Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Principal component analysis”

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
All sourcesNASA NTRSDOE OSTI
DOE OSTI2025

Petrology and geochemistry of the Middle Pennsylvanian (Langsettian) Clintwood coalbed, Pike County, Kentucky

Abstract The Middle Pennsylvanian, Langsettian sub-stage Clintwood coal, along with its correlatives, was one of the more important energy resources in eastern Kentucky. The coal thickens from the NE-SW-trending Belfry anticline in central Pike County to the southeast, towards the Virginia-Kentucky border. Much of the thickness increase is in the bright lithotypes below a dull lithology. The bright lithologies transition from a bright clarain + vitrain + fusain lithology to a clarain + fusain lithology to the southeast. Half of the lithotypes in the three analyzed sections exceed 1000-µg/g Rare earth elements + Y + Sc (REYSc) (ash basis). The La N /Sm N vs. Gd N /Yb N and La N /Sm N vs. Ce N /Ce N * plots suggest that the top lithotype may have had best indication of an oxidizing environment. Positive Gd N /Gd N * and Eu N /Eu N * in some of the lithologies suggests that there may have been a hydrothermal influence in the sediment sources and/or in the diagenesis of the coal. Consideration of La N /Sm N and Zr (µg/g; ash basis), Zr vs. Sr, and V/Cr vs. inertinite/(inertinite + vitrinite) indicates that the dull lithotypes existed in a distinct geochemical environment compared to the brighter lithotypes. The isolation of the dull lithotypes from the bright lithotypes is supported by principal components analysis on inertinite/(inertinite + vitrinite), La N /Sm N , V/Cr, Sr, and either Zr or Ln(100(TiO 2 )/Al 2 O 3 ).

Hower, James C. (ORCID:0000000346942776)↗
DOE OSTI2022

Non-intrusive reduced order modeling of natural convection in porous media using convolutional autoencoders: Comparison with linear subspace techniques

Natural convection in porous media is a highly nonlinear multiphysical problem relevant to many engineering applications (e.g., the process of CO 2 sequestration). Here, we extend and present a non-intrusive reduced order model of natural convection in porous media employing deep convolutional autoencoders for the compression and reconstruction and either radial basis function (RBF) interpolation or artificial neural networks (ANNs) for mapping parameters of partial differential equations (PDEs) on the corresponding nonlinear manifolds. To benchmark our approach, we also describe linear compression and reconstruction processes relying on proper orthogonal decomposition (POD) and ANNs. Further, we present comprehensive comparisons among different models through three benchmark problems. The reduced order models, linear and nonlinear approaches, are much faster than the finite element model, obtaining a maximum speed-up of 7 × 10 6 because our framework is not bound by the Courant–Friedrichs–Lewy condition; hence, it could deliver quantities of interest at any given time contrary to the finite element model. Our model’s accuracy still lies within a relative error of 7% in the worst-case scenario. We illustrate that, in specific settings, the nonlinear approach outperforms its linear counterpart and vice versa. We hypothesize that a visual comparison between principal component analysis (PCA) and t-Distributed Stochastic Neighbor Embedding (t-SNE) could indicate which method will perform better prior to employing any specific compression strategy.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Deep learning to estimate permeability using geophysical data

Time-lapse electrical resistivity tomography (ERT) is a popular geophysical method to estimate three-dimensional (3D) permeability fields from electrical potential difference measurements. Traditional inversion and data assimilation methods are used to ingest this ERT data into hydrogeophysical models to estimate permeability. Due to ill-posedness and the curse of dimensionality, existing inversion strategies provide poor estimates and low resolution of the 3D permeability field. Recent advances in deep learning provide us with powerful algorithms to overcome this challenge. This paper presents a deep learning (DL) framework to estimate the 3D subsurface permeability from time-lapse ERT data. To test the feasibility of the proposed framework, we train DL-enabled inverse models on simulation data. Each measurement in both synthetic and field data is standardized by removing the mean and scaling the time-series to unit variance. This pre-processing step is necessary to bring simulation data closer to field observations. Subsurface process models based on hydrogeophysics are used to generate this synthetic data. Training performed on limited simulation data resulted in the DL model over-fitting. An advanced data augmentation based on mixup is implemented to generate additional training samples to overcome this issue. This mixup technique creates weakly labeled (low-fidelity) samples from strongly labeled (high-fidelity) data. The weakly labeled training data is then used to develop DL-enabled inverse models and reduce over-fitting. As both time-lapse ERT (1133048 features/realization) and 3D permeability (585453 features/realization) data samples are from a high-dimensional space, principal component analysis (PCA) is employed to reduce dimensionality. Encoded ERT and encoded permeability are generated using the trained PCA estimators. A deep neural network is then trained to map the encoded ERT to encoded permeability. This mixup training and unsupervised learning allowed us to build a fast and reasonably accurate DL-based inverse model under limited simulation data. Results show that proposed weak supervised learning can capture salient spatial features in the 3D permeability field. Quantitatively, the average mean squared error (in terms of the natural log) on the strongly labeled training, validation, and test datasets is less than 0.5. The R 2 -score (global metric) is greater than 0.75, and the percent error in each cell (local metric) is less than 10%. Finally, an added benefit in terms of computational cost is that the proposed DL-based inverse model is at least O(10 4 ) times faster than running a forward model once it is trained. Data generation, DL model training, and hyperparameter tuning to identify optimal neural network architectures utilized high-performance computing resources while the DL inference is performed on a standard laptop. Approximately, O(10 5 ) processor hours are used for generating data and DL tuning and training. We acknowledge that the data generation and DL model development are expensive. But once a DL model is trained, it can be re-used for inversion rapidly for the given system, with set physics and domain. Note that traditional inversion may require multiple forward model simulations (e.g., in the order of 10 to 1000), which are very expensive. This computational savings ≈ O(10 5 ) – O(10 7 )) makes the proposed DL-based inverse model attractive for subsurface imaging and real-time ERT monitoring applications due to fast and yet reasonably accurate estimations of permeability field.

58 GEOSCIENCES↗
DOE OSTI2026

Maximizing machine learning interatomic potential transferability for the discovery of the novel stellated octadecagon Bi18-Pt24 cage structure

Achieving true transferability remains the central challenge for Machine Learning Interatomic Potentials (ML-IAPs) in modeling complex bimetallic nanoclusters across their vast potential energy surfaces. We systematically investigate data selection strategies to optimize the Chebyshev Interaction Model for Efficient Simulation (ChIMES) potential for the Bi-Pt nanoclusters by comparing three innovative sampling methods: Principal Component Analysis (PCA)/k-means (structural diversity), t-distributedStochasticNeighborEmbedding (t-SNE)/k-means (force-space diversity), and hierarchical clustering. Quantitatively, the PCA/k-means strategy proved most effective for global accuracy, yielding the lowest force errors and achieving energy root mean square errors (RMSE) values competitive with Density Functional Theory (DFT), demonstrating excellent accuracy (19.16meV/atom). Structural validation on 34 unique DFT-optimized isomers further confirmed the potential’s high fidelity, with the best model PCA/k-means reproducing structures with an average root mean square deviation (RMSD) of 0.10 Å. However, the t-SNE methods, by maximizing diversity in the force space, demonstrated superior extrapolative power, leading to the more precise prediction of a novel stellated octadecagon Bi18⁢Pt24 cage structure, demonstrating the potential for exploring previously unseen morphologies. Our results establish a clear methodology for strategic data sampling that successfully maximizes ML-IAP transferability, providing an accurate and computationally efficient tool that accelerates the theoretical discovery of complex bimetallic architectures.

Vangheluwe, Raphaël [Université Paris-Saclay, CNRS↗
DOE OSTI2022

Sizing ramping reserve using probabilistic solar forecasts: A data-driven method

Ramping products have been introduced or proposed in several U.S. power markets to mitigate the impact of load and renewable uncertainties on market efficiency and reliability. Current methods often rely on historical data to estimate the requirements of ramping products and fail to take into account the effects of the latest weather conditions and their uncertainties, which could lead to overly conservative or insufficient requirements. This study proposes a k-nearest-neighbor-based method to give weather-informed estimates of ramping needs based on short-term probabilistic solar irradiance forecasts. Forecasts from multiple sites are employed in conjunction with principal component analysis to derive numerical classifiers to characterize system-level weather conditions. In addition, we develop a data-driven method to optimize the model parameters in a rolling-forward manner. By using real-world data from the California Independent System Operator, we design two metrics to evaluate method performance: 1) frequency of shortage and 2) oversupply of ramping product. Our proposed method presents advantages in comparison with the baseline and a set of benchmark methods: without compromising system reliability, it reduces system ramping requirements by up to 25%, therefore improving both system reliability and economics.

14 SOLAR ENERGY↗
DOE OSTI2021

Aspects of rare earth element enrichment in Allegheny Plateau coals, Pennsylvania, USA

The Allegheny Plateau in western Pennsylvania, western Maryland, and northern West Virginia has relatively flat-lying strata with southwest-northeast-trending anticlines cored by splay faults originating from the thrust faults and southeast-northwest-trending lineaments. The lineaments define both the structural relief of the anticlines and the coal rank. Coal rank in the Pennsylvanian Allegheny Group coals is medium volatile bituminous in the region of less-prominent structural relief and low volatile bituminous in the region with more pronounced anticlines. Zinc and Cl exhibit high values in coals in the vicinity of the lineaments. Although coal rank was not a parameter in the Principal components analysis, the highest rank coals clustered together in a quadrant defined, in part, by the Zn concentration. Zinc was selected as a parameter due to its possible link with hydrothermal mineralization and Zr and TiO 2 /Al 2 O 3 were selected based on their presumed association with detrital mineral influences. The highest rare earths + Y values in benches of the Lower Kittanning coal were found at sites close to the lineaments. Negative or negligible Ce anomalies (δCe) and strongly positive Gd anomalies (δGd) suggest marine or hydrothermal influences and negative Eu anomalies (δEu) along with high TiO 2 /Al 2 O 3 values suggest a terrestrial input. In conclusion, the plots of δCe vs. δGd and δCe vs. δEu show that the low volatile bituminous coals are segregated from the other coals, suggesting that the factors that influenced coal rank also influenced the distribution of the rare earth elements.

58 GEOSCIENCES↗
DOE OSTI2024

A survey study on arsenic speciation in coal fly ash and insights into the role of coal combustion conditions

Coal fly ashes (CFAs) are the low-density byproducts of the coal combustion process. Improper or uncontrolled CFA disposal poses significant environmental and health concerns due to the potential leaching of toxic heavy metals such as arsenic (As). Previous studies have investigated the content and speciation of As in different CFA samples, yet systematic information on As speciation in CFA with representative coal source and combustion conditions is still missing. Based on a recent survey study on the typical coal sources and combustion conditions across the U.S., this study selected 19 representative CFA samples to systematically investigate As speciation and potential correlations with these parameters. The composition, morphology, mineralogy, and As speciation of these CFA samples were characterized by complementary analytical, microscopic, and spectroscopic techniques. Synchrotron X-ray spectroscopy and microscopy analyses revealed the dominant As oxidation state to be As(V) and with strong associations to Ca, with the exception of 3 samples that had 19–51% As(III), likely due to the use of selective catalytic reduction (SCR) process. Principal component analysis was conducted to identify potential correlations of As concentration and oxidation state with parameters such as major element content, loss on ignition (LOI), average particle size, coal source, and combustion condition. Al 2 O 3 and FeO content were found to capture a majority of the variability. Further, results from this study provide fundamental basis for understanding the correlations between coal source, combustion conditions, CFA characteristics, and As speciation, and providing insights for downstream beneficial utilization or disposal management.

01 COAL, LIGNITE, AND PEAT↗
DOE OSTI2021

Cosolvent enhanced lignocellulosic fractionation tailoring lignin chemistry and enhancing lignin bioconversion

Cosolvent Enhanced Lignocellulosic Fractionation (CELF) is an emerging solvolysis pretreatment to fractionate lignocellulosic biomass. Herein, the bioconversion performance of CELF lignin was fully evaluated for the first time. Results showed that CELF lignin possessed higher content of carboxylic acid OH, lower molecular weight, and disappeared β-O-4 and β-5 linkages compared to other two technical lignins including a conventional ethanol organosolv lignin (EOL) and a kraft lignin (KL). Rhodococcus opacus PD630 cell count from CELF lignin fermentation reached the highest value of 3.9 10 7 CFU/mL, representing a 62.5% and 77.3% improvement over EOL and KL, respectively. Correspondingly, lipid yield reached 143 mg/L from CELF lignin, which was 36.2% and 26.5% higher than from EOL and KL, respectively. Principal component analysis (PCA) revealed that more carboxylic acid groups and lower molecular weight contributed to the enhanced bioconversion performance of CELF lignin. This study demonstrates that CELF lignin is a promising candidate for bioconversion.

09 BIOMASS FUELS↗
DOE OSTI2022

Chemical mixture exposure patterns and obesity among U.S. adults in NHANES 2005–2012

The effect of chemical exposure on obesity has raised great concerns. Real-world chemical exposure always imposes mixture impacts, however their exposure patterns and the corresponding associations with obesity have not been fully evaluated. To discover obesity-related mixed chemical exposure patterns in the general U.S. population. Sparse Decompositional Regression (SDR), a model adapted from sparse representation learning technique, was developed to identify exposure patterns of chemical mixtures with exclusion (non-targeted model) and inclusion (targeted model) of health outcomes. We assessed the relationships between the identified chemical mixture patterns and obesity-related indexes. We also conducted a comprehensive evaluation of this SDR model by comparing to the existing models, including generalized linear regression model (GLM), principal component analysis (PCA), and Bayesian kernel machine regression (BKMR). Eight core exposure patterns were identified using the non-targeted SDR model. Patterns of high levels of MEP, high levels of naphthalene metabolites (ΣOH-Nap), and a pattern of high exposure levels of MCOP, MCNP, and MCPP were positively associated with obesity. Patterns of high levels of BP3, and a pattern of higher mixed levels of MPB, PPB, and MEP were found to have negative associations. Associations were strengthened using the targeted SDR model. In the single chemical analysis by GLM, BP3, MBP, PPB, MCOP, and MCNP showed significant associations with obesity or body indexes. The SDR model exceeded the performance of PCA in pattern identification. Both SDR and BKMR identified a positive contribution of ΣOH-Nap and MCOP, as well as a negative contribution of BP3 and PPB to obesity. Our study identified five core exposure patterns of chemical mixtures significantly associated with obesity using the newly developed SDR model. The SDR model could open a new avenue for assessing health effects of environmental mixture contaminants.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Emulation and detection of physical faults and cyber-attacks on building energy systems through real-time hardware-in-the-loop experiments

The increasing use of remote or mobile access, integrated wearable technologies, data exchange, and cloud-based data analytics in modern smart buildings is steering the building industry towards open communication technologies. The increased connectivity and accessibility could lead to more cyber-attacks in smart buildings. On the other hand, physical faults (e.g., HVAC -heating, ventilation, and air-conditioning faults) may have similar adverse impacts as those from the cyber-attacks on building energy systems, such as occupant discomfort, energy wastage, and equipment downtime. However, current physical behavior-based anomaly detection methods fail to differentiate between cyber-attacks and physical faults in building energy systems. Moreover, the challenge in collecting real-world threat data with ground truth has led researchers to rely on numerical models with user-defined assumptions, which may not accurately reflect real-world conditions due to the lack of in-situ experimental datasets. To address these challenges and gaps, this paper presents a flexible hardware-in-the-loop (HIL) testbed for generating cyber-attack and physical fault datasets and demonstrating threat detection algorithms in a real building automation system (BAS) environment. This testbed combines hardware (i.e., real BAS with local HVAC controllers and a physical network) with software (i.e., high-fidelity models to represent behaviors of building envelope and HVAC energy systems), enabling emulations of realistic threats. Five HIL experiments, including one baseline without any threats, two with physical faults, and two with cyber-attacks, were conducted to generate datasets containing detailed network traffic and system states. A joint classification framework, incorporating a network analyzer and a physical HVAC fault detector, was proposed to automatically detect cyber-physical abnormalities on BAS at both the network and the physical HVAC levels. The network analyzer comprises a conditional random fields (CRF) based command validator and a statistics-based detection strategy. The fault detector employs a weather and schedule-based pattern matching and feature-based principal component analysis (WPM-FPCA) method. Evaluation of the classification using four metrics from the multi-class confusion matrix revealed an average accuracy of 90.2%, recall of 89.7%, precision of 88.5% and F1-score of 89.2%. Finally, these results demonstrate that the proposed joint classification framework can effectively differentiate between specific types of cyber-attacks (e.g., device reinitialization attack, network Denial-of-Service attack) and physical faults (e.g., air handling unit operational fault, cooling coil valve stuck) in real time for improved building energy management.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2020

Data analytics for leak detection in a subcritical boiler

For decades, boiler leaks have been the leading cause of forced outages in the coal-fired unit. The leak occurrences are currently escalating since the existing plants must satisfy faster-ramping rates to support grid operation. Data analytics including Principal Component Analysis, Canonical Variate, and Fisher Discriminant Analysis were combined for detecting and characterizing the leak in a commercial 650 MW subcritical coal-fired power plant. The combined approach was shown to be highly effective in the fault investigation that would not have been easily achieved by an individual technique. The variability in both training and validation datasets was first evaluated using PCA. Then, the CV-FDA was employed to discriminate among faults, and to categorize the processed data into two main groups: no-leak (0) and leak (1), providing the timeframe and location of the leak occurrence. Furthermore, about 8,014 observations from 81 process variables were initially included in the calculation, while the variable count was reduced to 4 with less than 1% misclassification rate in total observations. Finally, the leak was isolated in the waterwall section. Thus, the outcome of this research may provide early detection and isolation of faulty operations in the coal-fired power plant that involves a considerable number of process variables.

20 FOSSIL-FUELED POWER PLANTS↗
DOE OSTI2021

Distribution of rare earth elements and other critical elements in beneficiated Pennsylvania anthracites

The Pennsylvania Anthracite Fields are in a complex tectonic and metamorphic terrain, historically hosting one of the largest concentrations of coal mining in the USA. Anthracite mining now largely consists of the surface mining of the pillars remaining from the prime years of underground mining. The geochemical study of the sized coal products and the refuse (largely rock) from three preparation plants (breakers) demonstrates that Principal components analysis (PCA) of select major oxide, minor element, and rare earth elements illustrates some differentiation among the products from the individual plants. The rock samples, with abundant quartz and metamorphic Al-Si minerals and with a lower ash-basis REE concentration than the coals, were distinctly separated from the coal samples on the PCA plots. Plots of Gd N /Gd N * vs. Eu N /Eu N * and Ce N /Ce N * vs. Eu N / Eu N * showed differentiation between the individual suites of coals showed that the refuse samples had distinct REE distributions compared to the associated coals. Further, several minor and trace elements show enrichments in the coal samples. Lithium, with concentrations of up to 314 ppm on an ash basis, is among the most promising of the critical elements, exceeding the enrichment of the REY and Sc.

58 GEOSCIENCES↗
DOE OSTI2022

Past fires and post-fire impacts reconstructed from a southwest Australian stalagmite

Speleothem records of past environmental change provide an important opportunity to explore fire frequency and intensity in the past, and the antecedent climatic conditions leading to fire events. Here, fire sensitive geochemical signals in a stalagmite from Yonderup Cave, a shallow cave in Western Australia, are compared to well-documented wildfire events that occurred in recent decades. The results are extended to identify wildfires during the growth interval of the speleothem (1760 CE–2005 CE). Principal component analysis of the stalagmite time-series revealed distinct peaks in a combination of phosphorus and metal (aluminium, zinc, copper and lead) concentrations in response to known fire events, which are interpreted to have come from ash. Varying responses in the geochemical signal in the pre- and post-European colonisation period are likely linked to changes in land management, fire frequency and fire intensities. Of note is what we infer to be a particularly intense fire event concluded to have occurred in 1897 ± 5 CE, which climate sensitive proxies (δ18O and bedrock-derived elements) indicate was preceded by a multi-decadal dry period that began in the late 1860s. The intensity of the deduced fire event is supported by a peak in P that is 6.3 times higher than elsewhere in the speleothem, accompanied by a peak in Zn. It was also inferred that changes to the surface-cave hydrology occurred as a result of this event, indicated by increased input of colloidal organic material onto the stalagmite that was interpreted to be caused by shallow karst bedrock fracturing from the heat-induced deformation that is observed after intense fires. These findings show the potential for speleothems to provide dated records of fire intensity and recurrence intervals. Further development could lead to a better understanding of the climate-fire relationship and the effects of land-management practices on wildfire frequency and intensity.

58 GEOSCIENCES↗
DOE OSTI2025

Optical vibrational spectroscopic signatures of ammonium diuranate process parameters

Ammonium diuranate (ADU) is commonly encountered in the nuclear fuel cycle; however, previous investigations have shown that ADU is a complex mixture of distinct compounds. Moreover, production parameters are known to heavily influence the composition of the resulting ADU. Here, we examine four samples of ADU prepared at Oak Ridge National Laboratory (ORNL), and one sample of ADU made at Pacific Northwest National Laboratory (PNNL), with the goal of further characterizing and elucidating the effect of processing parameters such as stir rate, strike direction, and temperature on material composition. Process parameters during ADU precipitation at ORNL and PNNL were well documented, and we relate process variables to optical vibrational spectroscopic signatures observed using Raman and infrared (IR) spectroscopy. In addition, powder X-ray diffraction (PXRD) reveals differences in the solid-phase composition of ADU precipitates, but we find that the primary phase is similar to the uranyl oxyhydroxyhydrate mineral metaschoepite. Despite the significant phase contributions of a metaschoepite-like phase, spectroscopic evidence of both nitrate and ammonium are observed for all samples. To gain a more holistic understanding of spectroscopic features of process parameters in ADU, principal component analysis (PCA) is employed and results in observable signatures that relate to the stir rate used during synthesis. These results provide further information about the process-dependence of ADU precipitate composition.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2021

Aerosol emissions from water-lean solvents for post-combustion CO 2 capture

Advanced water-lean solvents (WLS) for post-combustion CO 2 capture have been gaining interest due to their ability to reduce the parasitic penalty from energy needed for solvent regeneration. Commercial implementation of these novel CO 2 capture technologies hinges on successful control of amine emissions. RTI conducted a parametric study of fundamental and operational variables influence on overall amine aerosol and vapor emissions from our water-lean solvent eCO 2 Sol™ using our 6-kW equivalent bench-scale gas absorption system. The parametric testing used a simulated flue gas with 15 % CO 2 , 2.3–4.2 % H 2 O, and 0–6 ppm sulfite (SO 3 ) to examine the impact of the presence of aerosols to the capture performance and amine emissions from the system. The SO 3 reacts with water in the flue gas to create H 2 SO 4 , which forms liquid aerosol droplets and provide nucleation sites for growth of aerosols. Scanning Mobility Particle Sizer and Aerodynamic Particle Sizer instruments monitored the aerosol particle size distribution. Parametric testing results suggested that the presence of the aerosols in the flue gas could increase the overall amine emissions by 10X compared to the baseline emissions from WLS’s vapor pressure. Principal component analysis (PCA) and projection to latent squares (PLS) developed models to predict the aerosol-based amine emissions from process data. The predictive PLS model had a correlation coefficient (Q 2 ) of 0.92 and could predict the aerosol-based emissions from the NAS process with ±15 % accuracy (average absolute deviation, AAD). The PLS regression model also identified key variables affecting aerosol-based emissions from WLS.

42 ENGINEERING↗
DOE OSTI2021

A physics-informed and hierarchically regularized data-driven model for predicting fluid flow through porous media

This paper presents a new deep learning data-driven model for predicting structure dependent pore-fluid velocity fields in rock. The model is based on a Convolutional Auto-Encoder (CAE) artificial neural network capable of learning from image data generated by direct numerical simulations of fluid flow through pore-structures, such as by Lattice Boltzmann or molecular dynamics methods. The main novelty of the model in comparison to previous CAE-based data-driven approaches consists of three parts. The first is a methodology for decomposing the full-domain of the porous media into sub-regions, or “sub-domains”, in order to reduce the overall size of the CAE, batch process the sub-domains in parallel, and enable the CAE to learn local and generalizable nonlinear mappings of pore-fluid velocities. The second consists of embedding the finite difference solutions of the incompressible Navier-Stokes and continuity equations into convolutional layers prior to the CAE in order to provide the CAE with knowledge of fluid dynamics physics (PhyFlow). The third main novelty is that the training of the CAE is regularized with a hierarchical loss function that encourages the learning of fluid flow patterns (in a way similar to ranked modes in principal component analysis), ranking from most to least important. This is shown to increase the stability in learning, reduce over-fitting, and promote interpretability of the CAE neural network layers (HierCAE). The comprehensive new data-driven model, which we call the PhyFlow-HierCAE model, is shown to exhibit improved accuracy and generalizability of flow field predictions over conventional CAE models, attributable to the embedded physical knowledge and the hierarchical regularization, as well as realize orders of magnitude speed-ups in computation times as a surrogate for the direct numerical simulations. Examples of training and forward predictions on unseen pore-structures are provided and evaluated for data from Lattice Boltzmann and molecular dynamics simulations of pore-fluid flow. The model is shown to be a fast and accurate emulator (or “surrogate”) for predicting effective permeability of unseen pore-structures based on learning from relatively small direct numerical simulation datasets.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2021

Classification and regression models of audio and vibration signals for machine state monitoring in precision machining systems

Here we present a data-driven method for monitoring machine status in manufacturing processes. Audio and vibration data from precision machining are used for inference in two operating scenarios: (a) variable machine health states (anomaly detection); and (b) settings of machine operation (state estimation). Audio and vibration signals are first processed through Fast Fourier Transform and Principal Component Analysis to extract transformed and informative features. These features are then used in the training of classification and regression models for machine state monitoring. Specifically, three classifiers (K-nearest neighbors, convolutional neural networks and support vector machines) and two regressors (support vector regression and neural network regression) were explored, in terms of their accuracy in machine state prediction. It is shown that the audio and vibration signals are sufficiently rich in information about the machine that 100% state classification accuracy could be accomplished. Data fusion was also explored, showing overall superior accuracy of data-driven regression models.

42 ENGINEERING↗
DOE OSTI2024

Characterization of Fuel Cladding Chemical Interaction on a High Burnup U-10Zr Metallic Fuel via Electron Energy Loss Spectroscopy Enhanced by Machine Learning

Fuel cladding chemical interaction (FCCI) is one of the main performance limiting factors for metallic nuclear fuels. The interaction destabilizes the martensitic microstructure and deteriorates mechanical properties of HT-9 cladding. The detection of low atomic number elements (Z<10) and overlapping of elemental peaks can be problematic in interpreting energy dispersive X-ray spectroscopy (EDS) data. Electron energy loss spectroscopy (EELS) provides precise elemental edge energy values and can detect elements with a low atomic number. This work utilizes EELS to study the distribution of lanthanides and light elements at the interaction region. The sample was prepared from the FCCI region of a U-10Zr (wt.%) solid fuel with HT-9 cladding, irradiated to a burnup of 13.2 at.%. Processing the EELS data included three major steps: 1) enhance the signal to noise ratio by denoising the spectrum with principal component analysis (PCA) method, removing background and performing deconvolution; 2) identify chemical elements with core energy loss edges; 3) confirm different phases using a popular machine learning method, K-means. This work presents qualitative assessment of lanthanides and light elements like carbon (C) and oxygen (O) enhanced by the application of machine learning algorithms. By comparing with EDS elemental maps, EELS provides higher resolution chemical maps, reveals the distribution of carbon at the interaction region supporting the formation of zirconium carbide, a rind-like microstructure feature that was proposed to mitigate the chemical interaction. Furthermore, the plasmon peak map was also found to indicate an energy shift associated with the formation of phases/compounds. K-means clustering method was used on the processed electron energy loss (EEL) spectrum to automatically reveal different phases. The resulting clustered maps from K-means clustering align well with elemental maps confirming certain phases, especially Fe-Ce and Zr-C, in the FCCI region.

EELS↗