets place multiple engineering fields at the threshold of dramatic progress. However, a unique challenge in nuclear engineering is data scarcity because experimentation on nuclear systems is usually more expensive and time-consuming than most other disciplines. One potential way to resolve the data scarcity issue is deep generative learning, which uses certain ML models to learn the underlying distribution of existing data and generate synthetic samples that resemble the real data. In this way, one can significantly expand the dataset to train more accurate predictive ML models. In this study, our objective is to evaluate the effectiveness of data augmentation using variational autoencoder (VAE)-based deep generative models. We investigated whether the data augmentation leads to improved accuracy in the predictions of a deep neural network (DNN) model trained using the augmented data. Additionally, the DNN prediction uncertainties are quantified using Bayesian Neural Networks (BNN) and conformal prediction (CP) to assess the impact on predictive uncertainty reduction. To test the proposed methodology, we used TRACE simulations of steady-state void fraction data based on the NUPEC Boiling Water Reactor Full-size Fine-mesh Bundle Test (BFBT) benchmark. Here, we found that augmenting the training dataset using VAEs has improved the DNN model’s predictive accuracy, improved the prediction confidence intervals, and reduced the prediction uncertainties.

Bayesian neural network↗

SCALER MODE OF THE AUGER OBSERVATORY AND SUNSPOTS

Recent data from the Auger Observatory on low-energy secondary cosmic ray particles are analyzed to study temporal correlations together with data on the daily sunspot numbers and neutron monitor data. Standard spectral analysis demonstrates that the available data show 1/f {sup {beta}} fluctuations with {beta} Almost-Equal-To 1 in the low-frequency range. All data behave like Brownian fluctuations in the high-frequency range. The existence of long-range correlations in the data was confirmed by detrended fluctuation analysis. The real data confirmed the correlation between the scaling exponent of the detrended analysis and the exponent of the spectral analysis.

79 ASTRONOMY AND ASTROPHYSICS↗

Implementation of barycentric resampling for continuous wave searches in gravitational wave data

We describe an efficient implementation of a coherent statistic for searches of continuous gravitational wave from neutron stars. The algorithm works by transforming the data taken by a gravitational wave detector from a moving Earth bound frame to one that sits at the Solar System barycenter. Many practical difficulties arise in the implementation of this algorithm, some of which have not been discussed previously. These difficulties include constraints of small computer memory, discreteness of the data, losses due to interpolation, and gaps in real data. This implementation is considerably more efficient than previous implementations of these kinds of searches on Laser Interferometer Gravitational Wave (LIGO) detector data. The speed-up factors range from 10, when applied to Einstein-Home, to about 2000 for targeted searches which integrate over months of data.

79 ASTRONOMY AND ASTROPHYSICS↗

Detecting Gravitational Lensing of the Cosmic Microwave Background by Galaxy Clusters

Clusters of galaxies gravitationally lens the Cosmic Microwave Background (CMB) leading to a distinct signal in the CMB on arcminute scales. Measurement of the cluster lensing effect offers the exciting possibility of constraining the masses of galaxy clusters using CMB data alone. Improved constraints on cluster masses are in turn essential to the use of clusters as cosmological probes: uncertainties in cluster masses are currently the dominant systematic affecting cluster abundance constraints on cosmology. To date, however, the CMB cluster lensing signal remains undetected because of its small magnitude and angular size. In this thesis, we develop a maximum likelihood approach to extracting the signal from CMB temperature data. We validate the technique by applying it to mock data designed to replicate as closely as possible real data from the South Pole Telescope’s (SPT) Sunyaev-Zel’dovich (SZ) survey: the effects of the SPT beam, transfer function, instrumental noise and cluster selection are incorporated. We consider the effects of foreground emission on the analysis and show that uncertainty in amount of foreground lensing results in a small systematic error on the lensing constraints. Additionally, we show that if unaccounted for, the SZ effect leads to unacceptably large biases on the lensing constraints and develop an approach for removing SZ contamination. The results of the mock analysis presented here suggest that a 4σ first detection of the cluster lensing effect can be achieved with current SPT-SZ data.

79 ASTRONOMY AND ASTROPHYSICS↗

Machine learning applied to classifying neutron resonances

The performance of nuclear reactors and other nuclear systems depends on a precise understanding of the neutron interaction cross sections for materials used in these systems. These cross sections exhibit resonance structure whose shape is determined in part by the angular momentum quantum numbers of the resonances. The correct assignment of the quantum numbers of neutron resonances is therefore of paramount importance. In this project, we apply a machine learning technique, namely decision trees, to automate the quantum number assignments. The tree is trained from simulated data generated to mimic the errors found in real data. We explore the use of several physics-motivated features for training our trees, including the nearest neighbor spacing distribution, cumulative level distribution, and channel width distributions. Initial results using random matrix theory motivated fits which demonstrated that we can determine resonance spin groups somewhat reliably. If we use these fits as features in our trees, we can train them to spot outliers corresponding to misassigned resonances. We found that with the large number of features used in this project that the decision tree tended to over t training data resulting in poor performance with respect to the test data. By reducing the number of features, we can achieve nearly perfect assignment of quantum numbers with our training data.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

The Modeled Performance of the ORNL Matched Filter Detector for Seismic Signatures of Multi-Axle Tractor Trailers

The MINOS program seeks methods to passively monitor the transfer of special materials, which are reproducible and quantitatively defensible. To date, we have demonstrated that short-period seismometers deployed near vehicle tracks can detect and monitor signatures of multi-axle tractor trailers impacting small road obstacles as means to detect and identify some transfer activities. In this report, we model the seismic velocity signal output from a tractor-trailer impacting a small obstacle (a template waveform) to parameterize a matched filter that identifies signals with similar shapes buried in noise, and that indicate repeating events (target waveforms). We then use this model to perform thousands of synthetic experiments, in which we apply our detector to identify noisy waveforms output from a similarly modeled vehicle data set; we process all data in the same manner as we process real data collected from the Oak Ridge National Laboratory (ORNL) seismic network. Our experiments, while limited, reveal that the matched filter shows a high discrimination capability, and at very low SNR values (at visibility thresholds). This means that template waveforms can match target waveforms that are sourced by models that share the same vehicle speed (target vehicles), and have a low probability of mis-identification. In particular, a matched filter designed to detect a five-axle tractor-trailer will reliably identify that target vehicle down to an SNR of 11.25 dB, and within ±0.1 km/hr, while successfully screening target vehicles with three axles.

42 ENGINEERING↗

High-Energy X-ray Diffraction Microscopy for Nuclear Forensics FY2022 Project Report

Morphological information on nuclear material has been identified using visible light and scanning electron microscopy. These identify qualitative differences in particle morphology. Three-dimensional imaging of materials through alternating scanning electron microscopy imaging and focused ion beam milling has also been used. Unfortunately, these techniques are time- and labor-intensive, with significant sample preparation required and lengthy analysis times. Further, the resulting 3D images are qualitative, require manual identification, and do not capture statistically-representative populations. High energy X-ray 3D imaging using a direct-beam or diffracted-beam (High-Energy Diffraction Microscopy) have been developed at the Advanced Photon Source and can produce quantitative information on grains (phase, location, etc.) and pores (size distribution, sphericity) in a material. These techniques require only minutes to characterize a sample volume and are non-destructive, thus suitable for a wide range of existing samples and for confirmatory analyses to be carried out using conventional microscopy techniques. In this first year of the project, all uranium oxide samples were synthesized and characterized using conventional analyses by the analytical chemistry laboratory. Conventional analysis methods included powder x-ray diffraction, scanning electron microscopy, impurity analysis via inductively coupled plasma mass spectrometry, and infrared spectroscopy. Impurity analysis shows a drop in boron content from UO 3 to the lowest U 3 O 8 calcination temperature, but otherwise no appreciable difference in any sample. Analysis of diffraction data shows a flip of peaks from UO 3 dominated for the 600 °C calcined sample to U 3 O 8 dominated at 700 °C and 800 °C. Analysis of scanning electron microscopy images shows that with increased calcination temperature the size distribution of particles seems to increase and broaden. Both of these last findings are in line with previously published data, though this work used significantly fewer particles to simply show similar trends instead of getting truly quantitative particle analysis. Infrared analysis similarly shows ingrowth of U 3 O 8 as calcination temperature is increased, along with depression of peaks associated with UO 3 and water. Samples were prepared for analysis at the Advanced Photon Source at beamline 1-ID. It is anticipated that analysis will occur in November of 2022. AI/ML techniques to de-noise data coming out of 1-ID during the analyses was also developed during this time using previously gathered data. Preliminary results using a self-supervision technique called Noise2Selfshow good de-noising of data. Once the uranium oxide samples are analyzed, real data will be used to test the de-noising and other AI/ML techniques that may be developed in the second year of the project.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Machine Learning for Neutron Resonance Evaluations [Slides]

The performance of nuclear reactors and other nuclear systems depends on a precise understanding of the neutron interaction cross sections for materials used in these systems. These cross sections exhibit a resonance structure whose shape is determined in part by the angular momentum quantum numbers of the resonances. The correct assignment of the quantum numbers of neutron resonances is therefore of paramount importance. In this presentation, we describe the application of machine learning to automate the quantum number assignments. Scikit-learn classifiers were trained on simulated resonance data whose statistical properties were chosen to mimic real data. We explored the use of several physics (and random matrix theory)-motivated features for training the classifiers, including the nearest neighbor spacing distribution, cumulative level distribution, and channel width distributions. Initial results demonstrated that we can determine resonance spin groups somewhat reliably. We are now investigating the application of our approach to 52 Cr resonance data.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Mock Quasar-Lyman-α forest data-sets for the SDSS-III Baryon Oscillation Spectroscopic Survey

We describe mock data-sets generated to simulate the high-redshift quasar sample in Data Release 11 (DR11) of the SDSS-III Baryon Oscillation Spectroscopic Survey (BOSS). The mock spectra contain Lyα forest correlations useful for studying the 3D correlation function including Baryon Acoustic Oscillations (BAO). They also include astrophysical effects such as quasar continuum diversity and high-density absorbers, instrumental effects such as noise and spectral resolution, as well as imperfections introduced by the SDSS pipeline treatment of the raw data. The Lyα forest BAO analysis of the BOSS collaboration, described in Delubac et al. 2014, has used these mock data-sets to develop and cross-check analysis procedures prior to performing the BAO analysis on real data, and for continued systematic cross checks. Tests presented here show that the simulations reproduce sufficiently well important characteristics of real spectra. These mock data-sets will be made available together with the data at the time of the Data Release 11.

79 ASTRONOMY AND ASTROPHYSICS↗