Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “classification”

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 775 records · Page 43
All sourcesNASA NTRSDOE OSTI
DOE OSTI2017

Diagnosing Cloud Biases in the GFDL AM3 Model With Atmospheric Classification

In this paper, we define a set of 21 atmospheric states, or recurring weather patterns, for a region surrounding the Atmospheric Radiation Measurement Program's Southern Great Plains site using an iterative clustering technique. The states are defined using dynamic and thermodynamic variables from reanalysis, tested for statistical significance with cloud radar data from the Southern Great Plains site, and are determined every 6 h for 14 years, creating a time series of atmospheric state. The states represent the various stages of the progression of synoptic systems through the region (e.g., warm fronts, warm sectors, cold fronts, cold northerly advection, and high-pressure anticyclones) with a subset of states representing summertime conditions with varying degrees of convective activity. We use the states to classify output from the NOAA/Geophysical Fluid Dynamics Laboratory AM3 model to test the model's simulation of the frequency of occurrence of the states and of the cloud occurrence during each state. The model roughly simulates the frequency of occurrence of the states but exhibits systematic cloud occurrence biases. Comparison of observed and model-simulated International Satellite Cloud Climatology Project histograms of cloud top pressure and optical thickness shows that the model lacks high thin cloud under all conditions, but biases in thick cloud are state-dependent. Frontal conditions in the model do not produce enough thick cloud, while fair-weather conditions produce too much. Finally, we find that increasing the horizontal resolution of the model improves the representation of thick clouds under all conditions but has little effect on high thin clouds. However, increasing resolution also changes the distribution of states, causing an increase in total cloud occurrence bias.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2022

ThunderSecure: deploying real-time intrusion detection for 100G research networks by leveraging stream-based features and one-class classification network

Nowadays, data generated by large-scale scientific experiments are on the scale of petabytes per month. These data are transferred through dedicated high-bandwidth networks (40/100G) across distributed sites for processing, storage, and analysis. Like general purpose networks, research networks experience intrusions. However, monitoring anomalies in such high-speed network traffics is challenging given current cyber-infrastructure. Moreover, traditional network intrusion detection systems (NIDS) are signature based. However, anomaly patterns are difficult to define and that rulesets are often not updated frequently enough to reflect the changes of attack behaviors. We present ThunderSecure, a high-throughput, unsupervised learning-based intrusions detection system for 100G research networks. ThunderSecure implements an efficient packet processing and detection pipeline using multi-cores and GPUs. It extracts statistical and temporal features from real-time network data streams and feeds them to a one-class anomaly detection network. A baseline of normal distribution will be created based on the training observation. Testing traffic deviated from the learned profile will be marked as anomalies. We trained ThunderSecure on hundreds of billions of science data packets mirrored from two 100G network connections at Fermi National Accelerator Laboratory. The detection performance was evaluated on traffic captured from the same research network days and weeks after the training with different types of attack flows injected. Results show that ThunderSecure can recognize science data traffic captured long after the training and made nearly certain detection on the segment of the streams where anomalous flows were injected.

100G research network↗
DOE OSTI2018

Automated Classification and Analysis of Non-metallic Inclusion Data Sets

The aim of this study is to utilize principal component analysis (PCA), clustering methods, and correlation analysis to condense and examine large, multivariate data sets produced from automated analysis of non-metallic inclusions. Non-metallic inclusions play a major role in defining the properties of steel and their examination has been greatly aided by automated analysis in scanning electron microscopes equipped with energy dispersive X-ray spectroscopy. The methods were applied to analyze inclusions on two sets of samples: two laboratory-scale samples and four industrial samples from a near-finished 4140 alloy steel components with varying machinability. The laboratory samples had well-defined inclusions chemistries, composed of MgO-Al{sub 2}O{sub 3}-CaO, spinel (MgO-Al{sub 2}O{sub 3}), and calcium aluminate inclusions. The industrial samples contained MnS inclusions as well as (Ca,Mn)S + calcium aluminate oxide inclusions. PCA could be used to reduce inclusion chemistry variables to a 2D plot, which revealed inclusion chemistry groupings in the samples. Clustering methods were used to automatically classify inclusion chemistry measurements into groups, i.e., no user-defined rules were required.

36 MATERIALS SCIENCE↗
DOE OSTI2019

Existence and classification of pseudo-asymptotic solutions for Tolman–Oppenheimer–Volkoff systems

The Tolman–Oppenheimer–Volkoff (TOV) equations are a partially uncoupled system of nonlinear and non-autonomous ordinary differential equations which describe the structure of isotropic spherically symmetric static fluids. Nonlinearity makes finding explicit solutions of TOV systems very difficult and such solutions are very rare. In this paper we introduce the notion of pseudo-asymptotic TOV systems and we show that the space of such systems is at least fifteen-dimensional. We also show that if the system is defined in a suitable domain (meaning the extended real line), then well-behaved pseudo-asymptotic TOV systems are genuine TOV systems in that domain, ensuring the existence of new fourteen analytic solutions for extended TOV equations. The pseudo-asymptotic solutions are classified according to the nature of the matter (ordinary or exotic) and to the existence of cavities and singularities. It is shown that at least three of them are realistic, in the sense that they are formed only by ordinary matter and contain no cavities or singularities.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2022

Classification of animal sounds in a hyperdiverse rainforest using convolutional neural networks with data augmentation

To protect tropical forest biodiversity, we need to be able to detect it reliably, cheaply, and at scale. Automated detection of sound producing animals from passively recorded soundscapes via machine-learning approaches is a promising technique towards this goal, but it is constrained by the necessity of large training data sets. Using soundscapes from a tropical forest in Borneo and a Convolutional Neural Network model (CNN), we investigate i) the minimum viable training data set size for accurate prediction of call types (‘sonotypes’), and ii) the extent to which data augmentation and transfer learning can overcome the issue of small and imbalanced training data sets. We found that even relatively high sample sizes (>80 per sonotype) lead to mediocre accuracy, which however improved significantly with data augmentation and transfer learning, including at extremely small sample sizes (3 per sonotype), regardless of taxonomic group or call characteristics. Neither transfer learning nor data augmentation alone achieved high accuracy. Our results suggest that transfer learning and data augmentation could make the use of CNNs to classify species’ vocalizations feasible even for small soundscape-based projects with many rare species. Retraining our open-source model requires only basic programming skills which makes it possible for individual conservation initiatives to match their local context, in order to enable more evidence-informed management of biodiversity.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2018

Classification of machine learning frameworks for data-driven thermal fluid models

Thermal fluid processes are inherently multi-physics and multi-scale, involving mass-momentum-energy transport phenomena at multiple scales. Thermal fluid simulation (TFS) is based on solving conservative equations, for which – except for “first-principles” direct numerical simulation – closure relations (CRs) are required to provide microscopic interactions or so-called sub-grid-scale physics. In practice, TFS is realized through reduced-order modeling, and its CRs as low-fidelity models can be informed by observations and data from relevant and adequately evaluated experiments and high-fidelity simulations. This paper is focused on data-driven TFS models, specifically on their development using machine learning (ML). Five ML frameworks are introduced including physics-separated ML (PSML or Type I ML), physics-evaluated ML (PEML or Type II ML), physics-integrated ML (PIML or Type III ML), physics-recovered (PRML or Type IV ML), and physics-discovered ML (PDML or Type V ML). The frameworks vary in their performance for different applications depending on the level of knowledge of governing physics, source, type, amount and quality of available data for training. Notably, outlined for the first time in this paper, Type III models present stringent requirements on modeling, substantial computing resources for training, and high potential in extracting value from “big data” in thermal fluid research. Here the current paper demonstrates and investigates ML frameworks in three examples. First, we utilize the heat diffusion equation with a nonlinear conductivity model formulated by convolutional neural networks (CNNs) and feedforward neural networks (FNNs) to illustrate the applications of Type I, Type II, Type III, and Type V ML. The results indicate a preference for Type II ML under deficient data support. Type III ML can effectively utilize field data, potentially generating more robust predictions than Type I and Type II ML. CNN-based closures exhibit more predictability than FNN-based closures, but CNN-based closures require more training data to obtain accurate predictions. Second, we illustrate how to employ Type I ML and Type II ML frameworks for data-driven turbulence modeling using reference works. Third, we demonstrate Type I ML by building a deep FNN-based slip closure for two-phase flow modeling. The results show that deep FNN-based closures exhibit a bounded error in the prediction domain.

42 ENGINEERING↗
DOE OSTI2016

Precipitation regime classification for the Mojave Desert: Implications for fire occurrence

Mojave Desert ecosystem processes are dependent upon the amount and seasonality of precipitation. Multi-decadal periods of drought or above-average rainfall affect landscape vegetation condition, biomass and susceptibility to fire. The seasonality of precipitation events can also affect the likelihood of lightning, a key ignition source for fires. To develop an understanding of precipitation regimes and fire patterns we used monthly average precipitation data and GIS data representing burned areas from 1971-2010. We applied a K-means cluster analysis to the monthly precipitation data identifying three distinct precipitation seasons; winter (October – March), spring (April-June) and summer (July-September) and four discrete precipitation regimes within the Mojave ecoregion.

classification↗
DOE OSTI2013

Geometric classification of open-cell metal foams using X-ray micro-computed tomography

The geometry of foams has long been an area of interest, and a number of idealized geometric descriptions have been proposed. In order to acquire detailed, quantitative, geometric data for aluminum open-cell metal foams, X-ray {mu}CT is employed. The X-ray {mu}CT images are analyzed using specialized software, FoamView Registered-Sign , from which geometric information including strut length and pore shapes are extracted. The X-ray {mu}CT analysis allows comparison of the ideal geometric models to the actual geometric characteristics of the metal foam samples. The results reveal a high variability in ligament length, as well as features supporting the ideal geometry known as the Weaire-Phelan unit cell. The geometric findings provide information useful for improving current models of open-cell metal foam. Applications can range from predicting heat transfer or load failure to predicting liquid retention. Highlights: Black-Right-Pointing-Pointer Aluminum open-cell metal foams are geometrically classified Black-Right-Pointing-Pointer X-ray micro-computed tomography and specialized software are used to gather geometric data Black-Right-Pointing-Pointer The foams are shown to have a high variability in strut length Black-Right-Pointing-Pointer The Weaire-Phelan unit cell is shown to be a better representative of these foams.

36 MATERIALS SCIENCE↗
DOE OSTI2014

Classification of a Haemophilus influenzae ABC Transporter HI1470/71 through Its Cognate Molybdate Periplasmic Binding Protein, MolA

molA (HI1472) from H. influenzae encodes a periplasmic binding protein (PBP) that delivers substrate to the ABC transporter MolB{sub 2}C{sub 2} (formerly HI1470/71). The structures of MolA with molybdate and tungstate in the binding pocket were solved to 1.6 and 1.7 {angstrom} resolution, respectively. The MolA-binding protein binds molybdate and tungstate, but not other oxyanions such as sulfate and phosphate, making it the first class III molybdate-binding protein structurally solved. The {approx}100 {mu}M binding affinity for tungstate and molybdate is significantly lower than observed for the class II ModA molybdate-binding proteins that have nanomolar to low micromolar affinity for molybdate. The presence of two molybdate loci in H. influenzae suggests multiple transport systems for one substrate, with molABC constituting a low-affinity molybdate locus.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2014

Morphological classification of bioaerosols from composting using scanning electron microscopy

Highlights: • Bioaerosols were captured using the filter method. • Bioaerosols were analysed using scanning electron microscope. • Bioaerosols were classified on the basis of morphology. • Single small cells were found more frequently than aggregates and larger cells. • Smaller cells may disperse further than heavier aggregate structures. - Abstract: This research classifies the physical morphology (form and structure) of bioaerosols emitted from open windrow composting. Aggregation state, shape and size of the particles captured are reported alongside the implications for bioaerosol dispersal after release. Bioaerosol sampling took place at a composting facility using personal air filter samplers. Samples were analysed using scanning electron microscopy. Particles were released mainly as small (<1 μm) single, spherical cells, followed by larger (>1 μm) single cells, with aggregates occurring in smaller proportions. Most aggregates consisted of clusters of 2–3 particles as opposed to chains, and were <10 μm in size. No cells were attached to soil debris or wood particles. These small single cells or small aggregates are more likely to disperse further downwind from source, and cell viability may be reduced due to increased exposure to environmental factors.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2024

A Novel Method to Train Classification Models for Structure Detection in In Situ Spacecraft Data

We present a method for creating spacecraft-like data which can be used to train Machine Learning (ML) models to detect and classify structures in in situ spacecraft data. First, we use the Grad-Shafranov equation to numerically solve for several magnetohydrostatic equilibria which are variations on a known analytic equilibrium. These equilibria are then used as the initial conditions for Particle-In-Cell simulations in which the structures of interest are observed and labeled. We then take one-dimensional slices through the simulations to replicate what a spacecraft collecting data from the simulation would observe. This sliced data then can be used as training data for the initial training of ML models intended for use on spacecraft data. We demonstrate the method applied to the problem of detecting small-scale plasmoids in the magnetotail, which is important for understanding complex magnetotail reconnection dynamics. The simple 1D classifier we train is able to detect more than 70% of the plasmoid points in the data set but also produces a large number of false positives. Our further work on this example problem is detailed, and further potential uses of the method are discussed.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2010

Classification and stability of plasma motion in periodic linearly polarized relativistic waves

Based on a relativistic fluid-Maxwell model, laser-induced plasma dynamics is investigated for relativistic periodic waves. Within a one-dimensional (1D) description, the Akhiezer-Polovin model is applied to the existence of periodic, nonlinearly coupled electromagnetic and electrostatic waves, and the corresponding particle motion. Known existence criteria for periodic solutions are generalized. The corresponding stability behaviors are investigated by 1D integrators of the relativistic fluid-Maxwell model. It is shown that in contrast to the vacuum solution, linearly polarized coupled electromagnetic-electrostatic waves are unstable in plasmas. The magnitudes of the growth rates are investigated in terms of the maximum amplitudes and normalized phase velocities.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2010

Automated Classification of Disease Patterns from Echo-cardiography Images Based on Shape Features of the Left Ventricle

Computer assisted diagnosis using analysis of medical images is an area of active research in health informatics. This paper proposes a technique for indication of heart diseases by using information related to shapes of the left ventricle (LV). LV boundaries are tracked from echo-cardiography images taken from LV short axis view, corresponding to two disease conditions viz. dilated cardiomyopathy and hypertrophic cardiomyopathy, and discriminated from the normal condition. The LV shapes are modeled using shape histograms generated by plotting the frequency of normalized radii lengths drawn from the centroid to the periphery, against a specific number of bins. A 3-layer neural network activated by a log-sigmoid function is used to classify the shape histograms into one of the three classes. Experimentations on a dataset of 240 images show recognition accuracies of the order of 80%.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2012

Noncommutative Pfaffians and classification of states of five-dimensional quasi-spin

Noncommutative Pfaffians associated with an orthogonal algebra o{sub N} are some special elements of the universal enveloping algebra U(o{sub N}). Using Pfaffians we construct the fourth quantum number which together with the naturally defined three quantum numbers allows to classify the states of a five-dimensional quasi-spin. The Pfaffians are treated as creation operators for the new quantum number.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗