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.
Restructurable controls, failure classification, airframe design, failure type, control system type, post failure mission, plant identification, controller design techniques, and restructurable control concept validation tools are considered.
Successful project managers (PMs) have been shown to rely on 'intuition,' experience, and analogical reasoning heuristics. For new PMs to be trained and experienced PMs to avoid repeating others' mistakes, it is necessary to make the knowledge and heuristics of successful PMs more widely available. The preparers have evolved a model of PM thought processes over the last decade that is now ready to be implemented as a generic PM aid. This aid consists of a series of 'specialist' expert systems (CRITIC, LIBRARIAN, IDEA MAN, CRAFTSMAN, and WRITER) that communicate with each other via a 'blackboard' architecture. The various specialist expert systems are driven to support PM training and problem solving since any 'answers' they pass to the blackboard are subjected to conflict identification (AGENDA FORMULATOR) and GOAL SETTER inference engines.
A method for estimating an unknown domain in elliptic boundary value problems is considered. The problem is formulated as an inverse problem of integral equations of the second kind. A computational method is developed using a splice collocation scheme. The results can be applied to the inverse problem of impedance computed tomography (ICT) for image reconstruction.
The positive aspect of problem occurrences is the opportunity for learning and a challenge for innovation. The learning aspect is not restricted to the solution period of the problem occurrence, but can become the beacon for problem prevention on future programs. Problems/failures serve as a point of departure for scaling to new designs. To ensure that problems/failures and their solutions guide the future programs, a concerted effort has been expended to study these problems, their solutions, their derived lessons learned, and projections for future programs. This includes identification of technology thrusts, process changes, codes development, etc. However, they must not become an excuse for adding layers upon layers of standards, criteria, and requirements, but must serve as guidelines that assist instead of stifling engineers. This report is an extension of prior efforts to accomplish this task. Although these efforts only scratch the surface, it is a beginning that others must complete.
This paper considers factors that contribute to poor identification of unsteady aerodynamics from wind tunnel data for an airliner configuration. One approach to modeling a wing-tail configuration is considered and applied to both steady and large-amplitude forced pitch oscillation wind tunnel data taken over a wide range of angles of attack but a limited range of amplitude and frequencies. The identified models fit the measured data well but in some cases with inaccurate parameters. Only limited conclusions can be drawn from analysis of the current data set until further experiments can be performed to resolve the identification issues. The analysis of measured and simulated data provides some insights and guidance on how an effective experiment may be designed for wing-tail configurations with nonlinear unsteady aerodynamics.
Video surveillance is one of the most important technologies used by the International Atomic Energy Agency in international safeguards. At large, complicated facilities, multiple surveillance cameras are deployed to monitor the transfer of safeguards-relevant objects across the site. During inspections, all surveillance videos are reviewed to ensure the objects are not manipulated or diverted during transfer, a laborious, time-consuming task. This work describes using deep machine learning algorithms to track objects automatically across multiple cameras, greatly improving the efficiency of the review process. The fundamental problem in this object tracking task across multiple cameras is how to associate the same object, which may show extreme intra-class variations, such as viewpoints, occlusions, and various scales, in different and even non-overlapped cameras. Object re-identification (Re-ID) in nuclear facility video surveillance is even more challenging than classic person or vehicle Re-ID problems because different instances in the same category may display an identical appearance. One observation from nuclear facility surveillance videos is that all objects must be carted (e.g., via forklift) to move. Therefore, the spatial context information of an object, which provides the feature from the carrier, is critical for the object Re-ID task. This work proposes a two-stream convolutional neural networks model that takes features of objects and their surrounding regions into account. Moreover, the custom videos usually are gleaned from different scenes from the training data, which may have extreme variations in illumination changes and/or cluttered backgrounds. Directly applying the trained model to custom videos will dramatically decrease the performance. To tackle this problem, an advanced domain adaptation technique is proposed to mitigate the gap between the data taken from different scenes. The proposed framework will track objects of interest across a nuclear complex. The resulting tracks can be used in further analyses, such as event/activity recognition, anomaly detection, etc.
A model is developed for the evaluation of acreages (proportions) of different crop-types over a geographical area using a classification approach and methods for estimating the crop acreages are given. In estimating the acreages of a specific croptype such as wheat, it is suggested to treat the problem as a two-crop problem: wheat vs. nonwheat, since this simplifies the estimation problem considerably. The error analysis and the sample size problem is investigated for the two-crop approach. Certain numerical results for sample sizes are given for a JSC-ERTS-1 data example on wheat identification performance in Hill County, Montana and Burke County, North Dakota. Lastly, for a large area crop acreages inventory a sampling scheme is suggested for acquiring sample data and the problem of crop acreage estimation and the error analysis is discussed.
The NASA Short-term Prediction Research and Transition (SPoRT) Center developed the DustTracker-AI model for identifying and tracking dust in NASA/NOAA Geostationary Operational Environmental Satellite (GOES) imagery in a real-time framework. A training dataset consisting of day and night dust cases was gathered over the southwestern consisting of 115 distinct images and over a million dust pixels and 256 million no dust pixels. The dataset was separated into training (60%), testing (20%), and validation (20%). A simple random forest machine learning model was developed originally to overcome the problem of night-time dust detection and has been expanded to a comprehensive day/night model for dust identification and tracking. This physically-based machine-learning approach uses NASA/NOAA GOES-16 Advanced Baseline Imager infrared imagery as inputs to the model. The model probability of dust output achieves an Area-Under-Curve (AUC) of 0.97 with a standard deviation of 0.04 for dust cases. For images with dust present, the model correctly labels 85% of dust pixels for all dust images in the validation data set. In conjunction with developing the machine-learning model, the NASA Short-term Prediction Research and Transition Center (SPoRT) partnered with NOAA National Weather Service forecast offices to evaluate the model for utility in weather forecasting operations during the 2021 and 2023 late winter-spring seasons. Preliminary evaluation has indicated the majority of forecasters described the DustTracker-AI probabilities as having added confidence to interpreting the Dust RGB and other satellite products to objectively assess the dust extent and trends and increased the amount of time the dust plume could be tracked into the night as compared to use of the Dust RGB. More recently, SPoRT tested small scale events associated with thunderstorm outflow and burn scars to determine the model’s ability to capture local events. This presentation highlights design of the model, validation/evaluation of model performance, and example cases collected during end user product assessments.
The transition in confined rotating flows is a topical problem with many industrial and fundamental applications. The purpose of this study is to investigate the Taylor-Couette flow in a finite-length cavity with counter-rotating walls, for two aspect ratios L=5 or L=6. Two complex regimes of wavy vortex and spirals are emphasized for the first time via direct numerical simulation, by using a three-dimensional spectral method. The spatio-temporal behavior of the solutions is analyzed and compared to the few data actually available. c2001 Academie des sciences/Editions scientifiques et medicales Elsevier SAS.
This report summarizes the collective work of a five-person team from different organizations examining the problem of detecting foreign object damage (FOD) events in turbofan engines from gas path thermodynamic and bearing accelerometer sensors, and determining the severity of damage to each component (diagnosis). Several detection and diagnostic approaches were investigated and a software tool (FODID) was developed to assist researchers detect/diagnose FOD events. These approaches include (1) fan efficiency deviation computed from upstream and downstream temperature/ pressure measurements, (2) gas path weighted least squares estimation of component health parameter deficiencies, (3) Kalman filter estimation of component health parameters, and (4) use of structural vibration signal processing to detect both large and small FOD events. The last three of these approaches require a significant amount of computation in conjunction with a physics-based analytic model of the underlying phenomenon the NPSS thermodynamic cycle code for approaches 1 to 3 and the DyRoBeS reduced-order rotor dynamics code for approach 4. A potential application of the FODID software tool, in addition to its detection/diagnosis role, is using its sensitivity results to help identify the best types of sensors and their optimum locations within the gas path, and similarly for bearing accelerometers.
It is known that dynamic rotor inflow has a substantial effect on rotor dynamic loads. Despite the complexity of the unsteady flow problem, simple analytical models can be made useful by identifying their parameters from transient response tests without performing flow measurements. Two analytical inflow models are studied: the first is based on an equivalent blade Lock number, the second is based on a time delayed unsteady momentum inflow. In preparation for the experimental data analysis, identifications from simulated test data and an eigenvalue analysis are performed. The experimental results show that the first analytical inflow model is accurate for rotor advance ratios of 0.4 and above. For lower advance ratios, the second inflow model provides better accuracy. Prediction studies with experimental data not used for the identification are performed to determine the accuracy of the mathematical models.
Data management areas were studied to identify pertinent problems and issues that will affect future NASA data users in terms of performance and cost. Specific topics discussed include the identifications of potential NASA data users other than those normally discussed, consideration affecting the clustering of minicomputers, low cost computer system for information retrieval and analysis, the testing of minicomputer based data base management systems, ongoing work related to the use of dedicated systems for data base management, and the problems of data interchange among a community of NASA data users.
In the present automated system for the organization of the data and computational operations entailed by parallel problems, in ways that optimize multiprocessor performance, general heuristics for partitioning program data and control are implemented by capturing and manipulating representations of a computation at run time. These heuristics are directed toward the dynamic identification and allocation of concurrent work in computations with irregular computational patterns. An optimized static-workload partitioning is computed for such repetitive-computation pattern problems as the iterative ones employed in scientific computation.
Based on literature analysis, X-ray photoelectron spectroscopy (XPS) use continues to increase exponentially. This increased use is accompanied by anecdotal reports and systematic analyses indicating a growing presence of significantly flawed data analyses. Recognition of this problem within the surface analysis community has increased with an understanding that both inexperienced users and increased use of XPS outside the surface analysis community contribute to the problem. The XPS community has initiated several efforts to help address the problem, which is not unique to XPS. This paper describes some of the specific problems identified and some of the community efforts intended to address them. Here, we describe activities focused on three specific issues: (i) requests for detailed guides and protocols and bite-sized versions of information for non-experts, (ii) incomplete data and analysis reporting, and (iii) the high rate of peak fitting problems. A 2019 survey identified the need for guides, protocols, and standards to assist XPS users. One set of such guides has been published, and another is being assembled. Providing incremental bites of useful information is the goal of a series of papers on specific challenges to surface analysis with example solutions has been initiated as Notes and Insights papers in Surface and Interface Analysis. Examination of XPS-containing papers finds that information to establish the credibility and reproducibility of XPS results is often very incomplete. Unfortunately, ISO and ASTM standards require an amount of parameter reporting that seems excessive and unrealistic for many research publications. Initial approaches to develop and distribute a graded approach to parameter reporting are briefly described. Multiple efforts are underway to address the high rate of problems associated with photoelectron peak fitting. These include guides to peak fitting, guides to peak identification and fitting for specific elements, and the development of a peak fitting social network. The fitting social network is designed to facilitate interactions between new and experienced XPS users; analysts trying to fit XPS data (for publication or other reasons) can ask questions and establish dynamic conversations. Encouraging and enabling high-quality XPS analysis and reporting requires several different types of effort from all members of the surface and interface analysis community.
Abstract Machine learning and artificial intelligence (ML/AI) methods have been used successfully in recent years to solve problems in many areas, including image recognition, unsupervised and supervised classification, game-playing, system identification and prediction, and autonomous vehicle control. Data-driven machine learning methods have also been applied to fusion energy research for over 2 decades, including significant advances in the areas of disruption prediction, surrogate model generation, and experimental planning. The advent of powerful and dedicated computers specialized for large-scale parallel computation, as well as advances in statistical inference algorithms, have greatly enhanced the capabilities of these computational approaches to extract scientific knowledge and bridge gaps between theoretical models and practical implementations. Large-scale commercial success of various ML/AI applications in recent years, including robotics, industrial processes, online image recognition, financial system prediction, and autonomous vehicles, have further demonstrated the potential for data-driven methods to produce dramatic transformations in many fields. These advances, along with the urgency of need to bridge key gaps in knowledge for design and operation of reactors such as ITER, have driven planned expansion of efforts in ML/AI within the US government and around the world. The Department of Energy (DOE) Office of Science programs in Fusion Energy Sciences (FES) and Advanced Scientific Computing Research (ASCR) have organized several activities to identify best strategies and approaches for applying ML/AI methods to fusion energy research. This paper describes the results of a joint FES/ASCR DOE-sponsored Research Needs Workshop on Advancing Fusion with Machine Learning, held April 30–May 2, 2019, in Gaithersburg, MD (full report available at https://science.osti.gov/-/media/fes/pdf/workshop-reports/FES_ASCR_Machine_Learning_Report.pdf ). The workshop drew on broad representation from both FES and ASCR scientific communities, and identified seven Priority Research Opportunities (PRO’s) with high potential for advancing fusion energy. In addition to the PRO topics themselves, the workshop identified research guidelines to maximize the effectiveness of ML/AI methods in fusion energy science, which include focusing on uncertainty quantification, methods for quantifying regions of validity of models and algorithms, and applying highly integrated teams of ML/AI mathematicians, computer scientists, and fusion energy scientists with domain expertise in the relevant areas.
The ATLAS detector at the Large Hadron Collider reads out particle collision data from over 100 million electronic channels at a rate of approximately $100$ kHz, with a recording rate for physics events of approximately 1 kHz. Before being certified for physics analysis at computer centres worldwide, the data must be scrutinised to ensure they are clean from any hardware or software related issues that may compromise their integrity. Prompt identification of these issues permits fast action to investigate, correct and potentially prevent future such problems that could render the data unusable. This is achieved through the monitoring of detector-level quantities and reconstructed collision event characteristics at key stages of the data processing chain. This paper presents the monitoring and assessment procedures in place at ATLAS during 2015-2018 data-taking. Through the continuous improvement of operational procedures, ATLAS achieved a high data quality efficiency, with 95.6% of the recorded proton-proton collision data collected at $\sqrt{s}=13$ TeV certified for physics analysis.
Automated identification of protein conformational states from simulation of an ensemble of structures is a hard problem because it requires teaching a computer to recognize shapes. We adapt the naïve Bayes classifier from the machine learning community for use on atom-to-atom pairwise contacts. The result is an unsupervised learning algorithm that samples a ‘distribution’ over potential classification schemes. We apply the classifier to a series of test structures and one real protein, showing that it identifies the conformational transition with >95% accuracy in most cases. A nontrivial feature of our adaptation is a new connection to information entropy that allows us to vary the level of structural detail without spoiling the categorization. This is confirmed by comparing results as the number of atoms and time-samples are varied over 1.5 orders of magnitude. Further, the method’s derivation from Bayesian analysis on the set of inter-atomic contacts makes it easy to understand and extend to more complex cases.
As the radio frequency spectrum becomes increasingly overcrowded, interference with mission-critical DSN operations is rising at an alarming rate. To alleviate this problem the DSN is developing a wideband surveillance system for on-site detection and identification of potential sources of radio frequency interference (RFI), which will complement the existing frequency coordination activities. The RFI monitoring system is based on a wideband, multi-look discrete spectrum analyzer operating on fast Fourier transform principles. An extensive general statistical analysis is presented of such spectrum analyzers and derives threshold detection performance formulas for signals of interest. These results are then applied to the design of the RFI spectrum analyzer under development.