Search NASA⌕ Search

SEARCH · Search NASA

Results for “Image coding”

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 1,675 records · Page 93
All sourcesNASA NTRSDOE OSTI
DOE OSTI2017

Ignition and flame characteristics of cryogenic hydrogen releases

In this work, under-expanded cryogenic hydrogen jets were investigated experimentally for their ignition and flame characteristics. The test facility described herein, was designed and constructed to release hydrogen at a constant temperature and pressure, to study the dispersion and thermo-physical properties of cryogenic hydrogen releases and flames. In this study, a non-intrusive laser spark focused on the jet axis was used to measure the maximum ignition distance. The radiative power emitted by the corresponding jet flames was also measured for a range of release scenarios from 37 K to 295 K, 2–6 bar abs through nozzles with diameters from 0.75 to 1.25 mm. The maximum ignition distance scales linearly with the effective jet diameter (which scales as the square root of the stagnant fluid density). A 1-dimensional (stream-wise) cryogenic hydrogen release model developed previously at Sandia National Laboratories (although this model is not yet validated for cryogenic hydrogen) was exercised to predict that the mean mole fraction at the maximum ignition distance is approximately 0.14, and is not dependent on the release conditions. The flame length and width were extracted from visible and infra-red flame images for several test cases. The flame length and width both scale as the square root of jet exit Reynolds number, as reported in the literature for flames from atmospheric temperature hydrogen. As shown in previous studies for ignited atmospheric temperature hydrogen, the radiative power from the jet flames of cold hydrogen scales as a logarithmic function of the global flame residence time. The radiative heat flux from jet flames of cold hydrogen is higher than the jet flames of atmospheric temperature hydrogen, for a given mass flow rate, due to the lower choked flow velocity of low-temperature hydrogen. Lastly, this study provides critical information with regard to the development of models to inform the safety codes and standards of hydrogen infrastructure.

08 HYDROGEN↗
DOE OSTI2018

U 3 Si 2 Fabrication and Testing for Implementation into the BISON Fuel Performance Code

A creep test stand was designed and constructed for compressive creep testing of U 3 Si 2 pellets. This is described in Chapter 3. Creep testing of U 3 Si 2 pellets was completed. In total, 13 compressive creep tests of U 3 Si 2 pellets was successfully completed. This is reported in Chapter 3. Secondary creep model of U 3 Si 2 was developed and implemented in BISON. This is described in Chapter 4. Properties of U 3 Si 2 were implemented in BISON. This is described in Chapter 4. A resonant frequency and damping analyzer (RFDA) using impulse excitation technique (IET) was setup, tested, and used to analyze U 3 Si 2 samples to measure Young’s and Shear Moduli which were then used to calculate the Poisson ratio for U 3 Si 2 . This is described in Chapter 5. Characterization of U 3 Si 2 samples was completed. Samples were prepared and analyzed by XRD, SEM, and optical microscopy. Grain size analysis was conducted on images. SEM with EDS was used to analyze second phase precipitates. Impulse excitation technique was used to determine the Young’s and Shear Moduli of a tile specimen which allowed for the determination of the Poisson ratio. Helium pycnometry and mercury intrusion porosimetry was performed and used with image analysis to determine porosity size distribution. Vickers microindentation characterization method was used to evaluate the mechanical properties of U 3 Si 2 including toughness, hardness, and Vickers hardness. Electrical resistivity measurement was done using the four-point probe method. This is reported in Chapter 5.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2023

Erasure conversion in a high-fidelity Rydberg quantum simulator

Minimizing and understanding errors is critical for quantum science, both in noisy intermediate scale quantum (NISQ) devices and for the quest towards fault-tolerant quantum computation. Rydberg arrays have emerged as a prominent platform in this context with impressive system sizes and proposals suggesting how error-correction thresholds could be significantly improved by detecting leakage errors with single-atom resolution, a form of erasure error conversion. However, two-qubit entanglement fidelities in Rydberg atom arrays have lagged behind competitors and this type of erasure conversion is yet to be realized for matter-based qubits in general. Here we demonstrate both erasure conversion and high-fidelity Bell state generation using a Rydberg quantum simulator. When excising data with erasure errors observed via fast imaging of alkaline-earth atoms, we achieve a Bell state fidelity of $\ge 0.997{1}_{-13}^{+10}$ , which improves to $\ge 0.998{5}_{-12}^{+7}$ when correcting for remaining state-preparation errors. We further apply erasure conversion in a quantum simulation experiment for quasi-adiabatic preparation of long-range order across a quantum phase transition, and reveal the otherwise hidden impact of these errors on the simulation outcome. Our work demonstrates the capability for Rydberg-based entanglement to reach fidelities in the 0.999 regime, with higher fidelities a question of technical improvements, and shows how erasure conversion can be utilized in NISQ devices. These techniques could be translated directly to quantum-error-correction codes with the addition of long-lived qubits.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2021

Burst characteristics of advanced accident-tolerant FeCrAl cladding under temperature transient testing

Here assessment of burst characteristics of accident-tolerant fuel (ATF) claddings is essential to evaluate the safety margins of nuclear reactors, and better understanding can enable accelerated licensing for reactor concepts that include the new materials. Therefore, this study investigated the burst behavior of an ATF candidate of C26M, which is an iron–chromium–aluminum alloy (FeCrAl) under transient testing like the simulated loss-of-coolant accident (LOCA) conditions, except the water-quenching phase, in light-water reactors (LWRs). The effect of LOCA specimen length was assessed in terms of post-burst tube parameters. No critical length effect was determined on the burst pressure, burst location, and burst size measurements. Burst temperature showed larger variation likely due to its measurement approach used in this study. The post-test diametral strain was identified as the critical parameter to reduce the specimen length as compared to tube burst length or width. Postmortem optical metrology and digital image correlation were employed to determine the local strain-state during the simulated LOCA, which showed the loading path was close to equibiaxial conditions in regions away from the burst location, rather than internally pressurized conditions (or “plane-strain” tension). Pst-test microstructural characterizations of the FeCrAl revealed that ductile damage was present at the edge of the outer radial surface while the rest of the material was ruptured via cleavage at LOCA burst temperatures. Furthermore, internal grain boundary cracks were observed at locations away from the tube burst region. This behavior was considered to occur during the cooling down period of the simulated LOCA test. Overall, this study aimed to provide essential data for fuel performance code development by considering micromechanics of the deformation and failure.

36 MATERIALS SCIENCE↗
DOE OSTI2011

Introduction to the Geant4 Simulation toolkit

Geant4 is a Monte Carlo simulation Toolkit, describing the interactions of particles with matter. Geant4 is widely used in radiation physics research, from High Energy Physics, to medical physics and space science, thanks to its sophisticated physics component, coupled with advanced functionality in geometry description. Geant4 is widely used at the Centre for Medical Radiation Physics (CMRP), at the University of Wollongong, to characterise and optimise novel detector concepts, radiotherapy treatments, and imaging solutions. This lecture consists of an introduction to Monte Carlo method, and to Geant4. Particular attention will be devoted to the Geant4 physics component, and to the physics models describing electromagnetic and hadronic physics interactions. The second part of the lecture will be focused on the methodology to adopt to develop a Geant4 simulation application.

61 RADIATION PROTECTION AND DOSIMETRY↗
DOE OSTI2012

Evaluation of the potential utility of flat panel CT for quantifying relative contrast enhancement

Purpose: Certain directed oncologic therapies seek to take advantage of the fact that tumors are typically more susceptible to directed therapeutic agents than normal tissue owing to their extensive networks of poorly formed, leaky vasculature. If differences between the vascularity of normal and tumor tissues could be quantified, patients could be selected for or excluded from directed treatments on the basis of this difference. However, angiographic imaging techniques such as digital subtraction angiography (DSA) yield two-dimensional data that may be inadequate for this task. As a first step, the authors evaluated the feasibility of using a commercial implementation of flat panel computed tomography (FPCT) to quantify differences in enhancement of a simulated tumor compared with normal tissue based on differences in CT number measured in precontrast and postcontrast scans. Methods: To evaluate the FPCT scanner studied, the authors scanned several phantoms containing simulated normal and tumor tissues. In the first experiment, the authors used an anthropomorphic phantom containing inclusions representing normal, tumor, and bone tissue to evaluate the constancy of CT numbers in scans repeated at clinically relevant intervals of 1 and 3 min. The authors then scanned gelatin phantoms containing dilutions of iodinated contrast to evaluate the accuracy of relative contrast enhancement measurements for a clinical FPCT system. Data were analyzed using widely available software. Results: CT numbers measured in identical locations were constant over both scan intervals evaluated. Measured relative contrast enhancement values were accurate compared with known relative contrast enhancement values. Care must be taken to avoid artifacts in reconstructed images when placing regions of interest. Conclusions: Despite its limitations, FPCT in the interventional laboratory can be used to quantify relative contrast enhancement in phantoms. This is accomplished by measuring CT number in simulated tumor and normal tissue on precontrast and postcontrast scans. This information opens the door for refinement of technique in an effort to use such a technique to plan directed therapies.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2024

Measurement and calibration of non-linear shear terms in galaxy cluster fields

Introduction Galaxy cluster lensing is a powerful tool for measuring the mass of galaxy clusters, but accurate shear measurement and calibration are critical to obtaining reliable results. This study focuses on the measurement and calibration of weak lensing shears to improve mass estimates in cluster lensing. To deal with the problem, we first developed an image simulation pipeline, jedisim, which utilizes galaxy images extracted from the Hubble Space Telescope (HST) Ultra Deep Field (UDF) and the Cosmic Assembly Near-infrared Deep Extragalactic Legacy Survey (CANDELS). Methods The simulations represent realistic galaxy distributions and morphologies as input sources. The foreground halo with a Navarro–Frenk–White (NFW) profile is constructed such that the lensing signals of background galaxies can be measured by the Vera C. Rubin Observatory’s Legacy Survey of Space and Time (LSST) Science Pipelines. By comparing the measured reduced shear g meas and the true reduced shear g true , we observe non-linearity up to g ≲ 0.6 . We fit polynomials to the data with quadratic correction adequate to g ≲ 0.4 . Meanwhile, we conduct mass estimates using the pzmassfitter code on four different clusters. Results The mass estimate results are significantly improved after applying the shear calibration derived from the present work—from 4.954 ± 0.504 × 1 0 14 M ⊙ to 10.507 ± 0.498 × 1 0 14 M ⊙ after calibration for a simulated cluster with the mass of 10 × 1 0 14 M ⊙ . In multiple cases of validation, the estimated results are all consistent with true cluster mass. Discussion This study yields the first relationship between reality and shape measurement of the LSST Science Pipelines and serves as the first step toward the overall goal of mass calibration in cluster lensing. By addressing the challenges in shear measurement and calibration, we aim to enhance the accuracy and reliability of mass estimates in galaxy cluster lensing studies.

Astronomy & Astrophysics↗
DOE OSTI2016

Numerical Study of Vortex Core in Turbulent Swirling Flow

Swirling flow has attracted a great deal of research interest. An application of swirling flow can be found in many engineering areas; such as gas cyclone and heat transfer enhancement. In nuclear engineering, swirling flow is used as heat transfer enhancement in rod bundle by installing grid spacer and mixing vane. Because of its extensive applications, there has been a great number of experimental and numerical investigations of swirling flow. McClusky et al., experimentally studied mapping of the lateral velocity in rod bundle using Laser Doppler Velocimetry (LDV) and Conner et al. predicted the flow field in rod bundle by CFD. To understand the characteristics of swirling flow such as tangential velocity and vortex core swirling flow, some researchers investigated swirling flow in a pipe instead of in rod bundle. This is due to the similarity between swirling flow in a pipe and rod bundle. In recent study, flow mapping of 2D velocity field of single-phase swirling flow had been conducted experimentally by Particle Image Velocimetry (PIV). However, this method needs calibration and complicated. The new measurement technique for flow mapping in swirling flow was done by phased array Ultrasonic Velocity Profiler (PAUVP). Hamdani et al., successfully plotted velocity field and determined the swirling core region (vortex core). This technique is nonintrusive and no need calibration. However, there was a limitation of the measurement because phased array UVP cannot capture 360 deg. velocity field inside the pipe. In addition, tangential velocity in CFD results was not agree well with experiment data due asymmetric problem. To evaluate this limitation of experimental method, the numerical simulations using Computational Fluid Dynamics (CFD) needs to be performed. The objective of this study is to use the commercial CFD code Fluent to model the developing turbulent swirling flow induced by twisted tape inside a straight pipe, and to gain an understanding of such a flow and the turbulence models involved in the simulation. Straight pipe is used instead of rod bundle for a simplification. In this work, the Renormalized Group (RNG) k-{epsilon} turbulence model and Reynolds Stress Model (RSM) turbulence model, are performed to study the phenomena of vortex core in a pipe with twisted tape inserts. Effects of the Reynolds number and asymmetric of twisted tape (inclination, {alpha} = 5 deg. and {alpha} = 20 deg.) on swirling core position are examined using water as the testing fluid. Furthermore, the Navier-Stokes equations will be solved to obtain results that can be validated with the previous experimental results. (authors)

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2019

How to detect and correct myocardial creep in myocardial perfusion imaging using Rubidium-82 PET?

Reliability of myocardial blood flow (MBF) quantification in myocardial perfusion imaging (MPI) using PET can majorly be affected by the occurrence of myocardial creep when using pharmacologically induced stress. In this paper, we provide instructions on how to detect and correct for myocardial creep. For example, in each time frame of the PET images the myocardium contour and the observed activity have to be compared to check for misalignments. In addition, we provide an overview of the functionality of commonly used software packages to perform this quality control step as not all software packages currently provide this functionality. Furthermore, important clinical considerations to obtain accurate MBF measurements are given.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2013

DETECTION OF FAST TRANSIENTS WITH RADIO INTERFEROMETRIC ARRAYS

Next-generation radio arrays, including the Square Kilometre Array (SKA) and its pathfinders, will open up new avenues for exciting transient science at radio wavelengths. Their innovative designs, comprising a large number of small elements, pose several challenges in digital processing and optimal observing strategies. The Giant Metre-wave Radio Telescope (GMRT) presents an excellent test-bed for developing and validating suitable observing modes and strategies for transient experiments with future arrays. Here we describe the first phase of the ongoing development of a transient detection system for GMRT that is planned to eventually function in a commensal mode with other observing programs. It capitalizes on the GMRT's interferometric and sub-array capabilities, and the versatility of a new software backend. We outline considerations in the plan and design of transient exploration programs with interferometric arrays, and describe a pilot survey that was undertaken to aid in the development of algorithms and associated analysis software. This survey was conducted at 325 and 610 MHz, and covered 360 deg{sup 2} of the sky with short dwell times. It provides large volumes of real data that can be used to test the efficacies of various algorithms and observing strategies applicable for transient detection. We present examples that illustrate the methodologies of detecting short-duration transients, including the use of sub-arrays for higher resilience to spurious events of terrestrial origin, localization of candidate events via imaging, and the use of a phased array for improved signal detection and confirmation. In addition to demonstrating applications of interferometric arrays for fast transient exploration, our efforts mark important steps in the roadmap toward SKA-era science.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2022

Exploiting the Local Parabolic Landscapes of Adversarial Losses to Accelerate Black-Box Adversarial Attack

Existing black-box adversarial attacks on image classifiers update the perturbation at each iteration from only a small number of queries of the loss function. Since the queries contain very limited information about the loss, black-box methods usually require much more queries than white-box methods. We propose to improve the query efficiency of black-box methods by exploiting the smoothness of the local loss landscape. However, many adversarial losses are not locally smooth with respect to pixel perturbations. To resolve this issue, our first contribution is to theoretically and experimentally justify that the adversarial losses of many standard and robust image classifiers behave like parabolas with respect to perturbations in the Fourier domain. Our second contribution is to exploit the parabolic landscape to build a quadratic approximation of the loss around the current state, and use this approximation to interpolate the loss value as well as update the perturbation without additional queries. Since the local region is already informed by the quadratic fitting, we use large perturbation steps to explore far areas. We demonstrate the efficiency of our method on MNIST, CIFAR-10 and ImageNet datasets for various standard and robust models, as well as on Google Cloud Vision. The experimental results show that exploiting the loss landscape can help significantly reduce the number of queries and increase the success rate. Our codes are available at https://github.com/HoangATran/BABIES.

Tran, Hoang↗
DOE OSTI2020

Observation of shock-front separation in multi-ion-species collisional plasma shocks

We observe shock-front separation and species-dependent shock widths in multi-ion-species collisional plasma shocks, which are produced by obliquely merging plasma jets of a He/Ar mixture (97% He and 3% Ar by initial number density) on the plasma liner experiment. Visible plasma emission near the He-I 587.6-nm and Ar-II 476.5–514.5-nm lines is simultaneously recorded by splitting a single visible image of the shock into two different fast-framing cameras with different narrow bandpass filters (589 ± 5 nm for observing the He-I line and 500 ± 25 nm for the Ar-II lines). For conditions in these experiments (pre-shock ion and electron densities ≈5×10 14 cm –3 , ion and electron temperatures of ≈2.2 eV, and relative plasma-merging speed of 22 km/s), the observationally inferred magnitude of He/Ar shock-front separation and the shock widths themselves are <1 cm, which correspond to ~50 post-shock thermal ion–ion mean free paths. These experimental length scales are in reasonable qualitative and quantitative agreement with results from 1D multi-fluid simulations using the CHICAGO code. However, we note there are differences between the experimentally inferred and simulation-predicted ionization states and line emission intensities, particularly in the post-shock region. Overall, the experimental and simulation results are consistent with theoretical predictions that the lighter He ions diffuse farther ahead within the overall shock front than the heavier Ar ions.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2022

Unsupervised learning of ferroic variants from atomically resolved STEM images

An approach for the analysis of atomically resolved scanning transmission electron microscopy data with multiple ferroic variants in the presence of imaging non-idealities and chemical variabilities based on a rotationally invariant variational autoencoder (rVAE) is presented. We show that an optimal local descriptor for the analysis is a sub-image centered at specific atomic units, since materials and microscope distortions preclude the use of an ideal lattice as a reference point. The applicability of unsupervised clustering and dimensionality reduction methods is explored and is shown to produce clusters dominated by chemical and microscope effects, with a large number of classes required to establish the presence of rotational variants. Comparatively, the rVAE allows extraction of the angle corresponding to the orientation of ferroic variants explicitly, enabling straightforward identification of the ferroic variants as regions with constant or smoothly changing latent variables and sharp orientational changes. This approach allows further exploration of the chemical variability by separating the rotational degrees of freedom via rVAE and searching for remaining variability in the system. The code used in this article is available at https://github.com/saimani5/ferroelectric_domains_rVAE.

30 DIRECT ENERGY CONVERSION↗
DOE OSTI2017

Edge turbulence and divertor heat flux width simulations of Alcator C-Mod discharges using an electromagnetic two-fluid model

The BOUT++ code has been utilized in order to improve the understanding of the role of turbulent modes in controlling edge transport and resulting scaling of the scrape-off layer (SOL) heat flux width. For the C-Mod enhanced D α (EDA) H-mode discharges, BOUT++ six-field two-fluid nonlinear simulations show a reasonable agree- ment of upstream turbulence and divertor target heat flux behavior: a) The simulated quasi-coherent modes (QCMs) show consistent characteristics of the frequency versus poloidal wave number spectra of the electromagnetic fluctuations when compared with experimental measurements: frequencies are around 60-120 kHz (experiment: about 70-110 kHz), k θ are around 2.0 cm ₋1 which is similar to the Phase Contrast Imaging data; b) Linear spectrum analysis is consistent with the nonlinear phase relationship cal- culation which indicates the dominance of resistive-ballooning modes and drift-Alfven wave instabilities; c) The SOL heat flux width λ q vs current I p scaling is reproduced by turbulent transport: the simulations yield similar λ q to experimental measurements within a factor of 2. Yet the magnitudes of divertor heat fluxes can be varied, depending on the physics models, sources and sinks, sheath boundary conditions, or flux limiting coefficient; d) Simple estimate by the "2-point model" for λ q is consistent with simulation. Moreover, blobby turbulent spreading is confirmed for these relatively high B p shots.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2018

EXTINCTION DROPLET SIZING MEASUREMENTS IN DIESEL RELEVANT SPRAYS

A new diagnostic for the quantification of Sauter Mean Diameter (SMD) in high-pressure fuel sprays has been recently developed using combined optical and x-ray measurements at the Georgia Institute of Technology and Argonne National Laboratory, respectively. This diagnostic utilizes liquid scattering extinction measurements from diffuse back-illumination (DBI) imaging, conducted at Georgia Tech, and liquid absorption measurements from x-ray radiography, conducted at Argonne’s Advanced Photon Source. The new diagnostic, entitled the Scattering Absorption Measurement Ratio (SAMR), quantifies two-dimensional distributions of path-integrated SMD, enabling construction of the spatial history of drop size development within practical fuel sprays. This technique offers unique benefits over conventional drop-sizing methods in that it can be more robust in optically dense regions of the spray, while also providing high spatial resolution of the corresponding droplet field. The spatially-resolved SMD measurements that result from the SAMR diagnostic will be especially valuable to the engine modeling community for the quantitative validation of spray submodels in engine CFD codes. The methodology for quantification of SMD distributions using the SAMR technique has been previously introduced. This thesis aims to extend the initial development of the SAMR technique by presenting in detail: the experimental methodologies used in the SAMR technique, including the development of an ideal DBI setup for this technique; the data processing methodologies developed by the author; two-dimensional Sauter Mean Diameter measurements within diesel sprays for various experimental conditions; a summary of the various sources of measurement uncertainty; an assessment for how the sources of uncertainty affect the quantified SMD; and paths to improve the technique moving forward. The SMD results show that for low ambient density conditions droplets decrease in size as radial and axial position increases. For high ambient density conditions, however, the droplets show a stable size near the spray centerline and steadily increase in size as distance from the centerline increases. Additionally, this work provides 2-D droplet sizing maps which are particularly useful for model validation.

33 ADVANCED PROPULSION SYSTEMS↗
DOE OSTI2018

EXTINCTION DROPLET SIZING MEASUREMENTS IN DIESEL RELEVANT SPRAYS

A new diagnostic for the quantification of Sauter Mean Diameter (SMD) in high-pressure fuel sprays has been recently developed using combined optical and x-ray measurements at the Georgia Institute of Technology and Argonne National Laboratory, respectively. This diagnostic utilizes liquid scattering extinction measurements from diffuse back-illumination (DBI) imaging, conducted at Georgia Tech, and liquid absorption measurements from x-ray radiography, conducted at Argonne’s Advanced Photon Source. The new diagnostic, entitled the Scattering Absorption Measurement Ratio (SAMR), quantifies two-dimensional distributions of path-integrated SMD, enabling construction of the spatial history of drop size development within practical fuel sprays. This technique offers unique benefits over conventional drop-sizing methods in that it can be more robust in optically dense regions of the spray, while also providing high spatial resolution of the corresponding droplet field. The spatially-resolved SMD measurements that result from the SAMR diagnostic will be especially valuable to the engine modeling community for the quantitative validation of spray submodels in engine CFD codes. The methodology for quantification of SMD distributions using the SAMR technique has been previously introduced. This thesis aims to extend the initial developmen of SAMR technique by presenting in detail: the experimental methodologies used in the SAMR technique, including the development of an ideal DBI setup for this technique; the data processing methodologies developed by the author; two-dimensional Sauter Mean Diameter measurements within diesel sprays for various experimental conditions; a summary of the various sources of measurement uncertainty; an assessment for how the sources of uncertainty affect the quantified SMD; and paths to improve the technique moving forward. The SMD results show that for low ambient density conditions droplets decrease in size as radial and axial position increases. For high ambient density conditions, however, the droplets show a stable size near the spray centerline and steadily increase in size as distance from the centerline increases. Additionally, this work provides 2-D droplet sizing maps which are particularly useful for model validation.

33 ADVANCED PROPULSION SYSTEMS↗
DOE OSTI2025

Automated Waterbox Inspection for Nuclear Power Plants Using Computer Vision - Based Change Detection

Nuclear power plant waterboxes require regular inspection for leaks, missing components, and structural damage during maintenance outages. Traditional manual inspection is time-consuming and poses safety risks from confined space entry. We developed an automated computer vision system for drone-based waterbox inspection in partnership with Florida Light and Power. Our approach uses feature detection and matching to identify critical changes between baseline and current inspection images, automatically flagging additions (leaks/debris), removals (missing plugs), and translations (displaced components) while compensating for drone movement and environmental variations. We systematically evaluated six feature matching methods, from classical approaches (SIFT+BF) to state-of-the-art neural networks (SuperPoint+SuperGlue), using both standard benchmarks (HPatches) and waterbox-specific validation with real-world augmentations. SuperPoint+SuperGlue achieved superior performance with 7.82 pixels RMSE and 100% success rate—2.8x better accuracy than our baseline. While the pre-trained model has commercial licensing restrictions for nuclear deployment, our findings validate this architecture for custom training. We implemented a real-time GUI demonstrating the SIFT+BF approach for immediate deployment, processing drone feeds at 30 FPS with color-coded change visualization. Future work includes training a custom SuperPoint+SuperGlue model on waterbox data and integrating Vision-Language Models for automated reporting and maintenance guidance.

46 - INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AN↗