SEARCH · Engineering Papers
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.
This study benchmarks several FSS techniques and reports several studies where a large set metrics were reduced to a handful with little loss of detection accuracy. This result raises the possibility that software defect detection may be much simpler than previously believed.
Most work on preference learning has focused on pairwise preferences or rankings over individual items. In this paper, we present a method for learning preferences over sets of items. Our learning method takes as input a collection of positive examples--that is, one or more sets that have been identified by a user as desirable. Kernel density estimation is used to estimate the value function for individual items, and the desired set diversity is estimated from the average set diversity observed in the collection. Since this is a new learning problem, we introduce a new evaluation methodology and evaluate the learning method on two data collections: synthetic blocks-world data and a new real-world music data collection that we have gathered.
A requirement for effective aircraft engine performance estimation is the ability to account for engine degradation, generally described in terms of unmeasurable health parameters such as efficiencies and flow capacities related to each major engine module. This paper presents a linear point design methodology for minimizing the degradation-induced error in model-based aircraft engine performance estimation applications. The technique specifically focuses on the underdetermined estimation problem, where there are more unknown health parameters than available sensor measurements. A condition for Kalman filter-based estimation is that the number of health parameters estimated cannot exceed the number of sensed measurements. In this paper, the estimated health parameter vector will be replaced by a reduced order tuner vector whose dimension is equivalent to the sensed measurement vector. The reduced order tuner vector is systematically selected to minimize the theoretical mean squared estimation error of a maximum a posteriori estimator formulation. This paper derives theoretical estimation errors at steady-state operating conditions, and presents the tuner selection routine applied to minimize these values. Results from the application of the technique to an aircraft engine simulation are presented and compared to the estimation accuracy achieved through conventional maximum a posteriori and Kalman filter estimation approaches. Maximum a posteriori estimation results demonstrate that reduced order tuning parameter vectors can be found that approximate the accuracy of estimating all health parameters directly. Kalman filter estimation results based on the same reduced order tuning parameter vectors demonstrate that significantly improved estimation accuracy can be achieved over the conventional approach of selecting a subset of health parameters to serve as the tuner vector. However, additional development is necessary to fully extend the methodology to Kalman filter-based estimation applications.
AutoChem is a suite of Fortran 90 computer programs for the modeling of kinetic reaction systems. AutoChem performs automatic code generation, symbolic differentiation, analysis, and documentation. It produces a documented stand-alone system for the modeling and assimilation of atmospheric chemistry. Given databases of chemical reactions and a list of constituents defined by the user, AutoChem automatically does the following: 1. Selects the subset of reactions that involve a user-defined list of constituents and automatically prepares a document listing the reactions; 2. Constructs the ordinary differential equations (ODEs) that describe the reactions as functions of time and prepares a document containing the ODEs; 3. Symbolically differentiates the time derivatives to obtain the Jacobian and prepares a document containing the Jacobian; 4. Symbolically differentiates the Jacobian to obtain the Hessian and prepares a document containing the Hessian; and 5. Writes all the required Fortran 90 code and datafiles for a stand-alone chemical modeling and assimilation system (implementation of steps 1 through 5). Typically, the time taken for steps 1 through 5 is about 3 seconds. The modeling system includes diagnostic components that automatically analyze each ODE at run time, the relative importance of each term, time scales, and other attributes of the model.
Several feature extraction and selection methods for an existing automatic target recognition (ATR) system using JPLs Grayscale Optical Correlator (GOC) and Optimal Trade-Off Maximum Average Correlation Height (OT-MACH) filter were tested using MATLAB. The ATR system is composed of three stages: a cursory region of-interest (ROI) search using the GOC and OT-MACH filter, a feature extraction and selection stage, and a final classification stage. Feature extraction and selection concerns transforming potential target data into more useful forms as well as selecting important subsets of that data which may aide in detection and classification. The strategies tested were built around two popular extraction methods: Principal Component Analysis (PCA) and Independent Component Analysis (ICA). Performance was measured based on the classification accuracy and free-response receiver operating characteristic (FROC) output of a support vector machine(SVM) and a neural net (NN) classifier.
The Efficient Descent Advisor (EDA) controller automation tool generates trajectory-based speed, path, and altitude-profile advisories to facilitate efficient, continuous descents into congested terminal airspace. While prior field trials have assessed the trajectory prediction accuracy for large jet (i.e., Boeing and Airbus) types, smaller (i.e., regional and business) jet types present unique challenges involving different descent procedures and Flight Management System (FMS) capabilities. This paper quantifies the trajectory prediction accuracy for small jet revenue flight descents based on SkyWest Canadair Regional Jet 200, 700, and 900 aircraft arrivals to Denver in the fall of 2010. Post flight test data analysis and SkyWest pilot interviews uncovered unexpected variation between flight crews due to different interpretations of (1) which fixed flight path angle (FPA) to fly based on the flight trial procedure and (2) how to fly the descent to achieve the target FPA. Pilot reports were used to select a subset of flights where pilots indicated an FPA according to the flight trial procedure to remove the unexpected variation due to (1) to focus on (2). Results for the subset for en-route descents, from prior to top of descent to the meter fix 30 to 130 nmi downstream, indicate that aircraft arrived to the meter fix six seconds early with about a twelve second standard deviation. Large FPA errors up to one degree relative to the EDA flight trial procedure were detected after the flight trial as a characteristic of the unexpected variation. It is recommended that quantitative validation be performed during future flight trials so that experimental procedures can be adjusted if unexpected results are detected.
Hyper-spectral remote sensors, such as Atmospheric Infrared Sounder (AIRS) and Infrared Atmospheric Sounding Interferometer (IASI), provides top of atmospheric radiances with high information content on atmospheric and surface properties. In order to analyze these data in real time, fast, and accurate forward and inverse models are needed. We will describe a method of simultaneously retrieving atmospheric temperature, moisture, cloud, and surface properties using all available spectral channels without sacrificing computational speed. The method has been successfully applied to AIRS, IASI, and NAST-I data. By applying the same method to the current and future hyperspectral sounders, the derived products will have less errors due to biases introduced by different retrieval methods. We have compared the retrieved products to radiosondes, aircraft measurements and other validation dataset. Simulations have done to quantify retrieval errors associated with the retrievals. The essence of the method is to convert channel radiance spectra into super-channels by an Empirical Orthogonal Function (EOF) transformation. A Principal Component-based Radiative Transfer Model (PCRTM) developed at NASA Langley Research Center is used to calculate both the super-channel magnitudes and derivatives with respect to atmospheric profiles and other properties. The inversion algorithm is based on a non-linear Levenberg-Marquardt method with climatology covariance matrices and a priori information as constraints. One advantage of this approach is that it uses all information content from the hyper-spectral data so that the retrieval is less sensitive to instrument noise and eliminates the need for selecting a subset of the channels. The PCRTM forward model has also been used to performance end-to-end sensor performance simulations of the Climate Absolute Radiance and Refractivity Observatory (CLARREO).
This project developed tools to support the future use of Plankton, Aerosol, Cloud, ocean Ecosystem (PACE) hyperspectral imagery in water resource monitoring and research by NASA DEVELOP teams and members of the PACE applications community. We sought to address a need for support in processing and visualizing hyperspectral PACE Ocean Color Instrument (OCI) data among researchers and decision-makers working in coastal water quality management and harmful algal bloom (HAB) monitoring. To supplement the day of simulated PACE imagery available, we used Aqua MODIS earth observations with Level 3 processing from March 2022 to build a Python graphical user interface (GUI) for visualizing ocean biogeochemical parameters relevant to the early detection and monitoring of HABs. We used simulated PACE OCI Level 2 data derived from the Python Top of Atmosphere Simulation Tool (PyTOAST) to build Jupyter Notebooks for band subset and selection. The Level 3 PACE Viewer components support users with quick visualizations as well as the creation of geoTIFFs and time-series. The Level 2 Jupyter Notebooks address users’ concerns over the volume and complexity of hyperspectral imagery. The PACE Viewer is useful for visual inspection and netCDF data processing but should not be used for geospatial analysis. Once PACE launches, this tool will alleviate the technical burdens of working with hyperspectral data and support the early detection and monitoring of HABs using PACE satellite imagery.
Multitemporally registered LANDSAT MSS data from four acquisitions during the 1978 growing season were used in classification of eight sample segments in Iowa and Indiana. The results illustrate that use of LANDSAT acquisition when corn has tasseled is critical, as this is the optimum time for separation of corn and soybeans. An early season acquisition when the summer crops appear as bare soil can be beneficial in reducing the confusion between these two crops and other cover types. A subset of one visible and one infrared band from each date was found to produce results not significantly different from the use of all bands. Selection of a subset of these bands may also be feasible for multitemporal analysis.
The use of spectral attributes criteria was investigated, based on measures of statistical distance of separability between thematic classes in MSS digital LANDSAT imagery, in order to select the best subsets of channels in composite colors for the detection and discrimination of lithological units in the lower valley of Curaca River, State of Bahia, Brazil. Three situations were investigated: (1) selection of the three best channels, considering all of the original bands (channels 4, 5, 6, and 7); (2) selection of the three best bands, considering the six MSS band-ratios (channels 4/5, 4/6. 4/7, 5/6, 5/7, and 6/7); and (3) selection of the three best bands in a hybrid approach (the four original bands and the six ratios). A visual analysis was done on color composite images using the selected sets. Results show that the hybrid product (bands 4, 5/7, and 7 with green, blue, and red respectively) and the Normal Color Composite (bands 4, 5, and 7 with blue, green, and red colors respectively) had the best performance.
With the advent of the latest human spaceflight objectives, a series of prototype architectures for a new launch and reentry spacesuit that would be suited to the new mission goals. Four prototype suits were evaluated to compare their performance and enable the selection of the preferred suit components and designs. A consolidated approach to testing was taken: concurrently collecting suit mobility data, seat-suit-vehicle interface clearances, and qualitative assessments of suit performance within the volume of a Multi-Purpose Crew Vehicle mockup. It was necessary to maintain high fidelity in a mockup and use advanced motion-capture technologies in order to achieve the objectives of the study. These seemingly mutually exclusive goals were accommodated with the construction of an optically transparent and fully adjustable frame mockup. The construction of the mockup was such that it could be dimensionally validated rapidly with the motioncapture system. This paper describes the method used to create a space vehicle mockup compatible with use of an optical motion-capture system, the consolidated approach for evaluating spacesuits in action, and a way to use the complex data set resulting from a limited number of test subjects to generate hardware requirements for an entire population. Kinematics, hardware clearance, anthropometry (suited and unsuited), and subjective feedback data were recorded on 15 unsuited and 5 suited subjects. Unsuited subjects were selected chiefly based on their anthropometry in an attempt to find subjects who fell within predefined criteria for medium male, large male, and small female subjects. The suited subjects were selected as a subset of the unsuited medium male subjects and were tested in both unpressurized and pressurized conditions. The prototype spacesuits were each fabricated in a single size to accommodate an approximately average-sized male, so select findings from the suit testing were systematically extrapolated to the extremes of the population to anticipate likely problem areas. This extrapolation was achieved by first comparing suited subjects performance with their unsuited performance, and then applying the results to the entire range of the population. The use of a transparent space vehicle mockup enabled the collection of large amounts of data during human-in-the-loop testing. Mobility data revealed that most of the tested spacesuits had sufficient ranges of motion for the selected tasks to be performed successfully. A suited subject's inability to perform a task most often stemmed from a combination of poor field of view in a seated position, poor dexterity of the pressurized gloves, or from suit/vehicle interface issues. Seat ingress and egress testing showed that problems with anthropometric accommodation did not exclusively occur with the largest or smallest subjects, but also with specific combinations of measurements that led to narrower seat ingress/egress clearance.
Recent Large Language Models (LLMs) have demonstrated impressive capabilities at tasks that require human intelligence and are a significant step towards human-like artificial intelligence (AI). Yet the performance of LLMs at reasoning tasks have been subpar and the reasoning capability of LLMs is a matter of significant debate. While it has been shown that the choice of the prompting technique to the LLM can alter its performance on a multitude of tasks, including reasoning, the best performing techniques require human-made prompts with the knowledge of the tasks at hand. We introduce a framework for what we call Combinatorial Reasoning (CR), a fully-automated prompting method, where reasons are sampled from an LLM pipeline and mapped into a Quadratic Unconstrained Binary Optimization (QUBO) problem. The framework investigates whether QUBO solutions can be profitably used to select a useful subset of the reasons to construct a Chain-of-Thought style prompt. We explore the acceleration of CR with specialized solvers. We also investigate the performance of simpler zero-shot strategies such as linear majority rule or random selection of reasons. Our preliminary study indicates that coupling a combinatorial solver to generative AI pipelines is an interesting avenue for AI reasoning and elucidates design principles for future CR methods.
MCDUMP is a computer program that selects data from the SKYMAP SKY2000 Master Star Catalog a database about 150 MB in size, stored on a computer hard drive. The database describes about 300,000 stars, each by means of a 500-byte entry. MCDUMP reads all 300,000 entries, then generates an output file that comprises a subset of entries selected according to one or more criteria entered by the user. Examples of criteria that could be entered include: location in a selected portion of the sky; constancy or a specified degree of variability of brightness; absence of nearby, bright companion stars; a particular surface temperature; and brightness sufficient to enable detection by a specified astronomical instrument. The output of MCDUMP can be in the form of either a single 520-column file or multiple files that contain fewer columns to facilitate printing. MCDUMP has been configured and tested for use under the HP-UX 10.20 operating system (a Hewlett-Packard version of the UNIX operating system). It should also be possible to adapt MCDUMP to other versions of UNIX.
SOM_VIS is a computer program for analysis and display of multidimensional sets of Earth-image data typified by the data acquired by the Multi-angle Imaging Spectro-Radiometer [MISR (a spaceborne instrument)]. In SOM_VIS, an enhanced self-organizing-map (SOM) algorithm is first used to project a multidimensional set of data into a nonuniform three-dimensional lattice structure. The lattice structure is mapped to a color space to obtain a color map for an image. The Voronoi cell-refinement algorithm is used to map the SOM lattice structure to various levels of color resolution. The final result is a false-color image in which similar colors represent similar characteristics across all its data dimensions. SOM_VIS provides a control panel for selection of a subset of suitably preprocessed MISR radiance data, and a control panel for choosing parameters to run SOM training. SOM_VIS also includes a component for displaying the false-color SOM image, a color map for the trained SOM lattice, a plot showing an original input vector in 36 dimensions of a selected pixel from the SOM image, the SOM vector that represents the input vector, and the Euclidean distance between the two vectors.
In data mining, one often needs to analyze datasets with a very large number of attributes. Performing machine learning directly on such data sets is often impractical because of extensive run times, excessive complexity of the fitted model (often leading to overfitting), and the well-known "curse of dimensionality." In practice, to avoid such problems, feature selection and/or extraction are often used to reduce data dimensionality prior to the learning step. However, existing feature selection/extraction algorithms either evaluate features by their effectiveness across the entire data set or simply disregard class information altogether (e.g., principal component analysis). Furthermore, feature extraction algorithms such as principal components analysis create new features that are often meaningless to human users. In this article, we present input decimation, a method that provides "feature subsets" that are selected for their ability to discriminate among the classes. These features are subsequently used in ensembles of classifiers, yielding results superior to single classifiers, ensembles that use the full set of features, and ensembles based on principal component analysis on both real and synthetic datasets.
We survey the broadband spectral properties of approximately 500 X-ray-selected active galactic nuclei (AGNs) observed with the Einstein Observatory. Included in this survey are the approximately 450 AGNs in the Extended Medium Sensitivity Survey (EMSS) of Gioia et al. (1990) and the approximately 50 AGNs in the Ultrasoft Survey of Cordova et al. (1992). We present a revised version of the latter sample, based on the post publication discovery of a software error in the Einstein Rev-1b processing. We find that the mean spectral index of the AGNs between 0.1 and 0.6 keV is softer, and the distribution of indices wider, than previous estimates based on analyses of the X-ray spectra of optically selected AGNs. A subset of these AGNs exhibit flux variabiulity, some on timescales as short as 0.05 days. A correlation between radio and hard X-ray luminosity is confirmed, but the data do not support a correlation between the radio and soft X-ray luminosities, or between radio loudness and soft X-ray spectral slope. Evidence for physically distinct soft and hard X-ray components is found, along with the possibility of a bias in previous optically selected samples toward selection of AGNs with flatter X-ray spectra.
The TADPLOT Program, Version 2.0 is described. The TADPLOT program is a software package coordinated by a single, easy-to-use interface, enabling the researcher to access several standard file formats, selectively collect specific subsets of data, and create full-featured publication and viewgraph quality plots. The user-interface was designed to be independent from any file format, yet provide capabilities to accommodate highly specialized data queries. Integrated with an applications software network, data can be assessed, collected, and viewed quickly and easily. Since the commands are data independent, subsequent modifications to the file format will be transparent, while additional file formats can be integrated with minimal impact on the user-interface. The graphical capabilities are independent of the method of data collection; thus, the data specification and subsequent plotting can be modified and upgraded as separate functional components. The graphics kernel selected adheres to the full functional specifications of the CORE standard. Both interface and postprocessing capabilities are fully integrated into TADPLOT.
Regression analysis is applied to the problem of measuring water quality parameters from remote sensing spectral signature data. The equations necessary to perform regression analysis are presented and methods of testing the strength and reliability of a regression are described. An efficient algorithm for selecting an optimal subset of the independent variables available for a regression is also presented.